Power prediction method, system and equipment based on wind power plant super-capacity energy storage and medium

Abstract

The invention relates to the technical field of wind power plant power prediction, in particular to a power prediction method, system and device based on wind power plant super-capacity energy storage and a medium. The method comprises the following steps: acquiring actual output power value data of a wind power plant at historical moments to form a data set; constructing a prediction model, and performing training based on the data prediction model in the data set; acquiring output power value data of the wind power plant at the current moment, inputting the output power value data of the wind power plant at the current moment into the trained prediction model, and predicting the power value of the wind power plant at the next moment; wherein the prediction model is constructed by adopting a regularized aggregation movement algorithm. The prediction method can accurately predict the output power value of the next moment.

Description

Technical Field

[0001] The present invention relates to the technical field of wind farm power prediction, and particularly relates to a power prediction method, system, device and medium based on super-capacity energy storage in a wind farm. Background Art

[0002] Wind farms are new energy sources that are currently under development. However, due to their volatility and intermittency, they have an adverse impact on the power grid. Therefore, short-term power prediction is carried out through wind farms to improve the level of wind power consumption. Energy storage devices can suppress the power fluctuation of wind farms, help new energy power stations for consumption, peak regulation, frequency modulation and stable output, and thus reduce energy loss. Therefore, they are widely used in wind farms. Its basic principle is that energy storage devices (such as battery packs, supercapacitors or pumped-storage power stations) store excess electric energy during the peak of wind power generation and release electric energy during the shortage of wind power generation or the peak of grid demand, so as to balance power supply and demand and improve the stability and reliability of the power grid.

[0003] Currently, the method of power prediction is to utilize the non-linear mapping ability and self-learning ability of neural networks (such as BP neural network, LSTM, etc.), and establish a complex relationship between wind power and parameters such as meteorological conditions and equipment status through training the model to predict wind power. However, different prediction models have limitations in dealing with complex data and non-linear relationships, which may lead to inaccurate prediction results and thus affect the stable operation of wind farms. Summary of the Invention

[0004] The present invention provides a power prediction method, system, device and medium based on super-capacity energy storage in a wind farm, aiming to solve the problem of inaccurate power prediction in current wind farms.

[0005] The object of the present invention is achieved by the following technical solutions: In the first aspect, the present invention provides a power prediction method based on super-capacity energy storage in a wind farm, including: Obtain the output power value data of the wind farm at the current moment, input the output power value data of the wind farm at the current moment into the trained prediction model, predict the power value at the next moment of the wind farm at the current moment, and obtain the predicted power value at the next moment; Utilize the predicted power value at the next moment and the actual output power value at the historical moment, input them into the trained prediction model, and predict the output power value within a set time period; Use the probability error compensation coefficient to correct the predicted output power value within the set time period to obtain the power prediction result.

[0006] As a further improvement of the present invention, the training method of the trained prediction model includes: Obtain the actual output power value data at historical moments of the wind farm to form a data set; Construct a prediction model, and train the prediction model based on the data in the data set to obtain a trained prediction model, where the prediction model is constructed using the regularized aggregation moving algorithm. As a further improvement of the present invention, the regularized aggregation moving algorithm is:

[0007] In the formula, y m is the predicted power value at the m-th moment, y m-i is the output power value at the (m - i)-th previous moment, y m-n is the output power value at the (m - n)-th previous moment, s i is the slope coefficient corresponding to the output power value at the (m - i)-th previous moment, i = 1, 2, 3,..., n, where n is the total number of actual output power value data in the data set, and the slope coefficient is:

[0008] In the formula, Mid() is the average value of the slopes, ... respectively represent the time intervals between two output power values.

[0009] As a further improvement of the present invention, the method for determining the total number n of actual output power value data in the data set is: First, judge according to the output power value data at historical moments from m - 4 to m - 1. When the discrimination condition is met, the number of output power value data in the data set is 4. If not, use the output power value data at historical moments from m - 2 to m - 5 for judgment. If the discrimination condition is met, the number of output power value data is 5. Otherwise, continue to judge according to the discrimination condition until the discrimination condition is met. The discrimination condition is:

[0010] In the formula, is the output power value data at the (m - n)-th previous moment in history, is the output power value data at the (m - (n - 1))-th previous moment in history, is the output power value data at the (m - (n - 2))-th previous moment in history, is the output power value data at the (m - (n - 3))-th previous moment in history, m is the prediction moment, and n is the number of output power value data in the data set, n ≥ 4.

[0011] As a further improvement of the present invention, the output power values predicted for a set time period are corrected by using the probability error compensation coefficient to obtain a power prediction result. The probability error compensation coefficient is multiplied by each of the output power values predicted for the set time period to obtain the power prediction result.

[0012] As a further improvement of the present invention, the probability error compensation coefficient is:

[0013] In the formula, Q is the probability error compensation coefficient, a is the proportionality coefficient when the prediction error ratio is greater than 1, is the prediction error ratio corresponding to the m - 10th moment, is the prediction error ratio corresponding to the m - 9th moment, is the prediction error ratio corresponding to the m - 1st moment, is the prediction error ratio corresponding to the mth moment, is the predicted power value at the mth moment, is the predicted power value at the m - 1st moment, is the predicted power value at the m - 10th moment, is the actual output power value at the m - 10th moment, is the actual output power value at the mth moment.

[0014] As a further improvement of the present invention, after obtaining the power prediction result, it further includes controlling the capacitive energy storage system in the wind farm to charge and discharge according to the power prediction result, and performing power compensation on the wind farm.

[0015] In a second aspect, the present invention also provides a power prediction system based on super - capacitor energy storage in a wind farm, including: A first prediction module, which acquires the output power value data of the wind farm at the current moment, inputs the output power value data of the wind farm at the current moment into the trained prediction model, predicts the power value at the next moment of the wind farm at the current moment, and obtains the predicted power value at the next moment; A second prediction module, which inputs the predicted power value at the next moment and the actual output power values at historical moments into the trained prediction model, and predicts the output power values for a set time period; A prediction correction module, which corrects the predicted output power values for the set time period by using the probability error compensation coefficient to obtain a power prediction result.

[0016] In a third aspect, the present invention also provides an electronic device, including a processor and a memory. The processor is used to execute the computer program stored in the memory to implement the power prediction method based on super - capacitor energy storage in a wind farm as described above.

[0017] Fourthly, the present invention also provides a computer-readable storage medium storing at least one instruction, and when the at least one instruction is executed by a processor, the power prediction method based on the over-capacity energy storage of a wind farm as described above is implemented.

[0018] The beneficial effects of the present invention are as follows: In the power prediction method based on the over-capacity energy storage of a wind farm of the present invention, by inputting the output power value data of the wind farm at the current moment into the trained prediction model to predict the power value at the next moment, the dynamic changes of the wind farm power can be captured in a timely manner using real-time data. This short-term prediction based on real-time data helps to more accurately grasp the power change trend in the next extremely short period of time and reduce the prediction error caused by sudden changes. Using the probability error compensation coefficient to correct the output power value of the set time period obtained by prediction further improves the prediction accuracy.

[0019] Furthermore, the present invention can accurately predict the output power at the next moment through the regularized aggregation moving algorithm. This algorithm can make full use of the time series characteristics in historical data to capture the fluctuation rules and trends of wind power. This algorithm is suitable for processing data with time series characteristics, such as wind power data. Compared with the existing prediction methods, by obtaining the output power value data of the wind farm at the current moment and inputting it into the trained prediction model, the real-time prediction of the power value at the next moment of the wind farm can be realized. This real-time prediction ability is of great significance for the dispatching and operation of the power system. By adopting the regularized aggregation moving algorithm and an appropriate model training process, the prediction accuracy of wind power can be significantly improved. Accurate prediction results help the power system better arrange the power generation plan and power dispatching, and reduce the operation cost and risk. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0021] Figure 1 It is a schematic flowchart of the power prediction method based on the over-capacity energy storage of a wind farm in the embodiment of the present invention; Figure 2 It is a schematic structural diagram of the power prediction system based on the over-capacity energy storage of a wind farm in the embodiment of the present invention; Figure 3 It is a schematic structural diagram of an electronic device in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] To make the objectives and technical solutions of the present invention clearer and easier to understand, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. The specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0023] The concept of the present invention is to provide a power prediction method, system, device and medium based on super-capacity energy storage in a wind farm, mainly including: Obtain the output power value data of the wind farm at the current moment, input the output power value data of the wind farm at the current moment into the trained prediction model, predict the power value at the next moment of the wind farm at the current moment, and obtain the predicted power value at the next moment; Use the predicted power value at the next moment and the actual output power value at the historical moment, input them into the trained prediction model, and predict the output power value within a set time period; Use the probability error compensation coefficient to correct the predicted output power value within the set time period to obtain the power prediction result.

[0024] Through the prediction model of this method, the output power value at each moment can be accurately predicted, and then the output power value within a time period can be predicted. Using the probability error compensation coefficient to correct the predicted output power value within the set time period further improves the prediction accuracy.

[0025] Next, the technical solution of the present invention will be clearly and completely described with reference to the accompanying drawings and specific embodiments. Among them, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0026] Embodiment 1 As Figure 1 shown, this embodiment provides a power prediction method based on super-capacity energy storage in a wind farm, and the prediction method adopts the following specific implementation manners.

[0027] First, obtain the output power value data of the wind farm at the current moment, input the output power value data of the wind farm at the current moment into the trained prediction model, and predict the power value at the next moment of the wind farm at the current moment to obtain the predicted power value at the next moment.

[0028] Specifically, obtain the actual output power value data of the wind farm at the historical moment to form a data set. In this embodiment, when predicting the output power value at the m-th moment, it is necessary to obtain the actual output power value data at the previous m - n moments, and then n actual output power value data form a data set.

[0029] Then, the method for determining the number n of output power value data in the dataset is as follows: First, judge according to the output power value data at historical moments m-4 to m-1. When the discrimination condition is satisfied, the number of output power value data in the dataset is 4. If not, judge using the output power value data at moments m-2 to m-5. If the discrimination condition is satisfied, the number of output power value data is 5. Otherwise, continue to judge according to the discrimination condition until the discrimination condition is satisfied. The discrimination condition is:

[0030] In the formula, is the output power value data at the historical moment m-n, is the output power value data at the historical moment m-(n-1), is the output power value data at the historical moment m-(n-2), is the output power value data at the historical moment m-(n-3). m is the prediction moment, n is the number of output power value data in the dataset, and n≥4. For example, when n is 5, the corresponding discrimination condition is:

[0031] When this discrimination condition is satisfied, the corresponding n is 5. When not satisfied, let n = 6 for discrimination.

[0032] Construct a prediction model and train it based on the data in the dataset. The prediction model in this embodiment is constructed using the regularized aggregation moving algorithm.

[0033] Among them, the regularized aggregation moving algorithm is mainly:

[0034] In the formula, y m is the predicted power value at the mth moment, y m-i is the output power value at the previous m-i moment, y m-n is the output power value at the previous m-n moment, s i is the slope coefficient corresponding to the output power value at the previous m-i moment, i = 1, 2, 3,..., n, and n is the total number of actual output power value data in the dataset. The slope coefficient is:

[0035] In the formula, Mid() is the average value of the slope, ... respectively represent the time intervals between two output power values. For example, the time interval is the interval between the previous m-i moment and the previous m-i-1 moment.

[0036] During the training process, the output power value y at the previous m - i moments in history is used m-i to predict the output power value y at the previous m - (i + 1) moment. In this way, in this embodiment, n data in the dataset are respectively trained, and corresponding predicted power data are obtained. At the same time, based on the predicted power data at any moment and the actual power data at that moment, the prediction error ratio is obtained, that is: m-(i+1)

[0037] In the formula, is the predicted power value at the mth moment, is the actual output power value at the mth moment.

[0038] In this embodiment, the predicted power value at the next moment and the actual output power value at the historical moment are input into the trained prediction model to predict the output power value within a set time period.

[0039] The probability error compensation coefficient is used to correct the predicted output power value within the set time period to obtain the power prediction result. And this embodiment can perform error correction on the prediction model according to the probability error compensation coefficient calculated based on multiple prediction error ratios. The main method is to predict the output power values at multiple future moments according to the prediction model, and multiply the probability error compensation coefficient by each predicted output power value to complete the final error correction. Specifically, the probability error compensation coefficient is:

[0040] In the formula, Q is the probability error compensation coefficient, a is the proportionality coefficient when the prediction error ratio is greater than 1, is the prediction error ratio corresponding to the m - 10th moment, is the prediction error ratio corresponding to the m - 9th moment, is the prediction error ratio corresponding to the m - 1st moment, is the prediction error ratio corresponding to the mth moment, is the predicted power value at the mth moment, is the predicted power value at the m - 1st moment,..., is the predicted power value at the m - 10th moment, is the actual output power value at the m - 10th moment,..., is the actual output power value at the mth moment. In this embodiment, a refers to the proportion of the prediction error ratios greater than 1 among 10 prediction error ratio data.

[0041] The product between the probability error compensation coefficient and the power value obtained by the prediction model (i.e., the predicted power value at the next moment) is used to eliminate the error of the predicted power value.​

[0042] Assume that the actual output power value of the wind power is [y 1 , y 2 , y 3 ,.., y m (where m is greater than 300), and use the regularized aggregation movement method to calculate the actual output power value [y m-10 , y m-9 , y m-8 ,.., y m , and the obtained predicted values [y` m-10 , y` m-9 , y` m-8 ,.., y` m . The prediction error ratios (y` m-10 - y m-10 ) / y m-10 , (y` m-9 - y m-9 ) / y m-9 , (y` m-8 - y m-8 ) / y m-8 ,.., (y` m - y m ) / y m . Use the prediction error ratio to compensate for the error generated by the predicted value of the wind power output this time.

[0043] By analyzing the power prediction results, calculate the power value that needs to be compensated by the capacitor energy storage system in the wind farm, and then control the charging and discharging of the capacitor energy storage system in the wind farm to perform power compensation on the wind farm, so as to achieve economic optimization.

[0044] Specifically, the capacity that the capacitor energy storage can compensate is limited. If it participates in compensating for the defects of wind power prediction (that is, to achieve the purpose of avoiding fines), it cannot participate in secondary frequency modulation and peak shaving and valley filling to achieve the best economic purpose. In order to achieve the best economy, assume that the power that needs the capacitor energy storage to participate this time is P, where the power that needs to be compensated for power prediction is P1, and the power required for other design economic rewards is P2. If P ≥ P1 + P2, then there is no need to consider the optimization scheme. When P < P1 + P2, the power compensated by the capacitor energy storage system for the wind power is:

[0045] Based on the obtained expected output of the energy storage system of the wind farm, further control the wind farm based on the expected output, so as to achieve economic optimization.

[0046] In order to prove the practicality of the method, simulation verification was carried out for the traditional prediction method and the prediction method provided by the present invention, and the specific verification results are shown in Table 1. The prediction value is measured by using the mean absolute error percentage (Mean Absolute Percentage Error). The results are shown in Table 1: Table 1

[0047] It can be seen from Table 1 that the error percentage of the prediction method of this embodiment is only 12.1, which is more accurate than other prediction methods.

[0048] The prediction method provided by the present invention is further explained below in conjunction with application examples: For example, regarding the power prediction application of a wind farm including super-capacity energy storage, the power prediction method provided by the present invention is implemented and applied, as follows: Data collection and preprocessing: The output power value of a wind farm within a week is selected as the research object. The wind farm power output value data [y1, y2, ..., y300] (unit: kW) is collected in real time at fixed time intervals (such as 15 minutes) through existing equipment such as power sensors and data collectors in the wind farm. The collected data is synchronously transmitted to the data processing server to form a real-time data stream. Based on the collected power output value data, a power prediction experiment is conducted at 10 moments, each moment interval is 1 hour, and the actual power value fluctuates between 0-500kW.

[0049] Verify the validity of the collected real-time data and remove outliers and missing values. If there is missing data, use interpolation methods (such as linear interpolation and polynomial interpolation) to fill it in to ensure the continuity and integrity of the data.

[0050] Traverse forward from the prediction time m=301, first check whether the power output value corresponding to n=4 meets the discrimination condition. If the calculation result is 0.15, it does not meet the discrimination condition, then set n=5 and continue to check whether it meets the discrimination condition. When n=5, the discrimination condition is met, then the data set [y296, y297, y298, y299, y300], the power value in the data set is [190, 200, 180, 190, 200]. The power output value that meets the discrimination condition is used as the data set for the training data prediction model.

[0051] Next, use the regularized aggregate movement algorithm to make predictions:

[0052] Take time interval (Unit: hour). In this experiment, the slope coefficient at each moment is calculated with i=1 corresponding to y300: ; Similarly, ; ; ; = 7.5.

[0053] Therefore, using the regularized aggregation movement algorithm, kW.

[0054] Calculate the prediction error ratio during the training process: [0.05, -0.03, 0.08, -0.02, 0.1, 0.15, -0.05, 0.07, 0.09, -0.01]. The proportion of the prediction error ratio greater than 1 is 0. Therefore, trigger branch. Then, the compensation coefficient is:

[0055] Then, the corrected predicted power value is = 185.25kW.

[0056] If the total available power of the capacitor energy storage system P = 500kW; the power to be compensated for the prediction error P 1 = 300kW, the power required for economic reward P 2 = 300kW. Then, the compensation power is 249.52kW.

[0057] Finally, compare the prediction results of the present invention with those of the traditional prediction model and calculate the mean absolute error percentage. The results show that the error percentage of the prediction method of the present invention is only 12.1, while that of the traditional prediction method reaches 26.4. This effectively verifies the effectiveness and practicality of the present invention.

[0058] By dynamically adjusting the dataset size n and introducing a probabilistic error compensation coefficient Q , combined with the economic power distribution strategy of the energy storage system, this embodiment significantly reduces the wind farm power prediction error and realizes the optimal utilization of energy storage resources.

[0059] Embodiment 2 As Figure 2 shown, a power prediction system based on over-capacity energy storage in a wind farm in this embodiment includes: The first prediction module obtains the output power value data of the wind farm at the current moment, inputs the output power value data of the wind farm at the current moment into the trained prediction model, predicts the power value of the next moment of the wind farm at the current moment, and obtains the predicted power value of the next moment; In this embodiment, the method for determining the number of data collected by the first prediction module is as follows: First, judge according to the output power value data at historical moments m-4 to m-1. When the discrimination condition is satisfied, the number of output power value data in the dataset is 4. If not, judge using the output power value data at moments m-2 to m-5. If the discrimination condition is satisfied, the number of output power value data is 5. Otherwise, continue to judge according to the discrimination condition until the discrimination condition is satisfied. The discrimination condition is:

[0060] In the formula, is the output power value data at the historical moment m-n, is the output power value data at the historical moment m-(n-1), is the output power value data at the historical moment m-(n-2), is the output power value data at the historical moment m-(n-3). m is the prediction moment, n is the number of output power value data in the dataset, and n≥4.

[0061] The second prediction module inputs the predicted power value at the next moment and the actual output power value at the historical moment into the trained prediction model to predict the output power value for a set time period. Among them, the prediction model is constructed using the regularized aggregation moving algorithm: The regularized aggregation moving algorithm is:

[0062] In the formula, y m is the predicted power value at the mth moment, y m-1 is the output power value at the current moment, y m-n is the output power value at the moment m-n, s i is the slope coefficient corresponding to the output power value at the moment m-i, i = 1, 2, 3,..., n. n is the total number of actual output power value data in the dataset. The slope coefficient is:

[0063] In the formula, Mid() is the average value of the slope, ... respectively represent the time intervals between two output power values. The time interval ... is a set value, which is determined according to the time when two output power value data are obtained.

[0064] The prediction correction module corrects the output power values predicted for a set time period by using a probability error compensation coefficient to obtain a power prediction result. This module is constructed based on the probability error compensation coefficient. The error correction step is to multiply the probability error compensation coefficient by each of the output power values predicted for the set time period. The probability error compensation coefficient in this embodiment is calculated based on multiple prediction error ratios, and the probability error compensation coefficient is:

[0065] In the formula, Q is the probability error compensation coefficient, a is the proportionality coefficient when the prediction error ratio is greater than 1, is the prediction error ratio corresponding to the m - 10 moment, is the prediction error ratio corresponding to the m - 9 moment, is the prediction error ratio corresponding to the m - 1 moment, is the prediction error ratio corresponding to the m moment, is the predicted power value at the m moment, is the predicted power value at the m - 1 moment,..., is the predicted power value at the m - 10 moment, is the actual output power value at the m - 10 moment,..., is the actual output power value at the m moment.

[0066] Embodiment 3 As Figure 3 shown, this embodiment provides an electronic device for implementing the power prediction method based on wind farm over - capacity energy storage in Embodiment 1 above. The electronic device 100 includes at least one processor 102, a memory 101, a computer program 103 stored in the memory 101 and executable on at least one processor 102, and at least one communication bus 104. The memory 101 can be used to store the computer program 103. The processor 102 realizes the steps of the power prediction method based on wind farm over - capacity energy storage in Embodiment 1 by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101. The memory 101 mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function (such as a sound playback function, an image playback function, etc.); the data storage area can store data created according to the use of the electronic device 100 (such as audio data, etc.). In addition, the memory 101 can include non - volatile memory, such as a hard disk, a memory, a plug - in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, at least one magnetic disk storage device, a flash memory device, or other non - volatile solid - state storage devices.

[0067] At least one processor 102 may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 102 may be a microprocessor or the processor 102 may also be any conventional processor, etc. The processor 102 is the control center of the electronic device 100 and connects various parts of the entire electronic device 100 through various interfaces and lines.

[0068] The memory 101 in the electronic device 100 stores multiple instructions to implement a power prediction method based on over-capacity energy storage in a wind farm. The processor 102 can execute multiple instructions to implement: Obtain the output power value data of the wind farm at the current moment, input the output power value data of the wind farm at the current moment into the trained prediction model, predict the power value of the next moment of the wind farm at the current moment, and obtain the predicted power value of the next moment; Use the predicted power value of the next moment and the actual output power value of the historical moment, input them into the trained prediction model, and predict the output power value of the set time period; Use the probability error compensation coefficient to correct the predicted output power value of the set time period to obtain the power prediction result.

[0069] Embodiment 4 If the modules / units integrated in the electronic device 100 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, to implement all or part of the processes in the above embodiment methods of the present invention, it can also be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, and Read-Only Memory (ROM).

[0070] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in one or more flows and / or one or more blocks in the flowcharts and / or block diagrams. Figure 1 in one or more flows and / or one or more blocks Figure 1 in one or more blocks

[0071] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means for implementing the functions specified in one or more flows and / or one or more blocks in the flowcharts and / or block diagrams. Figure 1 in one or more flows and / or one or more blocks Figure 1 in one or more blocks

[0072] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more flows and / or one or more blocks in the flowcharts and / or block diagrams. Figure 1 in one or more flows and / or one or more blocks Figure 1 in one or more blocks

[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent replacements can still be made to the specific embodiments of the present invention, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered by the protection scope of the claims of the present invention.

Claims

1. A power prediction method based on wind farm excess energy storage, characterized in that: include: Obtain the output power value data of the wind farm at the current moment, input the output power value data of the wind farm at the current moment into the trained prediction model, predict the power value of the wind farm at the next moment at the current moment, and obtain the predicted power value at the next moment; The predicted power value at the next moment and the actual output power value at the historical moment are input into the trained prediction model to predict the output power value of the set time period; The probability error compensation coefficient is used to correct the output power value of the set time period and obtain the power prediction result.

2. The power prediction method based on wind farm excess energy storage according to claim 1 is characterized in that: The training method of the trained prediction model includes: Obtain the actual output power value data of the wind farm at historical moments to form a data set; A prediction model is constructed, and the prediction model is trained based on the data in the data set to obtain a trained prediction model, wherein the prediction model is constructed by using a regularized aggregated movement algorithm.

3. The power prediction method based on wind farm excess energy storage according to claim 2 is characterized in that: The regularized aggregation movement algorithm is: In the formula, y m is the predicted power value at the mth moment, y m-i is the output power value at the previous mi moment, y m-n is the output power value at the previous mn time, s i is the slope coefficient corresponding to the output power value at the previous mi moment, i=1,2,3,...,n, n is the total number of actual output power value data in the data set, and the slope coefficient is: Where Mid() is the average value of the slope, ... Respectively represent the time interval between two output power values.

4. The power prediction method based on wind farm excess energy storage according to claim 2 is characterized in that: The total number n of actual output power value data in the data set is determined as follows: First, the output power value data at the historical time m-4~m-1 is used for judgment. When the judgment condition is met, the number of output power value data in the data set is 4. If not, the output power value data at the time m-2~m-5 is used for judgment. If the judgment condition is met, the number of output power value data is 5. Otherwise, judgment is continued according to the judgment condition until the judgment condition is met. The judgment condition is: In the formula, is the output power value data at the previous mn moment in history, is the output power value data at the previous m-(n-1) moments in history, is the output power value data at the previous m-(n-2) moments in history, is the output power value data at m-(n-3) moments before the historical time, m is the prediction time, n is the number of output power value data in the data set, and n≥4.

5. The power prediction method based on wind farm excess energy storage according to claim 1 is characterized in that: The probability error compensation coefficient is used to correct the output power value predicted in the set time period to obtain the power prediction result, and the probability error compensation coefficient is multiplied by each output power value predicted in the set time period to obtain the power prediction result.

6. The power prediction method based on wind farm excess energy storage according to claim 5 is characterized in that: The probability error compensation coefficient is: In the formula, Q is the probability error compensation coefficient, a is the proportional coefficient when the prediction error ratio is greater than 1, is the prediction error ratio corresponding to time m-10, is the prediction error ratio corresponding to time m-9, is the prediction error ratio corresponding to time m-1, is the prediction error ratio corresponding to time m, is the predicted power value at time m, is the predicted power value at time m-1, is the predicted power value at time m-10, is the actual output power value at time m-10, is the actual output power value at time m.

7. The power prediction method based on wind farm excess energy storage according to claim 5 is characterized in that: After the power prediction result is obtained, the method further includes controlling the capacitor energy storage system in the wind farm to charge and discharge according to the power prediction result, so as to perform power compensation on the wind farm.

8. A power prediction system based on wind farm excess energy storage, characterized in that: include: The first prediction module obtains the output power value data of the wind farm at the current moment, inputs the output power value data of the wind farm at the current moment into the trained prediction model, predicts the power value of the wind farm at the next moment at the current moment, and obtains the predicted power value at the next moment; The second prediction module uses the predicted power value at the next moment and the actual output power value at the historical moment as inputs into the trained prediction model to predict the output power value for the set time period; The prediction correction module uses the probability error compensation coefficient and the predicted output power value of the set time period to perform correction to obtain the power prediction result.

9. An electronic device, characterized in that: It comprises a processor and a memory, wherein the processor is used to execute a computer program stored in the memory to implement a power prediction method based on excess energy storage in a wind farm as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one instruction, and when the at least one instruction is executed by the processor, the power prediction method based on excess energy storage in a wind farm is implemented as described in any one of claims 1 to 7.

Images