Cyclic error improved wind speed prediction super-capacitor grid-connected method and related products

Abstract

The application discloses a circulating error improved wind speed prediction super-capacity energy storage grid-connected method and related products, and belongs to the technical field of wind power generation. The method is characterized in that: the original wind speed sequence of a wind farm is obtained and is divided into a training set and a test set; a prediction model is inputted with the training set to obtain a predicted value, and an error coefficient is calculated accordingly; the original wind speed sequence is corrected by using the error coefficient, a wind speed correction sequence and a first-order residual sequence are generated, and the residual sequence is iteratively corrected until the residual error coefficient is less than a preset value; the wind speed correction sequence and all-order residual sequences are inputted into the prediction model to be superimposed to obtain a final wind speed prediction value; and finally, the prediction value is used to determine the shortage of the super-capacity energy storage system, and a charging and discharging plan is formulated to realize efficient dispatching of the power grid. The application continuously optimizes the wind speed prediction result through the circulating error correction mechanism, and provides a reliable basis for accurate grid-connected dispatching of the super-capacity energy storage system.

Description

Technical Field

[0001] The present invention belongs to the technical field of wind power generation, and relates to a super-capacity energy storage grid-connected method for cyclic error improved wind speed prediction and related products. Background Art

[0002] As global demand for renewable energy continues to grow, the development and utilization of wind power, a clean, renewable energy source, has become a crucial component of energy strategies worldwide. The construction and grid connection of large-scale wind farms not only helps reduce greenhouse gas emissions but also promotes the diversification of the energy mix. However, the inherent characteristics of wind power—the randomness, intermittency, and uncontrollability of wind speed—pose unprecedented challenges to the safe and stable operation of power grids.

[0003] In the grid dispatch process, accurately predicting wind farm output power is key to ensuring grid supply and demand balance, optimizing resource allocation, and improving system stability. Traditionally, neural network-based prediction methods have been widely used for wind speed and wind power forecasting. These methods construct complex network models to learn the mapping relationship between historical wind speed data and wind power, thereby predicting future wind speed and wind power. Although neural networks excel at nonlinear modeling, due to the complexity and uncertainty of wind speed variations, relying solely on neural network direct prediction often fails to achieve ideal prediction accuracy. In particular, prediction errors can increase significantly under extreme weather conditions.

[0004] Furthermore, traditional forecasting methods lack an effective closed-loop feedback mechanism. When there is a significant discrepancy between predicted and actual values, the forecasting model cannot be adjusted and optimized in a timely manner, resulting in forecasts that fail to accurately reflect the wind farm's actual generating capacity. This not only complicates grid scheduling but can also lead to frequency fluctuations, voltage instability, and other issues, impacting the overall performance of the power system. Summary of the Invention

[0005] The purpose of the present invention is to solve the technical problem in the prior art that inaccurate wind speed prediction affects power grid performance, and to provide a super-capacity energy storage grid-connected method and related products with cyclic error improved wind speed prediction.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A first aspect of the present invention provides a method for grid-connecting super-capacity energy storage using cyclic error improved wind speed prediction, comprising the following steps:

[0008] S1, obtain the original wind speed sequence of the wind farm operation; divide the original wind speed sequence into a training set and a test set;

[0009] S2, input the training set into the prediction model to obtain a predicted value, and calculate an error coefficient based on the predicted value and the test set;

[0010] S3, correct the original wind speed sequence based on the error coefficient to generate a wind speed correction sequence and a first-order residual sequence, wherein =1;

[0011] S4, input the first-order residual sequence into the prediction model to obtain a residual error coefficient, correct the first-order residual sequence to obtain a second-order residual sequence, and generate a first-order residual correction sequence and a second-order residual sequence; iterate S4 until the residual error coefficient is less than a preset value;

[0012] S5, input the wind speed correction sequence and all the residual sequences into the prediction model to obtain a final wind speed prediction value;

[0013] S6, determine an excess capacity of the energy storage system based on the final wind speed prediction value, and develop a charging and discharging plan to dispatch the power grid according to the excess capacity of the energy storage system.

[0014] Further, the prediction model is an SVR model.

[0015] Further, the calculation formula of the error coefficient is:

[0016]

[0017] wherein, MAPE is the average absolute percentage error of the predicted value and the test set; and MSE is the mean square error of the predicted value and the test set.

[0018] Further, the wind speed correction sequence is specifically:

[0019]

[0020] wherein, is the value of the wind speed correction sequence at the i th position; and the correction equation is ;

[0021] the balance coefficient ; is the value of the original wind speed sequence at the i th position; ; is the error coefficient; SELU and are activation functions; represents a sine function; is a hyperbolic tangent function; is Euler's number; ​​​​​​​is pi; sin() represents the sine function;

[0022] =1, The order residual sequence is described as:

[0023]

[0024] in, is the original wind speed sequence; is the wind speed correction sequence.

[0025] Further, the The order residual sequence is input into the prediction model to obtain the residual error coefficient, which is specifically:

[0026] Will The residual sequence of order is divided into The training set and The test set of the residual sequence of order ;

[0027] Will The training set of the order residual sequence is input into the prediction model, and we get The predicted value of the residual sequence of order; based on The predicted value and The test set of the order residual sequence is used to obtain the residual error coefficient; the residual error coefficient is:

[0028]

[0029] in, for The predicted value of the residual sequence of order The mean absolute percentage error of the test set of the residual sequence of order ; for The mean square error between the predicted value of the order residual sequence and the test set of the residual sequence.

[0030] Furthermore, the The order residual correction sequence is:

[0031]

[0032] in, for i order residual correction sequence j The value of the position; the correction equation is

[0033] The balance coefficient is ; for i Order residual sequence j The value of the position; is the residual error coefficient; SELU and Both are activation functions; tan() is a sine function; e is the Euler number; sin() represents the sine function.

[0034] Further, the The method for generating the order residual sequence is:

[0035]

[0036] in, for Order residual correction series; is the original wind speed sequence; is the wind speed correction sequence; express Order residual sequence.

[0037] Furthermore, the preset value is half of the error coefficient.

[0038] A second aspect of the present invention provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the super-capacity energy storage grid-connected method for cyclic error-improved wind speed prediction is implemented.

[0039] A third aspect of the present invention provides a super-capacity energy storage grid-connected system with cyclic error improved wind speed prediction, comprising:

[0040] The data acquisition module obtains the original wind speed sequence of the wind farm operation; divides the original wind speed sequence into a training set and a test set;

[0041] The error coefficient calculation module inputs the training set into the prediction model to obtain the predicted value, and calculates the error coefficient based on the predicted value and the test set;

[0042] The residual sequence generation module corrects the original wind speed sequence based on the error coefficient to generate the wind speed correction sequence and Order residual sequence, at this time =1;

[0043] The residual correction module will The order residual sequence is input into the prediction model to obtain the residual error coefficient, and the correction The residual sequence of order is obtained Order residual correction sequence, and generate Order residual sequence; iterate S4 until the residual error coefficient is less than the preset value;

[0044] Wind speed prediction module, wind speed correction sequence and all The final wind speed prediction value is obtained by superimposing the prediction model of the residual error sequence input;

[0045] The scheduling module determines the power shortage of the overcapacity energy storage system based on the final wind speed prediction value, and formulates the charging and discharging plan to schedule the power grid according to the power shortage of the overcapacity energy storage system.

[0046] Compared with the prior art, the present application has the following beneficial effects:

[0047] The application discloses an overcapacity energy storage grid-connected method for wind speed prediction with cycle error improvement. The method divides the original wind speed sequence into a training set and a test set, calculates an error coefficient in combination with a prediction model, and iteratively corrects the original wind speed sequence and the residual error sequence until the residual error coefficient is less than a preset value, thereby continuously reducing the prediction deviation. The multi-round error correction mechanism can peel off the error components in the data layer by layer, so that the wind speed correction sequence and the residual error sequence are closer to the actual wind speed variation law, and the accuracy of the wind speed prediction value obtained by superimposing the prediction model is greatly improved. Based on the high-precision final wind speed prediction value, the power shortage of the overcapacity energy storage system can be accurately judged. The charging and discharging plan formulated based on the above can realize efficient cooperation between the energy storage system and the power grid. When the wind speed is too high and the wind power output is excessive, the energy storage system can be reasonably arranged to store the excess power. When the wind speed is low and the wind power output is insufficient, the energy storage system can be timely discharged to supplement the gap, thereby avoiding the waste of energy storage resources or power supply shortage caused by inaccurate prediction in traditional scheduling, and improving the operation efficiency and economy of the overcapacity energy storage system.

[0048] The circulating error improvement type wind speed prediction super-capacity energy storage grid-connected system provided by the application has a data acquisition module providing basic data support for the whole system, accurately dividing the original wind speed sequence into a training set and a test set, ensuring the data reliability of subsequent error calculation and model training, and laying a data foundation for improving the prediction accuracy. The error coefficient calculation module obtains the predicted value by inputting the training set into the prediction model, and calculates the error coefficient combined with the test set, which can accurately capture the core features of the prediction deviation. The residual sequence generation module corrects the original wind speed sequence based on the error coefficient to generate a wind speed correction sequence and a first-order residual sequence, realizes the preliminary stripping of the original data error, makes the wind speed data closer to the actual situation, and creates conditions for further improving the prediction accuracy; the residual correction module processes the residual sequence through a circulating iteration method, continuously obtains the residual error coefficient and corrects the residual sequence until the residual error coefficient is less than the preset value. This continuous deepening error correction mechanism can layer by layer excavate and reduce the error components in the residual error, greatly improve the quality of the residual sequence, and thus provide a strong guarantee for the final accurate prediction. The wind speed prediction module inputs the wind speed correction sequence and all order residual sequences into the prediction model for superposition, fully integrates the correction results of each link, realizes the comprehensive and accurate prediction of the wind speed, and the final wind speed prediction value output by the module provides a high-precision decision basis for subsequent scheduling. The scheduling module determines the shortage of the super-capacity energy storage system based on the final wind speed prediction value and formulates the charging and discharging plan, which can realize the efficient collaborative operation of the super-capacity energy storage system and the power grid. Through accurate scheduling, the influence of wind power output fluctuation on the power grid is effectively suppressed, the energy utilization efficiency of the super-capacity energy storage system is improved, the stability and reliability of the power grid operation are enhanced, and the accommodation capacity of the entire power system to new energy is also improved. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0050] Figure 1 The figure is a flow chart of the circulating error improvement type wind speed prediction super-capacity energy storage grid-connected method of the application.

[0051] Figure 2 The figure is a flow chart of the circulating error improvement type wind speed prediction super-capacity energy storage grid-connected method of the application.

[0052] Figure 3 The figure is a block diagram of the circulating error improvement type wind speed prediction super-capacity energy storage grid-connected system of the application.

[0053] Wherein: 301-data acquisition module; 302-error coefficient calculation module; 303-residual sequence generation module; 304-residual correction module; 305-wind speed prediction module; 306-scheduling module. DETAILED DESCRIPTION

[0054] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some but not all of the embodiments of the present application. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0055] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative labor are within the scope of protection of the present application.

[0056] It should be noted that: similar reference numerals and letters represent similar items in the following drawings, therefore, once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0057] Noun explanation:

[0058] 1, SVR, support vector regression, English name is Support Vector Regression;

[0059] 2, MAPE, mean absolute percentage error, English name is Mean Absolute Percentage Error;

[0060] 3, RAM, random access memory, English name is Random Access Memory;

[0061] 4, ROM, read-only memory, English name is Read-Only Memory;

[0062] 5, EEPROM, electrically erasable programmable read-only memory, English name is Electrically Erasable Programmable Read-Only Memory;

[0063] 6, CPU, central processing unit, English name is Central Processing Unit;

[0064] 7. GPU, Graphics Processing Unit

[0065] 8. FPGA, Field Programmable Gate Array

[0066] 9. USB, flash drive, English name is Universal Serial Bus;

[0067] 10. HDMI, High Definition Multimedia Interface, English name is High Definition Multimedia Interface.

[0068] The present invention is described in further detail below with reference to the accompanying drawings:

[0069] See also Figure 1 The present invention discloses a method for grid-connecting super-capacity energy storage using cyclic error improved wind speed prediction, comprising the following steps:

[0070] S1, obtain the original wind speed sequence of the wind farm operation;

[0071] Use wind speed measurement equipment (such as anemometers) installed in the wind farm to continuously collect wind speed data, with a recording interval set to every 10 minutes. The collected wind speed data is arranged in chronological order to form the original wind speed sequence.

[0072] S2, input the training set into the prediction model to obtain the predicted value, and calculate the error coefficient based on the predicted value and the test set; the prediction model described in this embodiment is the SVR model.

[0073] S3, based on the error coefficient, corrects the original wind speed sequence to generate the wind speed correction sequence and Order residual sequence, at this time =1; at this time, the wind speed correction sequence and the first-order residual sequence are generated.

[0074] S4, will The order residual sequence is input into the prediction model to obtain the residual error coefficient, and the correction The residual sequence of order is obtained Order residual correction sequence, and generate ( ) order residual sequence; iterate S4 until the residual error coefficient is less than the preset value;

[0075] First, the first-order residual sequence is input into the prediction model to generate the residual error coefficient. When the residual error coefficient is greater than or equal to the preset value, the correction Order residual sequence, obtain the 1st order residual correction sequence, and generate the 2nd order residual sequence;

[0076] The generated 2-order residual sequence is input into the prediction model to generate a residual error coefficient, and when the residual error coefficient is greater than or equal to a preset value, the 2-order residual sequence is corrected to obtain a 2-order residual correction sequence, and a 3-order residual sequence is generated;

[0077] The above iteration is iterated until the residual error coefficient is less than the preset value, and the iteration is stopped. At each iteration, a ( ) order residual sequence is generated; all ( ) order residual sequences generated by iteration are stored for next calculation.

[0078] S5, the wind speed correction sequence and all order residual sequences are input into the prediction model to superimpose to obtain a final wind speed prediction value;

[0079] S6, based on the predicted wind speed, determine the shortage that the overcapacity energy storage system needs to bear; based on the shortage, control the dispatching of the overcapacity energy storage system; based on the final predicted wind speed, combined with the power generation capacity curve of the wind farm, calculate the expected power generation of the wind farm in a future period of time. Compared with the load demand of the power grid, the power shortage that the overcapacity energy storage system needs to bear is determined. According to the size of the shortage, the charging and discharging plan of the overcapacity energy storage system is made. When the wind power output is higher than the demand of the power grid, the excess power is stored; when the wind power output is insufficient, the stored power is released to supplement the demand of the power grid.

[0080] Referring to Figure 2 , the application discloses an overcapacity energy storage grid-connected method for a cycle error improved wind speed prediction, comprising the following steps:

[0081] S1, obtaining an original wind speed sequence of a wind farm operation; using a wind speed measuring device (such as an anemometer) installed in the wind farm to continuously collect wind speed data, and the time interval is set to record once every 15 minutes. The collected wind speed data is arranged in time sequence to form an original wind speed sequence.

[0082] S2, dividing the original wind speed sequence into a training set and a test set; dividing the original wind speed sequence into a training set and a test set according to a certain proportion (such as 90% training set and 10% test set).

[0083] S3, inputting the training set into the prediction model to obtain a prediction value; based on the prediction value and the test set, an error coefficient is obtained; selecting a suitable prediction model, and the SVR model is selected as the prediction model in this embodiment.

[0084] The error coefficient is:

[0085]

[0086] Among them, MAPE is the mean absolute percentage error between the predicted value and the test set; MSE is the mean square error between the predicted value and the test set.

[0087] S4, based on the error coefficient, the original wind speed sequence is corrected to obtain the wind speed correction sequence and Order residual sequence, at this time =1;

[0088] The original wind speed sequence is corrected based on the error coefficient to obtain the wind speed correction sequence; specifically:

[0089]

[0090] in, Wind speed correction sequence The value of the position; the correction equation is: ;

[0091] The balance coefficient is: ; is the original wind speed sequence The value of the position; is the error coefficient; SELU and sigmoid All are activation functions; sin() represents the sine function;

[0092] Based on the wind speed correction sequence and the original wind speed sequence, the first-order residual sequence is obtained:

[0093]

[0094] in, is the original wind speed sequence; is the wind speed correction sequence.

[0095] S5, will The order residual sequence is input into the prediction model to obtain the residual error coefficient; The order residual sequence is corrected to obtain Order residual correction sequence, and generate The order residual sequence is iterated until the residual error coefficient is less than the preset value. Specifically:

[0096] S501, will The residual sequence of order is divided into The training set and The test set of the residual sequence of order ;

[0097] S502, will The training set of the order residual sequence is input into the prediction model, and we get The predicted value of the residual sequence of order; based on the predicted value of the first-order residual sequence and a test set of the first-order residual sequence, to obtain a residual error coefficient; the residual error coefficient is:

[0098]

[0099] wherein, the predicted value of the first-order residual sequence and the mean absolute percentage error of the test set of the first-order residual sequence; the predicted value of the first-order residual sequence and the mean square error of the test set of the first-order residual sequence.

[0100] S503, when the residual error coefficient is greater than or equal to a preset value, the first-order residual sequence is corrected to obtain a first-order residual correction sequence, and at this time the predicted value of the first-order residual sequence and , specifically:

[0101]

[0102] wherein, the value of the first-order residual correction sequence j position; the correction equation is:

[0103] ; the balance coefficient is: ; the value of the first-order residual sequence i position; j the residual error coefficient; and SELU are all activation functions; sigmoid

[0104] based on the first-order residual correction sequence, a first-order residual sequence is obtained:

[0105]

[0106] wherein, the first-order residual correction sequence; the original wind speed sequence; the wind speed correction sequence; indicates the first-order residual sequence. The above process is iterated until the residual error coefficient is less than a preset value; in this embodiment, the preset value is half of the error coefficient.

[0107]

[0108] ​​​​​​​​S6, input the wind speed correction sequence and all order residual sequences into the prediction model to obtain the final predicted wind speed. Specifically, the wind speed correction sequence and all order residual sequences are input into the prediction model to predict respectively and then superimposed to obtain the final predicted wind speed. S6, input the wind speed correction sequence and all order residual sequences into the prediction model to obtain the final predicted wind speed. Specifically, the wind speed correction sequence and all order residual sequences are input into the prediction model to predict respectively and then superimposed to obtain the final predicted wind speed. S6, input the wind speed correction sequence and all order residual sequences into the prediction model to obtain the final predicted wind speed. Specifically, the wind speed correction sequence and all order residual sequences are input into the prediction model to predict respectively and then superimposed to obtain the final predicted wind speed.

[0109] S7, based on the final predicted wind speed, determine the shortage that the overcapacity energy storage system needs to bear; based on the shortage, control and dispatch the overcapacity energy storage system; based on the final predicted wind speed, combined with the power generation capacity curve of the wind farm, calculate the expected power generation of the wind farm in a future period of time. Compared with the load demand of the power grid, the power shortage that the overcapacity energy storage system needs to bear is determined. According to the size of the shortage, the charging and discharging plan of the overcapacity energy storage system is made. When the wind power output is higher than the demand of the power grid, the excess power is stored; when the wind power output is insufficient, the stored power is released to supplement the demand of the power grid.

[0110] In order to verify the advantages of the cycle error improved wind speed prediction method in practical application, the method proposed in the present embodiment is compared with the general prediction method (hereinafter referred to as "traditional method") through comparative experiments. By calculating the average absolute error percentage (MAPE) of the same group of wind speed sequences predicted by the two methods, the significant effect of the present invention in improving the prediction accuracy is intuitively displayed. Two representative wind farm original wind speed sequences are selected as experimental data, named wind speed sequence 1 and wind speed sequence 2 respectively. The experimental results are as follows:

[0111]

[0112] From the experimental results, it can be seen that whether it is wind speed sequence 1 or wind speed sequence 2, the cycle error improved wind speed prediction method proposed in the present invention has a lower MAPE value, i.e. higher prediction accuracy. Specifically, for wind speed sequence 1, the MAPE value of the present invention is reduced by about 2.8 percentage points compared with the traditional method; for wind speed sequence 2, it is reduced by about 4.5 percentage points. This result shows that the method of the present invention has stronger adaptability and robustness when dealing with complex and variable wind speed data, and can more effectively capture the rules of wind speed changes, thereby providing more accurate prediction results.

[0113] Through the comparative experiments of the present embodiment, the significant advantages of the cycle error improved wind speed prediction method in improving the prediction accuracy of wind speed are fully verified. The method not only has high prediction accuracy, but also has strong practicality and generalizability, which is of great significance for promoting the sustainable development of wind power and improving the stability of power grid operation.

[0114] One embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned cyclic error improved wind speed prediction super-capacity energy storage grid-connected method when executing the computer program; the device includes:

[0115] Memory: used for storing operating systems, application programs, data files, etc. In this embodiment, the memory also stores a computer program that implements the cyclic error improved wind speed prediction algorithm. The memory can be a combination of volatile memory (such as RAM) and non-volatile memory (such as ROM, EEPROM, etc.) to ensure data persistence and high-speed access.

[0116] Processor: as the core processing unit of the electronic device, responsible for executing the computer program stored in the memory. The processor can be a single-core or multi-core CPU, or a special processing chip such as GPU, FPGA, etc. to meet the efficient computing needs of complex algorithms.

[0117] Input and output interface: including but not limited to network interface, USB interface, HDMI interface, etc., used for data transmission and communication with external devices (such as wind speed measurement devices, energy storage system controllers, grid dispatching centers, etc.).

[0118] Power system: provides stable power supply for the electronic device to ensure continuous operation of the device.

[0119] The computer program is written for the cyclic error improved wind speed prediction super-capacity energy storage grid-connected method, which contains all the logic and instructions for algorithm implementation. When the processor executes the program, it will automatically complete the steps of the cyclic error improved wind speed prediction super-capacity energy storage grid-connected method.

[0120] Referring to Figure 3 , the present application discloses a cyclic error improved wind speed prediction super-capacity energy storage grid-connected system, including:

[0121] Data acquisition module 301 acquires the original wind speed sequence of the wind farm operation; divides the original wind speed sequence into a training set and a test set;

[0122] Error coefficient calculation module 302 inputs the training set into the prediction model to obtain the predicted value, and calculates the error coefficient based on the predicted value and the test set;

[0123] Residual sequence generation module 303 corrects the original wind speed sequence based on the error coefficient to generate a wind speed correction sequence and order residual sequence, at this time =1;

[0124] Residual correction module 304 corrects The residual error sequence is input into a prediction model to obtain residual error coefficients, and the residual error coefficients are corrected The residual error sequence is obtained The residual error correction sequence is obtained, and The residual error sequence is obtained; iteration S4 is performed until the residual error coefficients are less than a preset value

[0125] The wind speed prediction module 305 superimposes the wind speed correction sequence and all The residual error sequence is input into a prediction model to obtain residual error coefficients, and the residual error coefficients are corrected

[0126] The scheduling module 306 determines the lack of the overcapacity energy storage system based on the final wind speed prediction value, and formulates a charging and discharging plan to dispatch the power grid according to the lack of the overcapacity energy storage system.

[0127] The above is only a preferred embodiment of the present application and is not used to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for grid-connecting supercapacity energy storage using cyclic error improved wind speed prediction, characterized in that: The following steps are involved: S1, obtain the original wind speed sequence of the wind farm operation; divide the original wind speed sequence into a training set and a test set; S2, input the training set into the prediction model to obtain the predicted value, and calculate the error coefficient based on the predicted value and the test set; S3, based on the error coefficient, corrects the original wind speed sequence to generate the wind speed correction sequence and Order residual sequence, at this time =1; the wind speed correction sequence is specifically: in, Wind speed correction sequence Value of position; correction equation ; Balance coefficient ; is the original wind speed sequence The value of the position; is the error coefficient; SELU and Both are activation functions; represents the sine function; is the hyperbolic tangent function; is the Euler number; is pi; sin() represents the sine function; =1, The order residual sequence is described as: in, is the original wind speed sequence; is the wind speed correction sequence; S4, will The order residual sequence is input into the prediction model to obtain the residual error coefficient, and the correction The residual sequence of order is obtained Order residual correction sequence, and generate Order residual sequence; iterate S4 until the residual error coefficient is less than the preset value; Among them, the The order residual sequence is input into the prediction model to obtain the residual error coefficient, which is specifically: Will The residual sequence of order is divided into The training set and The test set of the residual sequence of order ; Will The training set of the order residual sequence is input into the prediction model, and we get The predicted value of the residual sequence of order; based on The predicted value and The test set of the order residual sequence is used to obtain the residual error coefficient; the residual error coefficient is: in, for The predicted value of the residual sequence of order The mean absolute percentage error of the test set of the residual sequence of order ; for The mean square error between the predicted value of the order residual sequence and the test set of the residual sequence; described The order residual correction sequence is: in, for order residual correction sequence The value of the position; the correction equation is The balance coefficient is ; for i Order residual sequence j The value of the position; is the residual error coefficient; SELU and Both are activation functions; tan() is a sine function; e is the Euler number; sin() represents the sine function; described The method for generating the order residual sequence is: in, for Order residual correction series; is the original wind speed sequence; is the wind speed correction sequence; express Order residual sequence; S5, wind speed correction sequence and all The order residual sequence is input into the prediction model and superimposed to obtain the final wind speed prediction value; S6, determining the excess capacity of the energy storage system based on the final wind speed forecast value, and formulating a charging and discharging plan to dispatch the power grid according to the excess capacity of the energy storage system.

2. The method for super-capacity energy storage grid connection based on cyclic error improved wind speed prediction according to claim 1, characterized in that: The prediction model is an SVR model.

3. The method for super-capacity energy storage grid connection based on cyclic error improved wind speed prediction according to claim 1, characterized in that: The calculation formula of the error coefficient is: Among them, MAPE is the mean absolute percentage error between the predicted value and the test set; MSE is the mean square error between the predicted value and the test set.

4. The method for super-capacity energy storage grid connection based on cyclic error improved wind speed prediction according to claim 1, characterized in that: The preset value is half of the error coefficient.

5. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the super-capacity energy storage grid-connected method for cyclic error improved wind speed prediction according to any one of claims 1 to 4 is implemented.

6. A super-capacity energy storage grid-connected system with cyclic error improved wind speed prediction, based on the super-capacity energy storage grid-connected method with cyclic error improved wind speed prediction according to claim 1, characterized in that: include: Data acquisition module, which obtains the original wind speed sequence of the wind farm operation; Divide the original wind speed sequence into training set and test set; The error coefficient calculation module inputs the training set into the prediction model to obtain the predicted value, and calculates the error coefficient based on the predicted value and the test set; The residual sequence generation module corrects the original wind speed sequence based on the error coefficient to generate the wind speed correction sequence and Order residual sequence, at this time =1; The residual correction module will The order residual sequence is input into the prediction model to obtain the residual error coefficient, and the correction The residual sequence of order is obtained Order residual correction sequence, and generate Order residual sequence; iterate S4 until the residual error coefficient is less than the preset value; Wind speed prediction module, wind speed correction sequence and all The order residual sequence is input into the prediction model and superimposed to obtain the final wind speed prediction value; The dispatching module determines the excess capacity of the energy storage system based on the final wind speed forecast value, and formulates a charging and discharging plan to dispatch the power grid according to the excess capacity of the energy storage system.

Images