Method for wind power system to participate in energy-frequency modulation market based on cloud energy storage

Abstract

The invention relates to the technical field of electricity markets, and discloses a method for a wind power system to participate in an energy-frequency modulation market based on cloud energy storage, and the method comprises the steps: carrying out the time-frequency synchronous decomposition and noise reduction of original wind power and multi-dimensional feature data, and predicting the power of the wind power; determining an optimal harmonic moment number and performing probability density estimation on prediction error data; constructing a prediction error fluctuation domain, and integrating to obtain a multiple uncertainty set; and based on the multiple uncertainty sets, constructing a master-slave game optimization model of a wind power plant and a cloud energy storage operator and a wind power system multi-target optimization model considering cloud energy storage and electric energy quality, and solving the optimal capacity and lease price of leasing cloud energy storage and an energy-frequency modulation market participation strategy of the wind power system. According to the method, the concept of cloud energy storage is introduced into the wind power plant, additional investment cost brought by independently building matched energy storage facilities for the wind power plant can be reduced, deviation punishment cost can be reduced, and frequency modulation performance benefits can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power market, and in particular to a method for a wind power system based on cloud energy storage to participate in an energy-frequency regulation market. Background Art

[0002] At present, the penetration rate of new energy represented by wind power is constantly increasing. New energy will undertake more frequency regulation tasks, and it has become a trend for new energy power generators to participate in the bidding competition in the energy-frequency regulation market. However, the volatility and uncertainty of wind power output make it face the risk of penalty due to bidding deviation in market participation, which weakens its market competitiveness. Energy storage, as a high-quality flexibility resource, can effectively alleviate the problem of wind power deviation. However, the construction of supporting energy storage equipment for wind farms will greatly increase the investment cost, and there is a risk of long cost recovery period or no cost recovery.

[0003] With the rise of the cloud energy storage business model, leasing cloud energy storage has become a practical solution, which can not only effectively improve the utilization rate of energy storage and reduce the cost of wind farms, but also promote the consumption of new energy and make wind farms more competitive in the market. Leasing cloud energy storage also provides favorable conditions for wind power and energy storage systems to jointly participate in the energy-frequency regulation market, further optimizing the participation model of wind farms in the electricity market.

[0004] From the above problems, it can be seen that the uncertainty of wind power output may cause wind farms to face higher penalty risk costs. As the basis for participating in the power market bidding, the accuracy of wind power forecasting directly affects the bidding strategy and benefits of wind farms. Therefore, it is necessary to adopt a more advanced combined forecasting model to improve the accuracy and stability of wind power forecasting results. In addition, the pricing strategy of wind power leasing cloud energy storage and the strategy of wind storage jointly participating in the energy-frequency regulation market are still imperfect, especially the impact of multiple uncertain factors is not fully considered, so it is necessary to improve them.

[0005] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0006] In response to the problems in the related technology, the present invention proposes a method for a wind power system based on cloud energy storage to participate in the energy-frequency regulation market, so as to overcome the above-mentioned technical problems existing in the existing related technology.

[0007] To this end, the specific technical solution adopted by the present invention is as follows:

[0008] A method for a wind power system based on cloud energy storage to participate in an energy-frequency regulation market comprises the following steps:

[0009] S1. Obtain the original wind power and multi-dimensional feature data, use the improved masking signal method combined with the robust local mean decomposition method to perform time-frequency synchronous decomposition and noise reduction on the original wind power and multi-dimensional feature data, and use the Transformer model based on local information enhancement and sparse attention mechanism to predict wind power;

[0010] S2, determine the optimal harmonic moment number of the density estimation method based on harmonic transformation, perform probability density estimation on the prediction error data of power and price, and construct multiple uncertainty sets;

[0011] S3. Based on multiple uncertainty sets, a master-slave game optimization model of wind farms and cloud energy storage operators and a multi-objective optimization model of wind power systems considering cloud energy storage and power quality are constructed, and the optimal capacity and rental price of cloud energy storage and the energy-frequency market participation strategy of wind power systems are solved.

[0012] Preferably, the method of obtaining original wind power and multi-dimensional feature data, performing time-frequency synchronous decomposition and noise reduction on the original wind power and multi-dimensional feature data using an improved masking signal method combined with a robust local mean decomposition method, and predicting wind power using a Transformer model based on local information enhancement and a sparse attention mechanism includes the following steps:

[0013] S11. Obtain original wind power and multidimensional feature data, and decompose the original wind power and multidimensional feature data into several optimal component product function components and residual signals based on the robust local mean decomposition method; use the Spearman correlation coefficient to remove components containing interference signals, and use approximate entropy to select components containing effective features;

[0014] S12. Use the improved masking signal method to mask the selected components, and reconstruct the optimal component product function components after processing to obtain the denoised wind power signal components; use the Transformer model based on local information enhancement and sparse attention mechanism, combined with the denoised wind power signal components to predict wind power.

[0015] Preferably, the obtaining of original wind power and multidimensional feature data, decomposing the original wind power and multidimensional feature data into a number of optimal component product function components and residual signals based on a robust local mean decomposition method; removing components containing interference signals using a Spearman correlation coefficient, and selecting components containing effective features using approximate entropy comprises the following steps:

[0016] S111, obtaining original wind power and multi-dimensional feature data, using the mirror extension method to process the boundary to search for all local extreme points in the original wind power signal; obtaining the local mean function and the local envelope function by the sliding average method, and calculating the zero mean signal and the pure frequency signal;

[0017] The calculation formula for the zero-mean signal is:

[0018] k 11 (t) = x(t) - m 11 (t)

[0019] The calculation formula for pure frequency signal is:

[0020] s 11 (t) = k 11 (t) / a 11 (t)

[0021] In the formula, k 11 (t) is a zero-mean signal, x(t) is the wind power signal to be decomposed, m 11 (t) is the local mean function, s 11 (t) is the processed pure frequency signal, a 11 (t) is the local envelope function;

[0022] S112, defining an objective function, extracting the first product function from the original wind power signal to obtain a residual signal, and treating the residual signal as a new signal to repeatedly iterate and extract all optimal component product function components until the residual becomes a constant or a monotonic function;

[0023] Among them, the expression of the objective function is:

[0024] f=RMS[z(t)]+EK[z(t)]

[0025]

[0026] Where f is the objective function value, RMS is the root mean square, EK is the empirical kurtosis calculation formula, z(t) is the zero baseline envelope signal, is the arithmetic mean of z(t), N s is the total number of signals;

[0027] S113. Use the Spearman correlation coefficient to remove components containing interference signals, and use approximate entropy to select components containing effective features.

[0028] Preferably, the improved masking signal method is used to perform masking processing on the selected components, and the optimal component product function components after processing are reconstructed to obtain the denoised wind power signal components; the Transformer model based on local information enhancement and sparse attention mechanism is used to predict wind power in combination with the denoised wind power signal components, including the following steps:

[0029] S121, calculating the masked mean signal, combining the Spearman correlation coefficient and the approximate entropy to select several components in the robust local mean decomposition result for masking processing; performing robust local mean decomposition on the masked processing result to obtain the signal component after masking processing, and performing final signal reconstruction based on the signal component after masking processing to obtain the wind power signal component after noise reduction;

[0030] Among them, the calculation formula of the masked mean signal is:

[0031]

[0032] The expression for masking is:

[0033]

[0034] The expression of the signal component after masking is:

[0035]

[0036] In the formula, s i (t) is the i-th masking signal, is the average amplitude, is the mean value of the instantaneous frequency, PF i+ (t), PF i- (t) are the signal components after averaging and subtracting, respectively, PF i (t) is the original component, IMS_PF i is the signal component after masking processing, pf i+ , pf i- PF i+ (t) and PF i- (t) the first component of the decomposition output;

[0037] S122, standardizing the wind power signal components after noise reduction, using a fully connected layer to perform high-dimensional representation on the wind power signal components after standardization, and extracting local information of enhanced data by stacking multiple layers of one-dimensional convolution to obtain a locally enhanced signal;

[0038] S123, using the samples obtained by the multi-layer convolution operation of the locally enhanced signal as the input of the encoder layer, using the position coding to mark the position information of each sample, inputting the position coding result into the encoder layer of the Transformer model, and performing layer normalization processing on the data in the encoder;

[0039] S124. Use a multi-head sparse attention mechanism to replace the fully connected attention mechanism in the standard Transformer model to capture the global information of the sequence, integrate the global and local information of wind power data, and map and learn the components of wind power through the middle layer of the network, and finally output the wind power prediction results.

[0040] Preferably, the determining of the optimal harmonic moment number of the density estimation method based on harmonic transformation, performing probability density estimation on the prediction error data of power and price, and constructing a multiple uncertainty set comprises the following steps:

[0041] S21. Calculate the forecast errors of wind power, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price respectively, determine the optimal harmonic moment number of the density estimation method based on harmonic transformation by using the weighted average algorithm, and perform probability density estimation on the forecast error data of power and price;

[0042] S22. Use the cumulative probability density function and preset confidence to select the upper and lower limits of the prediction error, construct the prediction error fluctuation domain of wind power, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price, and build a multiple uncertainty set for boundary adaptive optimization.

[0043] Preferably, the method of respectively calculating the forecast errors of wind power, energy market capacity clearing price, frequency modulation market capacity clearing price and mileage clearing price, using a weighted average algorithm to determine the optimal harmonic moment number of a density estimation method based on harmonic transformation, and performing probability density estimation on the forecast error data of power and price comprises the following steps:

[0044] S211, respectively calculating the forecast errors of wind power, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price, where the forecast error is the difference between the forecast value and the true value;

[0045] S212, initializing parameters, constructing a fitness function of the optimal harmonic moment number based on relative entropy; in each iteration, calculating the weighted average position of the current harmonic moment number, and taking the weighted average position of the current harmonic moment number as the representative position of the entire search space; adopting different mobile exploration strategies, respectively aiming at exploring a wide solution space and fine-tuning the optimal harmonic moment number;

[0046] Among them, the expression of the fitness function of the optimal harmonic moment number based on relative entropy is:

[0047]

[0048] The weighted average position is calculated as:

[0049]

[0050] Where Fitness(h) is the fitness function of the optimal harmonic moment number based on relative entropy, f(h) is the probability density function based on harmonic transformation, is the probability density function of the kernel density estimation, h is the number of harmonic moments to be optimized, N Candidate is the number of candidate solutions selected, P is the population size, it is the current iteration number, Max It is the maximum number of iterations, Sum Fitness is the sum of all fitness values ​​of the selected candidates, h i is the i-th candidate optimal harmonic moment number, h Miu is the weighted average position;

[0051] S213, selecting and adjusting the mobile strategy based on random constants and dynamic parameters related to the iteration number, achieving a balance between exploration and development, obtaining the optimal harmonic moment number, and using a density estimation method based on harmonic transformation to perform probability density estimation on the forecast errors of wind power probability, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price;

[0052] Among them, the expression of probability density function is:

[0053]

[0054] The expression of the probability density function based on harmonic transformation is:

[0055]

[0056] In the formula, f e (e) is the probability density function obtained by inverse Mellin transformation, α is a fixed value, which is taken as 2, δ is an assumed value, and the complex number ω = α + βi, then δ = 2β, e is the prediction error, and f(e) is the probability density function based on harmonic transformation. It is a harmonic transformation.

[0057] Preferably, the method of selecting the upper and lower limits of the prediction error by using the cumulative probability density function and the preset confidence, constructing the prediction error fluctuation domain of wind power, energy market capacity clearing price, frequency modulation market capacity clearing price and mileage clearing price, and constructing the multiple uncertainty set of boundary adaptive optimization includes the following steps:

[0058] S221. Using the cumulative probability density function and combining with the preset confidence, select the upper and lower limits of the prediction error and construct the prediction fluctuation domain of the uncertain variable;

[0059] Among them, the expression of the cumulative probability density function is:

[0060]

[0061] The confidence interval is:

[0062] e lower =F -1 (0.025)

[0063] e upper =F -1 (0.975)

[0064] The expression for predicting the fluctuation range is:

[0065]

[0066] Where F(e) is the cumulative probability density function, e is the prediction error, and e lower 、e upper are the lower and upper limits of the forecast error fluctuation, respectively, and F -1 is the inverse of the cumulative distribution function, is the predicted value;

[0067] S222, integrating the forecast error fluctuation domains of wind power, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price into a multiple uncertainty set;

[0068] Among them, the expression of multiple uncertainty set is:

[0069]

[0070] In the formula, They are wind power, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price. are the upper and lower limits of wind power forecast fluctuation at time τ, are the upper and lower limits of the energy market capacity clearing price at time t, are the upper and lower limits of the frequency modulation market capacity clearing price at time t, They are the upper and lower limits of the mileage clearing price at time t respectively.

[0071] Preferably, the master-slave game optimization model of wind farms and cloud energy storage operators and the multi-objective optimization model of wind power systems considering cloud energy storage and power quality are constructed based on multiple uncertainty sets, and the optimal capacity and rental price of renting cloud energy storage and the energy-frequency modulation market participation strategy of the wind power system are solved, including the following steps:

[0072] S31. Establish a master-slave game optimization model between wind farms and cloud energy storage operators that takes into account multiple uncertainties, and with the goal of maximizing the benefits of the main cloud energy storage operator and minimizing the cost of the subordinate wind farm, solve the optimal capacity and rental price of leasing cloud energy storage;

[0073] S32. Establish a multi-objective optimization model for wind power systems that takes cloud energy storage and power quality into consideration, combine the optimized cloud energy storage rental price and rental capacity, take maximizing the benefits of the wind power system and optimizing the power quality at the grid connection point as the dual objectives, and solve the energy-frequency market participation strategy of the wind power system.

[0074] Preferably, the master-slave game optimization model between the wind farm and the cloud energy storage operator includes a game subject cloud energy storage operator optimization model and a game slave wind power cluster robust optimization model;

[0075] Among them, the objective function expression of the optimization model of the cloud energy storage operator of the game subject is:

[0076]

[0077] The objective function expression of the game-based wind power cluster robust optimization model is:

[0078]

[0079]

[0080] In the formula, I c Profits for cloud energy storage operators, To rent out energy storage income, The auction revenue of the frequency modulation market is is the energy market bidding revenue, is the cloud energy storage operating cost, are the power price and capacity price of cloud energy storage for leasing respectively. are the charging and discharging power and capacity of cloud energy storage respectively, Δt is the day-ahead market time scale, is the frequency modulation performance index, Bidding capacity for cloud storage in the frequency regulation market, are the capacity clearing price and mileage clearing price of the frequency modulation market, η is the mileage call rate, is the capacity clearing price for the energy market, is the bidding capacity of cloud storage in the energy market, τ is a certain moment in the real-time market, t is a certain moment in the day-ahead market, are respectively the cloud energy storage capacity cost and mileage cost, C w is the total cost of the wind farm, They are respectively the cost of leasing cloud energy storage for wind power, the penalty cost for wind power deviation, and the opportunity cost. are the bidding income of wind power in the frequency regulation market and the bidding income in the electric energy market, respectively; λ1 and λ2 are the application penalty coefficients in the energy market and the frequency regulation market, respectively. are the day-ahead bidding capacity of wind power in the energy market and frequency regulation market, are the real-time bidding capacity of wind storage in the energy market and frequency regulation market, Δτ is the real-time market time scale, N T is the total time, The bidding price for frequency regulation capacity, is the bidding price for frequency regulation mileage, φ(t) is the time-of-use electricity price in the electricity market, P w.fm.c Declare capacity for wind power cluster, P w.fm.c.max To allow wind power clusters to declare a capacity cap in the frequency regulation market, K W (t) is the frequency modulation performance index value, P w.e.c (t), R w,e,c (t) are the bidding capacity and price of wind power cluster in the energy market respectively.

[0081] Preferably, the objective function of the wind power system multi-objective optimization model considering cloud energy storage and power quality includes the optimal wind power system revenue and the optimal wind power system grid-connected power quality;

[0082] Among them, the optimal expression of wind power system benefit is:

[0083]

[0084] The optimal expression for the grid-connected power quality of wind power system is:

[0085]

[0086] Where F1 is the wind power system revenue function, is the total revenue of the energy and frequency regulation market, Penalty cost for wind power deviation, For the optimized wind power leasing cloud energy storage cost, is the opportunity cost of the second stage, H is the set of 24 periods in a day, They are the day-ahead bidding capacity for the second phase wind power energy market and the frequency regulation market. They are the real-time bidding capacity of wind storage in the second phase energy market and frequency regulation market, is the wind power frequency regulation performance index, They are the day-ahead bidding capacity of wind power in the energy market and frequency regulation market in the second phase, are the real-time bidding capacity of wind storage in the second phase energy market and frequency regulation market, respectively. w.fm.c.o is the capacity declared by the wind power cluster in the second phase, F2 is the power quality function of the wind power system grid-connected electricity, M is the total number of time nodes throughout the year, and P w (t) is the output power of the wind farm at time t, and Q(t) is the comprehensive evaluation value of power quality at time t.

[0087] The beneficial effects of the present invention are:

[0088] 1) The present invention combines the improved masked signal method (IMS) with the robust local mean decomposition method (RLMD) to perform time-frequency synchronous decomposition and denoising on wind power and multi-dimensional feature data, thereby effectively removing noise in the data while retaining the main features of the signal. In addition, by introducing the local information enhancement module and the Transformer model with a sparse attention mechanism, the model can enhance its ability to capture local change characteristics in wind power time series data while retaining the learning of global dependencies. The combined prediction model (IMS-RLMD-LSA-Transformer) improves the accuracy and robustness of wind power forecasting by integrating time-frequency denoising with deep learning feature extraction, providing a more reliable data foundation for subsequent market participation strategies.

[0089] 2) The present invention adopts a new meta-heuristic optimization algorithm based on the concept of weighted average holding to determine the optimal harmonic moment of the density estimation method based on harmonic transformation. It has adaptive characteristics, can effectively balance the deviation and variance, and improve the estimation accuracy and reliability. The probability density estimation of the prediction error provides a more accurate way to help better deal with errors and uncertainties in the optimization process, avoiding the limitations of traditional methods. Using the generated fluctuation domain, a multiple uncertainty set of boundary adaptive optimization is constructed, providing a more accurate optimization solution for wind power systems in different complex and dynamic scenarios.

[0090] 3) The present invention can optimize the rental electricity price and rental capacity between the wind farm and the cloud energy storage by establishing a master-slave game optimization model between the wind farm and the cloud energy storage operator. This game optimization model can not only balance the interests of the wind farm and the cloud energy storage operator, but also effectively reduce the uncertainty caused by wind power fluctuations and improve the overall efficiency of the system. By using a robust optimization method, combined with a multi-objective optimization model that considers cloud energy storage and power quality, the system's energy market and frequency regulation market participation strategies can be optimized under a certain degree of conservatism. It ensures that even in the face of forecast errors and market fluctuations, the system can still achieve the optimal market participation strategy, thereby improving the stability and benefits of market participation. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0092] Figure 1It is a flow chart of a method for a wind power system based on cloud energy storage to participate in an energy-frequency regulation market according to an embodiment of the present invention. DETAILED DESCRIPTION

[0093] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in the field should be able to understand other possible implementations and advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0094] According to an embodiment of the present invention, a method for a wind power system based on cloud energy storage to participate in an energy-frequency regulation market is provided.

[0095] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, the method for a wind power system based on cloud energy storage to participate in the energy-frequency regulation market according to an embodiment of the present invention includes the following steps:

[0096] S1. Obtain the original wind power and multi-dimensional feature data, use the improved masking signal method combined with the robust local mean decomposition method to perform time-frequency synchronous decomposition and noise reduction on the original wind power and multi-dimensional feature data, and use the Transformer model based on local information enhancement and sparse attention mechanism to predict wind power;

[0097] In this embodiment, the RLMD method (Robust Local Mean Decomposition) is first used to decompose the acquired original wind power and multidimensional feature data into multiple components, the Spearman correlation coefficient is used to remove the components containing interference signals, and then the approximate entropy is used to select the components containing effective features. Secondly, the IMS method (i.e., the improved masked signal method) is used to mask the selected components, and the processed PF (optimal component product function) components are reconstructed to obtain the denoised signal. Using the sparse attention mechanism, the global correlation between the virtual sample and the wind power data is calculated to capture its long-distance dependency. At the same time, by calculating the correlation of adjacent samples, the local dependency characteristics are further extracted, and the global and local information of the data are fully integrated. After feature extraction and global-local information fusion, the model maps and learns the wind power components through the middle layer of the network, and finally realizes the prediction of wind power.

[0098] The method of obtaining original wind power and multi-dimensional feature data, performing time-frequency synchronous decomposition and noise reduction on the original wind power and multi-dimensional feature data using an improved masking signal method combined with a robust local mean decomposition method, and predicting wind power using a Transformer model based on local information enhancement and sparse attention mechanism includes the following steps:

[0099] S11. Obtain original wind power and multidimensional feature data, and decompose the original wind power and multidimensional feature data into several optimal component product function components and residual signals based on the robust local mean decomposition method; use the Spearman correlation coefficient to remove components containing interference signals, and use approximate entropy to select components containing effective features;

[0100] Specifically, the original wind power and multidimensional feature data are obtained, and the original wind power and multidimensional feature data are decomposed into several optimal component product function components and residual signals based on the robust local mean decomposition method; the components containing interference signals are removed by using the Spearman correlation coefficient, and the components containing effective features are selected by using the approximate entropy, including:

[0101] (1) Obtain the original wind power and multi-dimensional feature data, and use RLMD to decompose the obtained original wind power and multi-dimensional feature data into multiple components. The steps of the RMLD algorithm are as follows:

[0102] 1) Initial signal processing and local extreme point identification:

[0103] Assume that the wind power signal to be decomposed is x(t), and all local extreme points in x(t) are n by processing the boundary through the mirror extension method. i ;

[0104]

[0105] In the formula, m i For two adjacent extreme points n i and n i+1 The mean value of i is the local envelope estimate;

[0106] 2) Obtain the local mean function and local envelope function through the sliding average method, and calculate the zero mean signal and pure frequency signal;

[0107] The calculation formula for the zero-mean signal is:

[0108] k 11 (t) = x(t) - m 11 (t)

[0109] The calculation formula for pure frequency signal is:

[0110] s 11 (t) = k 11(t) / a 11 (t)

[0111] In the formula, k 11 (t) is a zero-mean signal, x(t) is the wind power signal to be decomposed, m 11 (t) is the local mean function, s 11 (t) is the processed pure frequency signal, a 11 (t) is the local envelope function;

[0112] 3) Define the objective function. The expression of the objective function is:

[0113] f=RMS[z(t)]+EK[z(t)]

[0114]

[0115] Where f is the objective function value, RMS is the root mean square, EK is the empirical kurtosis calculation formula, z(t) is the zero baseline envelope signal, is the arithmetic mean of z(t), N s is the total number of signals;

[0116] When the iteration condition is met, the calculated pure FM signal s1(t) and envelope signal a 1j (t) multiply to obtain the first product function;

[0117] s1(t)=s 1j (t)

[0118]

[0119] In the formula, s 1j (t) is the pure FM signal of the jth iteration, PF1(t) is the first optimal component product function of the envelope signal;

[0120] 4) Extract the first product function PF1(t) from the wind power signal x(t) to obtain the residual signal u1(t), and then treat the residual signal u1(t) as a new signal and repeat the iterative steps to extract all the optimal component product function components until u1(t) is obtained. j (t) becomes a constant or a monotonic function; finally, the original signal x(t) is decomposed into p product function PF components and a residual u j (t);

[0121]

[0122] Where p is the number of product function components, u j (t) is the residual of the jth iteration;

[0123] (2) The Spearman correlation coefficient is used to remove the components containing interference signals, and the approximate entropy is used to select the components containing effective features.

[0124] The Spearman correlation coefficient is a nonparametric correlation measure used to assess the correlation between the ranks or orderings of two sets of data and is calculated as follows:

[0125]

[0126] In the formula, r s is the Spearman correlation coefficient, d i is the rank of the difference between the ranks of two variables, and n is the number of data pairs;

[0127] Approximate entropy is a statistical tool to measure the complexity and regularity of time series data. The calculation principle is as follows:

[0128]

[0129] Where ApEn is the approximate entropy, Φm(r) is the domain probability of m-dimensional embedding, m is the embedding dimension, and r is the threshold;

[0130] S12. Use the improved masking signal method to mask the selected components, and reconstruct the optimal component product function components after processing to obtain the denoised wind power signal components; use the Transformer model based on local information enhancement and sparse attention mechanism, combined with the denoised wind power signal components to predict wind power.

[0131] In this embodiment, the improved masked signal method means that before using the masked signal method, the Spearman correlation coefficient is first used to remove the components containing the interference noise signal, and then the approximate entropy is used to select the components containing the effective features; secondly, the masked signal method is used to mask the selected components, thereby improving the efficiency and accuracy of the masking. However, each component obtained from the signal decomposition of the existing masked signal method needs to be manually solved for different masked signals, which is very computationally intensive and cannot achieve the noise reduction effect.

[0132] Specifically, the improved masking signal method is used to perform masking processing on the selected components, and the optimal component product function components after processing are reconstructed to obtain the denoised wind power signal components; the Transformer model based on local information enhancement and sparse attention mechanism is used to predict wind power in combination with the denoised wind power signal components, including the following steps:

[0133] (1) Using the IMS method to mask the selected components and reconstruct the processed PF components to obtain the denoised signal;

[0134] 1) Calculate the masked mean signal. The calculation formula of the masked mean signal is:

[0135]

[0136] In the formula, s i (t) is the i-th masking signal, is the average amplitude, is the mean of the instantaneous frequency;

[0137] 2) Combine the Spearman correlation coefficient and approximate entropy to select several components in the robust local mean decomposition result for masking processing. The expression of masking processing is:

[0138]

[0139] In the formula, PF i+ (t), PF i- (t) are the signal components after averaging and subtracting, respectively, PF i (t) is the original component;

[0140] 3) PF i+ (t), PF i- (t) Perform RLMD decomposition again to obtain the signal component after masking. The expression of the signal component after masking is:

[0141]

[0142] Where, IMS_PF i is the signal component after masking processing, pf i+ , pf i- PF i+ (t) and PF i- (t) the first component of the decomposition output;

[0143] 4) The final signal is reconstructed according to the signal component after masking to obtain the wind power signal component after noise reduction. The expression of the reconstructed signal is:

[0144]

[0145] Where irx(t) is the reconstructed signal;

[0146] (2) A Transformer model based on local information enhancement and sparse attention mechanism is used to predict wind power combined with the denoised wind power signal components;

[0147] Specifically, the denoised wind power signal components are input into the model, and multiple layers of one-dimensional convolution (Conv1D) are stacked. Each layer uses convolution kernels of different sizes to capture local change characteristics in wind power data from multiple scales and strengthen the expression of local information. The front-layer normalization is used to replace the traditional back-layer normalization. The sparse attention mechanism is used to calculate the global correlation between virtual samples and wind power data to capture their long-range dependencies. At the same time, by calculating the correlation between adjacent samples, the local dependency characteristics are further extracted to fully integrate the global and local information of wind power data. After feature extraction and global-local information fusion, the model maps and learns the components of wind power through the middle layer of the network, and finally outputs the wind power prediction results.

[0148] 1) Local information enhancement based on Conv1D, that is, standardizing the wind power signal components after noise reduction, using the fully connected layer to perform high-dimensional representation of the standardized wind power signal components, and extracting the local information of the enhanced data by stacking multiple layers of one-dimensional convolution to obtain the locally enhanced signal;

[0149] First, the input wind power data component X = [x1, x2, x3, ..., x N ] Perform Z-score standardization to obtain X Z-score ∈R N×M , R is a set of real numbers, M is the number of features;

[0150]

[0151] Where, X Z-score To perform Z standardization on wind power data, x is the mean value of all samples, S is the standard deviation, and x i is the i-th sample, i∈[1,N], N is the total number of samples;

[0152] Then, using the fully connected layer, X Z-score With feature dimension M being mapped to d model dimensional to obtain its high-dimensional representation Secondly, stacking multiple layers of Conv1D extracts and enhances the local information of the data, and obtains

[0153] X Conv =X Z-score-d +max(0,BN(Conv1D n (X Z-score-d )))

[0154] Where, X Conv is the signal after local enhancement, X Z-score-d It is the high-dimensional representation of the normalized original data, n is the number of convolutional layers, and BN is the batch normalization function.

[0155] Finally, X Conv The samples obtained after multi-layer convolution operations are used as the input of the encoder layer. The position information of each sample is marked using position encoding. The position encoding result is input into the encoder layer of the Transformer model. In the encoder, the data is layer-normalized, as shown in the formula:

[0156] X Conv ′=LayerNorm(ReLU(X Conv ))

[0157] Where, X Conv ′ is the signal after layer normalization, LayerNorm is the layer normalization, and ReLU is the activation function;

[0158] 2) Global information extraction based on sparse attention mechanism

[0159] When X Conv ′ is integrated into the encoder layer, and a multi-head sparse attention mechanism is used to replace the fully connected attention mechanism in the standard Transformer model to capture the global information of the sequence. For the vector sequence H = X Conv ,H=[h1,...h i ...,h N ], Represents the high-dimensional embedding of the i-th sample data. The formula for updating sample data is:

[0160]

[0161] MHSAttention(q,H)=Concat(head1,...head a ...,head h )·W O

[0162] head a =Attention(qW Q ,KW K ,VW V )

[0163]

[0164] In the formula, represents the i-th sample data of the context information, Concat represents the connection function, are the left and right neighbor states of the current sample, respectively. S t-1 are the previous state, initial value and previous state of the virtual data respectively, is the state of each sample data, MHSAttention is the multi-head coefficient attention mechanism, q represents the high-dimensional embedding of each sample data, h represents the number of heads, and head a represents the ath head in the calculation result of the multi-head sparse attention mechanism, W Q ,W K ,W V ,W O are different learnable parameters, Q, K, V, O are different value phasors, d K is the projection space dimension, softmax is the normalization function, and T is the length of the input sequence;

[0165] Introduction Represents sample data, the initialization state is S 0 =average(H) and initialization value H 0 =H, the process of updating virtual data is:

[0166] S t =MultiAttention(S t-1 ,Concat[S t-1 ;H t ])

[0167] S t =LayerNorm(ReLU(St))

[0168] In the formula, S t is the final state of the sample data, MultiAttention is the dot product attention mechanism, H t is the final state of the virtual data, S t-1 is the state at the previous moment;

[0169] Then it enters the feed-forward neural network (FFN) to increase the high dimensionality of the data, filter out useful information, and then reduce it to a low dimensionality to maintain the consistency of the dimensions before and after the FFN operation;

[0170] H t =LayerNorm(ReLU(H t ))

[0171] FFN=max(0,H t ·W1+b1)·W2+b2

[0172] Where FFN is a feedforward neural network; W1, W2, b1, and b2 are the weights and offsets of the two linear transformations of FFN respectively;

[0173] Finally, d modelThe data is reduced to one dimension through the fully connected layer to obtain the final wind power prediction result X trm ∈R N×1 ;

[0174]

[0175] Where, X trm This is the final wind power prediction result.

[0176] S2, determine the optimal harmonic moment number of the density estimation method based on harmonic transformation, perform probability density estimation on the prediction error data of power and price, and construct multiple uncertainty sets;

[0177] The method of determining the optimal harmonic moment number of the density estimation method based on harmonic transformation, performing probability density estimation on the prediction error data of power and price, and constructing a multiple uncertainty set includes the following steps:

[0178] S21. Calculate the forecast errors of wind power, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price respectively, determine the optimal harmonic moment number of the density estimation method based on harmonic transformation by using the weighted average algorithm, and perform probability density estimation on the forecast error data of power and price;

[0179] Specifically, respectively calculating the forecast errors of wind power, energy market capacity clearing price, frequency modulation market capacity clearing price and mileage clearing price, using a weighted average algorithm to determine the optimal harmonic moment number of a density estimation method based on harmonic transformation, and performing probability density estimation on the forecast error data of power and price includes the following steps:

[0180] (1) Relative entropy is selected as the fitness function, and the weighted average algorithm is used to determine the optimal harmonic moment number based on the harmonic transform density estimation method;

[0181] 1) Initialization phase:

[0182] h i,j = rand·(h maxj -h minj )+h minj

[0183] In the formula, h i,j is the position of the ith solution in the jth dimension, rand is a random number between 0 and 1, and h max,j 、h min,j are the lower and upper limits of the harmonic moment number respectively;

[0184] Define the optimal harmonic moment fitness function based on relative entropy:

[0185]

[0186] Where Fitness(h) is the fitness function of the optimal harmonic moment number based on relative entropy, f(h) is the probability density function based on harmonic transformation, is the probability density function of the kernel density estimation, h is the number of harmonic moments to be optimized;

[0187] 2) Calculate the weighted average position:

[0188]

[0189] Where N Candidate is the number of candidate solutions selected, P is the population size, it is the current iteration number, Max It is the maximum number of iterations, Sum Fitness is the sum of all fitness values ​​of the selected candidates, h i is the i-th candidate optimal harmonic moment number, h Miu is the weighted average position;

[0190] 3) Define the search phase:

[0191]

[0192] Where f(it) is the candidate solution and A is a constant that controls the balance between the exploration and exploitation phases;

[0193] 4) Search phase:

[0194]

[0195] In the formula, h i,j (it+1) is the i-th solution at the j-th position at the (it+1)th iteration, S is the step size, is the position in the global optimal solution at the it iteration;

[0196] h i1 (it+1)=rand·(Uh min -Lh min )+Lh min

[0197] Where, Lh min , Uh min are the minimum values ​​of the lower and upper limits of all dimensions respectively;

[0198] 5) Development phase:

[0199] The exploitation strategy simulates how a population of search agents moves around the search space with high probability to exploit new global optima;

[0200]

[0201] In the formula, w 11 、w 12 、w 13 、w 21 、w 22 、w 31 、w 32 are random numbers between 0 and 1, is the best search position for an individual, h i,2 (it+1),h i,3 (it+1),h i,4 (it+1) are the position update formulas under three development strategies respectively, is the global best search position, is the global best position;

[0202] (2) Calculate the forecast errors of wind power, energy market capacity clearing price, frequency modulation market capacity clearing price and mileage clearing price respectively, and perform probability density estimation on the forecast error data of power and price based on the density estimation method of harmonic transformation;

[0203] 1) Compare the prediction results with the actual data and calculate the prediction error:

[0204]

[0205] In the formula, is the predicted value at time t, y(t) is the true value at time t, and e(t) is the prediction error at time t;

[0206] 2) Using the density estimation method based on harmonic transformation to estimate the probability density of the prediction error data;

[0207] According to the Euler formula and the inverse Mellin transformation, the derived probability density function is obtained, and the expression of only the real term is retained:

[0208]

[0209] In the formula, f e (e) is the probability density function obtained by inverse Mellin transformation, α is a fixed value, which is 2 in this embodiment, e is the prediction error; δ is an assumed value, assuming that the complex number ω = α + βi, then δ = 2β;

[0210] Define the harmonic transform:

[0211]

[0212] In the formula, is harmonic transformation;

[0213] The probability density function based on harmonic transformation is:

[0214]

[0215] Where f(e) is the probability density function based on harmonic transformation.

[0216] S22. Use the cumulative probability density function and preset confidence to select the upper and lower limits of the prediction error, construct the prediction error fluctuation domain of wind power, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price, and build a multiple uncertainty set for boundary adaptive optimization.

[0217] Specifically, the method of selecting the upper and lower limits of the prediction error by using the cumulative probability density function and the preset confidence, constructing the prediction error fluctuation domain of wind power, energy market capacity clearing price, frequency modulation market capacity clearing price and mileage clearing price, and constructing the multiple uncertainty set of boundary adaptive optimization includes the following steps:

[0218] (1) Based on the cumulative probability density function, the upper and lower limits of the prediction error are selected with a confidence level of 0.95 to construct the prediction fluctuation domain of the uncertain variable;

[0219] The expression of the cumulative probability density function is:

[0220]

[0221] Select the confidence interval:

[0222] e lower =F -1 (0.025)

[0223] e upper =F -1 (0.975)

[0224] Where F(e) is the cumulative probability density function, e is the prediction error, and e lower 、e upper are the lower and upper limits of the forecast error fluctuation, respectively, and F -1 is the inverse function of the cumulative distribution function;

[0225] Construct the predicted volatility domain:

[0226]

[0227] In the formula, is the predicted value at time t;

[0228] (2) Integrate the forecast error fluctuation domains of wind power, energy market capacity clearing price, frequency regulation market capacity clearing price, and mileage clearing price into a multi-uncertainty set;

[0229] Among them, the expression of multiple uncertainty set is:

[0230]

[0231] In the formula, They are wind power, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price. are the upper and lower limits of wind power forecast fluctuation at time τ, are the upper and lower limits of the energy market capacity clearing price at time t, are the upper and lower limits of the frequency modulation market capacity clearing price at time t, They are the upper and lower limits of the mileage clearing price at time t respectively.

[0232] S3. Based on multiple uncertainty sets, a master-slave game optimization model of wind farms and cloud energy storage operators and a multi-objective optimization model of wind power systems considering cloud energy storage and power quality are constructed, and the optimal capacity and rental price of cloud energy storage and the energy-frequency market participation strategy of wind power systems are solved.

[0233] In this embodiment, a master-slave game optimization model of wind farms and cloud energy storage operators is established considering multiple uncertainties, and the optimal capacity and rental price of leasing cloud energy storage are solved with the goals of maximizing the benefits of the main cloud energy storage operator and minimizing the cost of the slave wind farm; a multi-objective optimization model of the wind power system considering cloud energy storage and power quality is established, combined with the preliminary optimized cloud energy storage rental electricity price and rental capacity, with the dual goals of maximizing the benefits of the wind power system and optimizing the power quality of the grid connection point, the energy-frequency modulation market participation strategy of the wind power system under a certain degree of conservatism is solved to improve the market participation benefits and enhance its robustness to deal with uncertainties.

[0234] Among them, the master-slave game optimization model of wind farms and cloud energy storage operators and the multi-objective optimization model of wind power systems considering cloud energy storage and power quality are constructed based on multiple uncertainty sets, and the optimal capacity and rental price of renting cloud energy storage and the energy-frequency market participation strategy of wind power systems are solved, including the following steps:

[0235] S31. Establish a master-slave game optimization model between wind farms and cloud energy storage operators that takes into account multiple uncertainties, and with the goal of maximizing the benefits of the main cloud energy storage operator and minimizing the cost of the subordinate wind farm, solve the optimal capacity and rental price of leasing cloud energy storage;

[0236] Specifically, the master-slave game optimization model between the wind farm and the cloud energy storage operator includes a game subject cloud energy storage operator optimization model and a game slave wind power cluster robust optimization model;

[0237] (1) Optimization model of cloud energy storage operators

[0238] 1) The objective function expression is:

[0239]

[0240]

[0241] In the formula, I c Profits for cloud energy storage operators, To rent out energy storage income, The auction revenue of the frequency modulation market is is the energy market bidding revenue, is the cloud energy storage operating cost, are the power price and capacity price of cloud energy storage for leasing respectively. are the charging and discharging power and capacity of cloud energy storage respectively, Δt is the day-ahead market time scale, is the frequency modulation performance index, Bidding capacity for cloud storage in the frequency regulation market, are the capacity clearing price and mileage clearing price of the frequency modulation market, η is the mileage call rate, is the capacity clearing price for the energy market, is the bidding capacity of cloud storage in the energy market, τ is a certain moment in the real-time market, t is a certain moment in the day-ahead market, They are the cloud energy storage capacity cost and mileage cost respectively;

[0242] 2) Constraints:

[0243] Cloud energy storage aggregator operating capacity constraints:

[0244]

[0245] In the formula, is the rated capacity of cloud energy storage, They are the upper and lower limits of the energy storage state of charge constraints respectively;

[0246] Cloud energy storage initial state and final state constraints:

[0247]

[0248] In the formula, are the charges of the initial and final states of the cloud energy storage respectively;

[0249] The actual energy storage constraints that can be configured for cloud energy storage are:

[0250]

[0251] Where F is the set of all distributed energy storage, θ is a 0-1 decision variable, and C cl(i) is the i-th distributed energy storage capacity;

[0252] (2) Game-based robust optimization model for wind power clusters

[0253] 1) The objective function expression is:

[0254]

[0255] In the formula, C w is the total cost of the wind farm, They are respectively the cost of leasing cloud energy storage for wind power, the penalty cost for wind power deviation, and the opportunity cost. are the bidding income of wind power in the frequency regulation market and the bidding income in the electric energy market, respectively; λ1 and λ2 are the application penalty coefficients in the energy market and the frequency regulation market, respectively. are the day-ahead bidding capacity of wind power in the energy market and frequency regulation market, are the real-time bidding capacity of wind storage in the energy market and frequency regulation market, Δτ is the real-time market time scale, N T is the total time, The bidding price for frequency regulation capacity, is the bidding price for frequency regulation mileage, φ(t) is the time-of-use electricity price in the electricity market, P w.fm.c Declare capacity for wind power cluster, P w.fm.c.max To allow wind power clusters to declare a capacity cap in the frequency regulation market, K W (t) is the frequency modulation performance index value, P w.e.c (t), R w,e,c (t) are the capacity and price of wind power clusters bidding in the energy market;

[0256] 2) Constraints

[0257] Frequency regulation market constraints on wind turbine capacity declaration:

[0258] 0.03P fm.e ≤P w.fm.c (t)≤0.25P fm.e

[0259] Where P w.fm.c (t) is the capacity reported by the wind power cluster, P fm.e Rated capacity for frequency regulation market;

[0260] Wind power leasing cloud energy storage needs to meet the constraints to improve frequency regulation performance indicators:

[0261] K min ≤a i r c (t)×100+b i ≤K w.max

[0262] In the formula, r c (t) is the proportion of rented cloud storage, a i 、b i are the piecewise fitting function coefficients, K min It is the entry threshold of the comprehensive frequency modulation performance index standard for the frequency modulation market, K w.max It is the maximum value of the comprehensive frequency modulation performance index;

[0263] Wind power system wind curtailment constraints:

[0264] P w.fm.c (t)+P w.e.c (t)≥(1-ε W.max ) w (t)

[0265] In the formula, ε W.max is the maximum allowable wind curtailment ratio, P w (t) is the capacity of the wind power system;

[0266] Wind power bidding capacity constraints in the two markets:

[0267] 0≤P w.fm.c (t)+P w.e.c (t)≤P w (t)

[0268] Where P w.fm.c Declare capacity for wind power clusters.

[0269] S32. Establish a multi-objective optimization model for wind power systems that takes cloud energy storage and power quality into consideration, combine the optimized cloud energy storage rental price and rental capacity, take maximizing the benefits of the wind power system and optimizing the power quality at the grid connection point as the dual objectives, and solve the energy-frequency market participation strategy of the wind power system.

[0270] Specifically, the objective functions of the wind power system multi-objective optimization model considering cloud energy storage and power quality include optimal wind power system revenue and optimal wind power system grid-connected power quality;

[0271] (1) The optimal expression for wind power system revenue is:

[0272]

[0273] Where F1 is the wind power system revenue function, is the total revenue of the energy and frequency regulation market, Penalty cost for wind power deviation, For the optimized wind power leasing cloud energy storage cost, is the opportunity cost of the second stage, H is the set of 24 periods in a day, They are the day-ahead bidding capacity for the second phase wind power energy market and the frequency regulation market. They are the real-time bidding capacity of wind storage in the second phase energy market and frequency regulation market, is the wind power frequency regulation performance index, They are the day-ahead bidding capacity of wind power in the energy market and frequency regulation market in the second phase, are the real-time bidding capacity of wind storage in the second phase energy market and frequency regulation market, respectively. w.fm.c.o Declare capacity for the second phase of wind power clusters;

[0274] (2) The optimal expression for the grid-connected power quality of wind power system is:

[0275]

[0276] Where F2 is the power quality function of the wind power system grid-connected power, M is the total number of time nodes throughout the year, and P w (t) is the output power of the wind farm at time t, and Q(t) is the comprehensive evaluation value of power quality at time t.

[0277] (3) Constraints:

[0278] 1) The wind power capacity bid is subject to the following constraints:

[0279]

[0280] In the formula, is the predicted output power of wind power, They are the bid capacities of wind power in the energy and frequency regulation markets a day ago;

[0281] 2) Wind storage real-time bidding capacity:

[0282]

[0283] In the formula, The real-time bidding capacity of the wind storage system in the energy and frequency regulation markets, respectively. The real-time output power of wind power, Discharge capacity for cloud energy storage;

[0284] 3) Wind power output and reserved frequency regulation capacity constraints in the energy market:

[0285]

[0286] In the formula, is the maximum wind power output at time τ;

[0287] 4) Maximum real-time frequency regulation capacity constraints of wind power:

[0288]

[0289] In the formula, is the rated power of wind power, δ is the wind power ramp rate limiting coefficient;

[0290] 5) The maximum capacity constraint that can be used for leased energy storage within the Δk time period:

[0291]

[0292] In the formula, They are respectively the charging and discharging power capacity of cloud energy storage, is the energy storage rated power;

[0293] 6) Power quality index calculation constraints:

[0294]

[0295] Where: P i (t), Q i (t) are the active and reactive power injected into node i at time t, respectively, U i (t), U j (t) are the actual voltages of nodes i and j at time t, G ij , B ij ,θ ij are the conductance, susceptance and power angle between node i and node j respectively, and D is the total number of nodes in the system.

[0296] In summary, with the help of the above technical solutions of the present invention, the present invention combines the improved masked signal method (IMS) with the robust local mean decomposition method (RLMD) to perform time-frequency synchronous decomposition and noise reduction on wind power and multi-dimensional feature data, thereby effectively removing the noise in the data while retaining the main features of the signal. In addition, by introducing the local information enhancement module and the Transformer model with a sparse attention mechanism, the model can enhance the capture of local change characteristics in wind power time series data while retaining the learning of global dependencies. The combined prediction model (IMS-RLMD-LSA-Transformer) improves the accuracy and robustness of wind power forecasting by integrating time-frequency noise reduction with deep learning feature extraction, providing a more reliable data foundation for subsequent market participation strategies.

[0297] In addition, the present invention adopts a new meta-heuristic optimization algorithm based on the concept of weighted average holding to determine the optimal harmonic moment of the density estimation method based on harmonic transformation, which has the characteristics of adaptability, can effectively balance the deviation and variance, and improve the estimation accuracy and reliability. The probability density estimation of the prediction error provides a more accurate way to help better deal with errors and uncertainties in the optimization process, avoiding the limitations of traditional methods. Using the generated fluctuation domain, a multiple uncertainty set of boundary adaptive optimization is constructed, providing a more accurate optimization solution for wind power systems in different complex and dynamic scenarios.

[0298] In addition, the present invention can optimize the rental electricity price and rental capacity between the wind farm and the cloud energy storage by establishing a master-slave game optimization model between the wind farm and the cloud energy storage operator. This game optimization model can not only balance the interests of the wind farm and the cloud energy storage operator, but also effectively reduce the uncertainty caused by wind power fluctuations and improve the overall efficiency of the system. By using a robust optimization method, combined with a multi-objective optimization model that considers cloud energy storage and power quality, the system's energy market and frequency regulation market participation strategies can be optimized under a certain degree of conservatism. It ensures that even in the face of forecast errors and market fluctuations, the system can still achieve the optimal market participation strategy, thereby improving the stability and benefits of market participation.

[0299] 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 principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for a wind power system based on cloud energy storage to participate in the energy-frequency regulation market, characterized in that: The following steps are involved: S1. Obtain the original wind power and multi-dimensional feature data, use the improved masking signal method combined with the robust local mean decomposition method to perform time-frequency synchronous decomposition and noise reduction on the original wind power and multi-dimensional feature data, and use the Transformer model based on local information enhancement and sparse attention mechanism to predict wind power; S2, determine the optimal harmonic moment number of the density estimation method based on harmonic transformation, perform probability density estimation on the prediction error data of power and price, and construct multiple uncertainty sets; S3. Based on multiple uncertainty sets, a master-slave game optimization model of wind farms and cloud energy storage operators and a multi-objective optimization model of wind power systems considering cloud energy storage and power quality are constructed, and the optimal capacity and rental price of cloud energy storage and the energy-frequency market participation strategy of wind power systems are solved.

2. A method for a wind power system based on cloud energy storage to participate in the energy-frequency regulation market according to claim 1, characterized in that: The method of obtaining original wind power and multi-dimensional feature data, performing time-frequency synchronous decomposition and noise reduction on the original wind power and multi-dimensional feature data by using an improved masking signal method combined with a robust local mean decomposition method, and predicting wind power by using a Transformer model based on local information enhancement and a sparse attention mechanism includes the following steps: S11. Obtain original wind power and multidimensional feature data, and decompose the original wind power and multidimensional feature data into several optimal component product function components and residual signals based on the robust local mean decomposition method; use the Spearman correlation coefficient to remove components containing interference signals, and use approximate entropy to select components containing effective features; S12. Use the improved masking signal method to mask the selected components, and reconstruct the optimal component product function components after processing to obtain the denoised wind power signal components; use the Transformer model based on local information enhancement and sparse attention mechanism, combined with the denoised wind power signal components to predict wind power.

3. A method for a wind power system based on cloud energy storage to participate in the energy-frequency modulation market according to claim 2, characterized in that: The method of obtaining original wind power and multidimensional feature data, decomposing the original wind power and multidimensional feature data into a plurality of optimal component product function components and residual signals based on a robust local mean decomposition method; removing components containing interference signals by using a Spearman correlation coefficient, and selecting components containing effective features by using approximate entropy includes the following steps: S111, obtaining original wind power and multi-dimensional feature data, using the mirror extension method to process the boundary to search for all local extreme points in the original wind power signal; obtaining the local mean function and the local envelope function by the sliding average method, and calculating the zero mean signal and the pure frequency signal; The calculation formula for the zero-mean signal is: k 11 (t)=x(t)-m 11 (t) The calculation formula for pure frequency signal is: s 11 (t)=k 11 (t) / a 11 (t) In the formula, k 11 (t) is a zero-mean signal, x(t) is the wind power signal to be decomposed, m 11 (t) is the local mean function, s 11 (t) is the processed pure frequency signal, a 11 (t) is the local envelope function; S112, defining an objective function, extracting the first product function from the original wind power signal to obtain a residual signal, and treating the residual signal as a new signal to repeatedly iterate and extract all optimal component product function components until the residual becomes a constant or a monotonic function; Among them, the expression of the objective function is: f=RMS[z(t)]+EK[z(t)] Where f is the objective function value, RMS is the root mean square, EK is the empirical kurtosis calculation formula, z(t) is the zero baseline envelope signal, is the arithmetic mean of z(t), N s is the total number of signals; S113. Use the Spearman correlation coefficient to remove components containing interference signals, and use approximate entropy to select components containing effective features.

4. A method for a wind power system based on cloud energy storage to participate in the energy-frequency modulation market according to claim 2, characterized in that: The improved masking signal method is used to perform masking processing on the selected components, and the optimal component product function components after processing are reconstructed to obtain the denoised wind power signal components; the Transformer model based on local information enhancement and sparse attention mechanism is used to predict wind power in combination with the denoised wind power signal components, including the following steps: S121, calculating the masked mean signal, combining the Spearman correlation coefficient and the approximate entropy to select several components in the robust local mean decomposition result for masking processing; performing robust local mean decomposition on the masked processing result to obtain the signal component after masking processing, and performing final signal reconstruction based on the signal component after masking processing to obtain the wind power signal component after noise reduction; Among them, the calculation formula of the masked mean signal is: The expression for masking is: The expression of the signal component after masking is: In the formula, s i (t) is the i-th masking signal, is the average amplitude, is the mean value of the instantaneous frequency, PF i+ (t), PF i- (t) are the signal components after averaging and subtracting, respectively, PF i (t) is the original component, IMS_PF i is the signal component after masking processing, pf i+ , pf i- PF i+ (t) and PF i- (t) the first component of the decomposition output; S122, standardizing the wind power signal components after noise reduction, using a fully connected layer to perform high-dimensional representation on the wind power signal components after standardization, and extracting local information of enhanced data by stacking multiple layers of one-dimensional convolution to obtain a locally enhanced signal; S123, using the samples obtained by the multi-layer convolution operation of the locally enhanced signal as the input of the encoder layer, using the position coding to mark the position information of each sample, inputting the position coding result into the encoder layer of the Transformer model, and performing layer normalization processing on the data in the encoder; S124. Use a multi-head sparse attention mechanism to replace the fully connected attention mechanism in the standard Transformer model to capture the global information of the sequence, integrate the global and local information of wind power data, and map and learn the components of wind power through the middle layer of the network, and finally output the wind power prediction results.

5. The method for a wind power system based on cloud energy storage to participate in the energy-frequency regulation market according to claim 1, characterized in that: The method of determining the optimal harmonic moment number of the density estimation method based on harmonic transformation, performing probability density estimation on the prediction error data of power and price, and constructing a multiple uncertainty set includes the following steps: S21. Calculate the forecast errors of wind power, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price respectively, determine the optimal harmonic moment number of the density estimation method based on harmonic transformation by using the weighted average algorithm, and perform probability density estimation on the forecast error data of power and price; S22. Use the cumulative probability density function and preset confidence to select the upper and lower limits of the prediction error, construct the prediction error fluctuation domain of wind power, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price, and build a multiple uncertainty set for boundary adaptive optimization.

6. A method for a wind power system based on cloud energy storage to participate in the energy-frequency regulation market according to claim 5, characterized in that: The method of respectively calculating the forecast errors of wind power, energy market capacity clearing price, frequency modulation market capacity clearing price and mileage clearing price, using a weighted average algorithm to determine the optimal harmonic moment number of the density estimation method based on harmonic transformation, and performing probability density estimation on the forecast error data of power and price includes the following steps: S211, respectively calculating the forecast errors of wind power, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price, where the forecast error is the difference between the forecast value and the true value; S212, initializing parameters, constructing a fitness function of the optimal harmonic moment number based on relative entropy; in each iteration, calculating the weighted average position of the current harmonic moment number, and taking the weighted average position of the current harmonic moment number as the representative position of the entire search space; adopting different mobile exploration strategies, respectively aiming at exploring a wide solution space and fine-tuning the optimal harmonic moment number; Among them, the expression of the fitness function of the optimal harmonic moment number based on relative entropy is: The weighted average position is calculated as: Where Fitness(h) is the fitness function of the optimal harmonic moment number based on relative entropy, f(h) is the probability density function based on harmonic transformation, is the probability density function of the kernel density estimation, h is the number of harmonic moments to be optimized, N Candidate is the number of candidate solutions selected, P is the population size, it is the current iteration number, Max It is the maximum number of iterations, Sum Fitness is the sum of all fitness values ​​of the selected candidates, h i is the i-th candidate optimal harmonic moment number, h Miu is the weighted average position; S213, selecting and adjusting the mobile strategy based on random constants and dynamic parameters related to the iteration number, achieving a balance between exploration and development, obtaining the optimal harmonic moment number, and using a density estimation method based on harmonic transformation to perform probability density estimation on the forecast errors of wind power probability, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price; Among them, the expression of probability density function is: The expression of the probability density function based on harmonic transformation is: In the formula, f e (e) is the probability density function obtained by inverse Mellin transformation, α is a fixed value, δ is an assumed value, e is the prediction error, and f(e) is the probability density function based on harmonic transformation. It is a harmonic transformation.

7. A method for a wind power system based on cloud energy storage to participate in the energy-frequency regulation market according to claim 5, characterized in that: The method of selecting the upper and lower limits of the prediction error by using the cumulative probability density function and the preset confidence, constructing the prediction error fluctuation domains of wind power, energy market capacity clearing price, frequency modulation market capacity clearing price and mileage clearing price, and constructing the multiple uncertainty sets of boundary adaptive optimization includes the following steps: S221. Using the cumulative probability density function and combining the preset confidence level to select the upper and lower limits of the prediction error, construct the prediction fluctuation domain of the uncertain variable; Among them, the expression of the cumulative probability density function is: The confidence interval is: e lower =F -1 (0.025) e upper =F -1 (0.975) The expression for predicting the fluctuation range is: Where F(e) is the cumulative probability density function, e is the prediction error, and e lower 、e upper are the lower and upper limits of the forecast error fluctuation, respectively, and F -1 is the inverse of the cumulative distribution function, is the predicted value at time t; S222, integrating the forecast error fluctuation domains of wind power, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price into a multiple uncertainty set; Among them, the expression of multiple uncertainty set is: In the formula, They are wind power, energy market capacity clearing price, frequency regulation market capacity clearing price and mileage clearing price. are the upper and lower limits of wind power forecast fluctuation at time τ, are the upper and lower limits of the energy market capacity clearing price at time t, are the upper and lower limits of the frequency modulation market capacity clearing price at time t, They are the upper and lower limits of the mileage clearing price at time t respectively.

8. The method for a wind power system based on cloud energy storage to participate in the energy-frequency regulation market according to claim 1, characterized in that: The method of constructing a master-slave game optimization model of wind farms and cloud energy storage operators and a multi-objective optimization model of wind power systems considering cloud energy storage and power quality based on multiple uncertainty sets, and solving the optimal capacity and rental price of renting cloud energy storage and the energy-frequency regulation market participation strategy of the wind power system includes the following steps: S31. Establish a master-slave game optimization model between wind farms and cloud energy storage operators that takes into account multiple uncertainties, and with the goal of maximizing the benefits of the main cloud energy storage operator and minimizing the cost of the subordinate wind farm, solve the optimal capacity and rental price of leasing cloud energy storage; S32. Establish a multi-objective optimization model for wind power systems that takes cloud energy storage and power quality into consideration, combine the optimized cloud energy storage rental price and rental capacity, take maximizing the benefits of the wind power system and optimizing the power quality at the grid connection point as the dual objectives, and solve the energy-frequency market participation strategy of the wind power system.

9. A method for a wind power system based on cloud energy storage to participate in the energy-frequency regulation market according to claim 8, characterized in that: The master-slave game optimization model between the wind farm and the cloud energy storage operator includes a game subject cloud energy storage operator optimization model and a game slave wind power cluster robust optimization model; Among them, the objective function expression of the optimization model of the cloud energy storage operator of the game subject is: The objective function expression of the game-based wind power cluster robust optimization model is: In the formula, I c Profits for cloud energy storage operators, To rent out energy storage income, The auction revenue of the frequency modulation market is is the energy market bidding revenue, is the cloud energy storage operating cost, are the power price and capacity price of cloud energy storage for rent, P t cl , are the charging and discharging power and capacity of cloud energy storage respectively, Δt is the day-ahead market time scale, is the frequency modulation performance index, Bidding capacity for cloud storage in the frequency regulation market, are the capacity clearing price and mileage clearing price of the frequency modulation market, η is the mileage call rate, is the capacity clearing price for the energy market, is the bidding capacity of cloud storage in the energy market, τ is a certain moment in the real-time market, t is a certain moment in the day-ahead market, are respectively the cloud energy storage capacity cost and mileage cost, C w is the total cost of the wind farm, They are respectively the cost of leasing cloud energy storage for wind power, the penalty cost for wind power deviation, and the opportunity cost. are the bidding income of wind power in the frequency regulation market and the bidding income in the electric energy market, respectively; λ1 and λ2 are the application penalty coefficients in the energy market and the frequency regulation market, respectively. are the day-ahead bidding capacity of wind power in the energy market and frequency regulation market, are the real-time bidding capacity of wind storage in the energy market and frequency regulation market, respectively; Δτ is the real-time market time scale, The bidding price for frequency regulation capacity, is the bidding price for frequency regulation mileage, φ(t) is the time-of-use electricity price in the electricity market, P w.fm.c Declare capacity for wind power cluster, P w.fm.c.max To allow wind power clusters to declare a capacity cap in the frequency regulation market, K W (t) is the frequency modulation performance index value, P w.e.c (t), R w,e,c (t) are the bidding capacity and price of wind power cluster in the energy market respectively.

10. A method for a wind power system based on cloud energy storage to participate in the energy-frequency regulation market according to claim 9, characterized in that: The objective functions of the wind power system multi-objective optimization model considering cloud energy storage and power quality include optimal wind power system revenue and optimal wind power system grid-connected power quality; Among them, the optimal expression of wind power system benefit is: The optimal expression for the grid-connected power quality of wind power system is: Where F1 is the wind power system revenue function, is the total revenue of the energy and frequency regulation market, Penalty cost for wind power deviation, For the optimized wind power leasing cloud energy storage cost, is the opportunity cost of the second stage, H is the set of 24 periods in a day, They are the day-ahead bidding capacity for the second phase wind power energy market and the frequency regulation market. They are the real-time bidding capacity of wind storage in the second phase energy market and frequency regulation market, is the wind power frequency regulation performance index, They are the day-ahead bidding capacity of wind power in the energy market and frequency regulation market in the second phase, are the real-time bidding capacity of wind storage in the second phase energy market and frequency regulation market, respectively. w.fm.c.o is the capacity declared by the wind power cluster in the second phase, F2 is the power quality function of the wind power system grid-connected electricity, M is the total number of time nodes throughout the year, and P w (t) is the output power of the wind farm at time t, and Q(t) is the comprehensive evaluation value of power quality at time t.

Images