Method and system suitable for wind power prediction
By combining numerical weather forecast data and data flow classification model in wind power prediction and combined with manual intervention, the problem of limited accuracy of existing wind power prediction models is solved, and higher prediction accuracy and reliability are achieved.
Patent Information
- Application Number
- CN202411285384.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-13
- Publication Date
- 2025-02-11
AI Technical Summary
The existing wind power prediction model relies on machine learning methods and lacks automated corrections from manual experience, resulting in limited prediction accuracy. Especially during unstable periods of weather systems, numerical weather forecasts are difficult to accurately track fluctuations in meteorological elements, resulting in relatively high power prediction deviations. big.
Based on the numerical weather forecast data of wind farms, a wind power population prediction model CM1 is established, and a wind power power prediction model is established based on the initial data set through the data flow classification model EMC, and a wind power power prediction model is established based on the initial data set, combining manual intervention and machine learning to improve prediction accuracy.
Through this method, the level of wind power prediction can be effectively improved, model accuracy and reliability can be improved, and more reliable support for the planning, operation and scheduling of wind farms.
Smart Images
Figure CN119416009A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of wind power prediction, and more specifically, to a method and system suitable for wind power prediction. Background Art
[0002] Wind power forecasting is an important part of wind farm planning, operation and dispatching, and is of great significance for ensuring the safe and stable operation of wind farms. At present, wind power forecasting technology has been widely used. The existing new energy power forecasting is mainly based on objective forecasting. The forecasting model is mainly trained by machine learning, and the mapping ability has limitations, mainly including:
[0003] 1) The prediction model only achieves the best global prediction effect based on the training data set, lacks extraction, and occasionally makes huge prediction errors;
[0004] 2) There is a strong correlation between renewable energy output and wind / solar resources. During periods of unstable weather systems, numerical weather forecasts are difficult to accurately track the fluctuations of meteorological elements. The prediction model is based on numerical weather forecasts, which leads to large deviations in power forecasts.
[0005] The improvement of numerical weather forecasting and prediction modeling technology is a slow and continuous process. Under the current development level of numerical weather forecast and prediction models, establishing a hybrid enhanced artificial intelligence prediction model, embedding prior experience and expert cognition into the prediction model, and manually intervening in the objective prediction results are important ways to improve the prediction level of new energy in the short term. It is also a common method adopted by foreign new energy developed countries, domestic and foreign meteorological forecasting, load forecasting and other professionals. The existing wind power prediction model mainly relies on machine learning methods, and manual correction is limited to daily manual correction and curve evaluation. It fails to make full use of manual experience for automatic correction, resulting in limited accuracy of the prediction model. Summary of the invention
[0006] In view of the above problems, the present invention proposes a method suitable for wind power prediction, comprising:
[0007] Based on the numerical weather forecast data of the wind farm, a wind power group prediction model CM1 is established, and a power prediction value within a preset time in the future is predicted according to the wind power group prediction model CM1;
[0008] Taking the power prediction value as an initial data set, and establishing a wind power prediction model based on a data stream classification model EMC and the initial data set;
[0009] Based on the wind power prediction model, the wind power of the wind farm in the future period is predicted.
[0010] Optionally, based on the numerical weather forecast data of the wind farm, a wind power group prediction model CM1 is established, including:
[0011] Gridded numerical weather forecast data covering the spatial geographic location distribution of all wind farms in the group are collected, and the gridded numerical weather forecast data are input into the autoencoder model EM for dimensionality reduction processing. According to the collected total power data and predicted power data of the wind power group, and the gridded numerical weather forecast data with dimensionality reduction processing, the graph convolution model is trained to generate a wind power group prediction model CM1.
[0012] Optionally, the data dimension of the gridded numerical weather forecast data is: M×N×K×T, where M is the number of grids in the latitude direction, N is the number of grids in the longitude direction, K is the number of meteorological elements, and T is the number of time sections;
[0013] Optionally, the gridded numerical weather forecast data is input into an autoencoder model EM for dimensionality reduction processing, including: performing data dimensionality reduction processing on the gridded numerical weather forecast data according to time sections.
[0014] Optionally, after data dimensionality reduction processing, the data dimension of the gridded numerical weather forecast data is reduced from M×N×K×T to 25×K.
[0015] Optionally, based on the data stream classification model EMC, a wind power prediction model is established according to the initial data set, including:
[0016] Based on the initial data set, establishing an initial model;
[0017] Using the K-means method to classify the initial data set into multiple data sets, and extracting feature information of each data set, based on the feature information, determining cluster information;
[0018] Based on the data stream classification model EMC, sample classification is performed on multiple sets of data sets according to the cluster information to obtain multiple classes of samples, and the confidence of each class of samples is determined;
[0019] According to the confidence level of each type of samples, the initial model is updated to generate a wind power prediction model.
[0020] In another aspect, the present invention provides a system suitable for wind power forecasting, comprising:
[0021] A data acquisition unit, used to establish a wind power group prediction model CM1 based on the numerical weather forecast data of the wind farm, and predict the power prediction value within a preset time in the future according to the wind power group prediction model CM1;
[0022] A modeling unit, configured to use the power prediction value as an initial data set, and establish a wind power prediction model based on a data stream classification model EMC and the initial data set;
[0023] The prediction unit is used to predict the wind power of the wind farm in the future period based on the wind power prediction model.
[0024] Optionally, the data acquisition unit establishes a wind power group prediction model CM1 based on the numerical weather forecast data of the wind farm, including:
[0025] Gridded numerical weather forecast data covering the spatial geographic location distribution of all wind farms in the group are collected, and the gridded numerical weather forecast data are input into the autoencoder model EM for dimensionality reduction processing. According to the collected total power data and predicted power data of the wind power group, and the gridded numerical weather forecast data with dimensionality reduction processing, the graph convolution model is trained to generate a wind power group prediction model CM1.
[0026] Optionally, the data dimension of the gridded numerical weather forecast data is: M×N×K×T, where M is the number of grids in the latitude direction, N is the number of grids in the longitude direction, K is the number of meteorological elements, and T is the number of time sections;
[0027] Optionally, the gridded numerical weather forecast data is input into an autoencoder model EM for dimensionality reduction processing, including: performing data dimensionality reduction processing on the gridded numerical weather forecast data according to time sections.
[0028] Optionally, after data dimensionality reduction processing, the data dimension of the gridded numerical weather forecast data is reduced from M×N×K×T to 25×K.
[0029] Optionally, based on the data stream classification model EMC, a wind power prediction model is established according to the initial data set, including:
[0030] Based on the initial data set, establishing an initial model;
[0031] Using the K-means method to classify the initial data set into multiple data sets, and extracting feature information of each data set, based on the feature information, determining cluster information;
[0032] Based on the data stream classification model EMC, sample classification is performed on multiple sets of data sets according to the cluster information to obtain multiple classes of samples, and the confidence of each class of samples is determined;
[0033] According to the confidence level of each type of samples, the initial model is updated to generate a wind power prediction model.
[0034] In yet another aspect, the present invention further provides a computing device, comprising: one or more processors;
[0035] a processor for executing one or more programs;
[0036] When the one or more programs are executed by the one or more processors, the above-described method is implemented.
[0037] In yet another aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed, the method described above is implemented.
[0038] Compared with the prior art, the present invention has the following beneficial effects:
[0039] The present invention provides a method suitable for wind power prediction, including: establishing a wind power group prediction model CM1 based on numerical weather forecast data of a wind farm, and predicting a power prediction value within a preset time in the future according to the wind power group prediction model CM1; using the power prediction value as an initial data set, and establishing a wind power prediction model based on a data stream classification model EMC according to the initial data set; predicting the wind power of the wind farm in the future time period based on the wind power prediction model. The present invention can effectively improve the level of wind power prediction, improve the accuracy and reliability of the model, and provide more reliable support for the planning, operation and scheduling of wind farms. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] Figure 1 is a flow chart of the method of the present invention;
[0041] Figure 2 An update model flow chart of an embodiment of the method of the present invention;
[0042] Figure 3 It is a structural diagram of the system of the present invention. DETAILED DESCRIPTION
[0043] Now, exemplary embodiments of the present invention are described with reference to the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to disclose the present invention in detail and completely and to fully convey the scope of the present invention to those skilled in the art. The terms used in the exemplary embodiments shown in the accompanying drawings are not intended to limit the present invention. In the accompanying drawings, the same units / elements are marked with the same reference numerals.
[0044] Unless otherwise specified, the terms (including technical terms) used herein have the commonly understood meanings to those skilled in the art. In addition, it is understood that the terms defined in commonly used dictionaries should be understood to have the same meanings as those in the context of the relevant fields, and should not be understood as idealized or overly formal meanings.
[0045] Embodiment 1:
[0046] The present invention proposes a method suitable for wind power prediction, such as Figure 1 As shown, including:
[0047] Step 1: Based on the numerical weather forecast data of the wind farm, a wind power group prediction model CM1 is established, and a power prediction value within a preset time in the future is predicted according to the wind power group prediction model CM1;
[0048] Step 2: Using the power prediction value as an initial data set, and establishing a wind power prediction model based on the data stream classification model EMC and the initial data set;
[0049] Step 3: Based on the wind power prediction model, predict the wind power of the wind farm in the future period.
[0050] Among them, based on the numerical weather forecast data of wind farms, a wind power group prediction model CM1 is established, including:
[0051] Gridded numerical weather forecast data covering the spatial geographic location distribution of all wind farms in the group are collected, and the gridded numerical weather forecast data are input into the autoencoder model EM for dimensionality reduction processing. According to the collected total power data and predicted power data of the wind power group, and the gridded numerical weather forecast data with dimensionality reduction processing, the graph convolution model is trained to generate a wind power group prediction model CM1.
[0052] Among them, the data dimension of gridded numerical weather forecast data is: M×N×K×T, where M is the number of grids in the latitude direction, N is the number of grids in the longitude direction, K is the number of meteorological elements, and T is the number of time sections;
[0053] The gridded numerical weather forecast data is input into the autoencoder model EM for dimensionality reduction processing, including: performing data dimensionality reduction processing on the gridded numerical weather forecast data according to time sections.
[0054] Among them, after data dimensionality reduction processing, the data dimension of gridded numerical weather forecast data is reduced from M×N×K×T to 25×K.
[0055] Wherein, based on the data stream classification model EMC, a wind power prediction model is established according to the initial data set, including:
[0056] Based on the initial data set, establishing an initial model;
[0057] Using the K-means method to classify the initial data set into multiple data sets, and extracting feature information of each data set, based on the feature information, determining cluster information;
[0058] Based on the data stream classification model EMC, sample classification is performed on multiple sets of data sets according to the cluster information to obtain multiple classes of samples, and the confidence of each class of samples is determined;
[0059] According to the confidence level of each type of samples, the initial model is updated to generate a wind power prediction model.
[0060] The present invention will be further described below in conjunction with embodiments:
[0061] Establish a wind farm power prediction model based on numerical weather forecast:
[0062] (1) Taking a wind power group consisting of N wind farms as the object, collect gridded numerical weather forecast data covering the spatial geographical distribution of all wind farms in the group, the data dimension is M×N×K×T, where M is the number of grids in the latitude direction, N is the number of grids in the longitude direction, and K is the number of meteorological elements. In the present invention, K=4, T is the number of time sections, and the data at the same time includes the longitude and latitude of the upper left corner of the grid, all meteorological parameters of different grids (including 10m wind speed, 10m wind direction, 100m wind speed, 100m wind direction, a total of 4 parameters), and the data of different time sections are superimposed to form the final data. Based on autoencoding, the gridded numerical weather forecast historical data is subjected to data dimensionality reduction processing according to the time section, and the autoencoder model EM is obtained through training, and the original data dimension is reduced from M×N×K to 25×K, K is the number of meteorological elements, that is, the data dimension of each meteorological element is reduced to 25;
[0063] (2) Collect the total power data and predicted power data of the wind power group. The numerical weather forecast data processed by the autoencoder in the previous step needs to be input. The graph convolution model is used to establish the wind power group prediction model CM1 to obtain the power forecast value for the next ten days.
[0064] Multi-model collaborative optimization based on data stream classification model EMC, such as Figure 2 As shown, including:
[0065] (1) Initial model establishment:
[0066] The power prediction value obtained in step 1 is used as the initial data set D. The K-means method is first used to classify the initial data set D into K groups Ci = {C1, C2, .. Ck}, and then the micro-cluster MC is used to save the feature information of each cluster Ci. The micro-cluster MC is defined as a feature tuple (LS, SS, N, θ, W, T), where LS, SS, and θ represent the linear sum of the samples marked as β in the cluster Ci, respectively. The sum of squares SS = x; the covariance matrix N represents the total number of samples of each class in each cluster, W represents the importance of micro-cluster MC over time, initially set to 1, and T represents the time of micro-cluster MC update, initially set to 0. Based on the feature information, the center C of the cluster can be calculated MC and radius R MC , calculated as shown in formula (1) and (2):
[0067]
[0068] The center and radius of each micro-cluster MC form a hypersphere, and the union of all micro-cluster hyperspheres constitutes the entire decision space.
[0069] (2) Sample classification:
[0070] New samples are input into M in the form of blocks. When the classification model classifies the input sample x, it first determines whether X is an outlier. An outlier is defined as a sample far away from the decision boundary, that is, the distance between the sample and any micro-cluster center is greater than the corresponding micro-cluster radius. The samples are classified according to the cluster structure distance, the purity of the micro-cluster, and the ratio of the classes in each micro-cluster, as shown in equations (3) and (4):
[0071] H(x)=-∑ β∈L p(x) β log p(x) β (3)
[0072]
[0073] in Represents the sample x and the microfamily MC i The cluster structure distance CSD (Cluster Structure Distance) of the center of the sample of category β in is calculated using the Euclidean distance; Pur(MC i ) represents micro cluster MC i The frequency of samples with the highest number of categories, i.e., the purity of the micro-cluster; Indicates Microtribe MC i The ratio of the middle category to be β; represents the k nearest microclusters of sample x.
[0074] (3) Confidence setting:
[0075] In this paper, we design a classifier confidence based on information entropy based on the classification rules based on micro-cluster structure distance and purity, as shown in equations (5) and (6):
[0076] H(x)=-∑ β∈L p(x) β log p(x) β (5)
[0077]
[0078] where p(x) β is the normalized probability that sample x is predicted to be class β, and L is all the categories contained in the nearest micro-cluster. H(x) is the information entropy of all predicted categories, which is used as the confidence of the classifier. The higher the information entropy, the lower the confidence; the lower the information entropy, the higher the confidence.
[0079] (4) Model update:
[0080] Updating the model is divided into two steps:
[0081] The first step is to manually process microclusters with low confidence. Specifically, the sample labels are manually judged. If the classified sample is considered correct, the result is output without correction; otherwise, it is manually corrected and the corrected result is fed back to the model training data.
[0082] The second step is to update the statistics of the micro-cluster. If the sample falls inside a micro-cluster, the statistics of the micro-cluster closest to it are updated using the sample information.
[0083] In terms of wind power forecast error analysis, the present invention analyzes the spatiotemporal characteristics of historical wind power forecast errors, explores the temporal transmission characteristics and spatial distribution characteristics of forecast errors, divides different weather processes, and proposes a typical error scenario division method to achieve visualization of typical error scenarios that match weather processes for manual correction processes, and conducts machine learning based on wind power forecast error characteristics and cause analysis. In terms of power forecast optimization based on human-machine collaboration, the present invention considers manual correction of error characteristics of different power forecast methods, establishes a comprehensive evaluation system for forecast correction, and intelligently optimizes the results of manual correction, studies a method for extracting human-in-the-loop features based on optimal manual correction experience, and constructs a multi-model machine learning technology based on the integration of human-in-the-loop intelligent framework with human-in-the-loop experience, so as to improve the prediction performance of different methods. With the help of cognitive computing, the present invention realizes the integrated fusion of multiple prediction results, repeats the iterative process, and improves the prediction method based on human-in-the-loop and cognitive computing.
[0084] The beneficial effect of the present invention is that it can effectively improve the level of wind power prediction, improve the accuracy and reliability of the model, and provide more reliable support for the planning, operation and scheduling of wind farms.
[0085] Embodiment 2:
[0086] The present invention proposes a system 200 suitable for wind power prediction, such as Figure 1 As shown, including:
[0087] The data acquisition unit 201 is used to establish a wind power group prediction model CM1 based on the numerical weather forecast data of the wind farm, and predict the power prediction value within a preset time in the future according to the wind power group prediction model CM1;
[0088] A modeling unit 202 is used to use the power prediction value as an initial data set, and establish a wind power prediction model based on the data stream classification model EMC and the initial data set;
[0089] The prediction unit 203 is used to predict the wind power of the wind farm in a future period based on the wind power prediction model.
[0090] The data acquisition unit 201 establishes a wind power group prediction model CM1 based on the numerical weather forecast data of the wind farm, including:
[0091] Gridded numerical weather forecast data covering the spatial geographic location distribution of all wind farms in the group are collected, and the gridded numerical weather forecast data are input into the autoencoder model EM for dimensionality reduction processing. According to the collected total power data and predicted power data of the wind power group, and the gridded numerical weather forecast data with dimensionality reduction processing, the graph convolution model is trained to generate a wind power group prediction model CM1.
[0092] Among them, the data dimension of gridded numerical weather forecast data is: M×N×K×T, where M is the number of grids in the latitude direction, N is the number of grids in the longitude direction, K is the number of meteorological elements, and T is the number of time sections;
[0093] The gridded numerical weather forecast data is input into the autoencoder model EM for dimensionality reduction processing, including: performing data dimensionality reduction processing on the gridded numerical weather forecast data according to time sections.
[0094] Among them, after data dimensionality reduction processing, the data dimension of gridded numerical weather forecast data is reduced from M×N×K×T to 25×K.
[0095] Wherein, based on the data stream classification model EMC, a wind power prediction model is established according to the initial data set, including:
[0096] Based on the initial data set, establishing an initial model;
[0097] Using the K-means method to classify the initial data set into multiple data sets, and extracting feature information of each data set, based on the feature information, determining cluster information;
[0098] Based on the data stream classification model EMC, sample classification is performed on multiple sets of data sets according to the cluster information to obtain multiple classes of samples, and the confidence of each class of samples is determined;
[0099] According to the confidence level of each type of samples, the initial model is updated to generate a wind power prediction model.
[0100] The present invention can effectively improve the level of wind power prediction, increase model accuracy and reliability, and provide more reliable support for the planning, operation and scheduling of wind farms.
[0101] Embodiment 3:
[0102] Based on the same inventive concept, the present invention also provides a computer device, which includes a processor and a memory, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding functions, so as to implement the steps of the method in the above embodiment.
[0103] Embodiment 4:
[0104] Based on the same inventive concept, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both a built-in storage medium in a computer device and an extended storage medium supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the steps of the method in the above embodiment.
[0105] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The schemes in the embodiments of the present invention may be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.
[0106] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes 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 a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0107] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0108] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0109] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.
[0110] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.
Claims
1. A method for wind power prediction, characterized in that: include: Based on the numerical weather forecast data of the wind farm, a wind power group prediction model CM1 is established, and a power prediction value within a preset time in the future is predicted according to the wind power group prediction model CM1; Taking the power prediction value as an initial data set, and establishing a wind power prediction model based on a data stream classification model EMC and the initial data set; Based on the wind power prediction model, the wind power of the wind farm in the future period is predicted.
2. The method according to claim 1, characterized in that The wind power group prediction model CM1 is established based on the numerical weather forecast data of the wind farm, including: Gridded numerical weather forecast data covering the spatial geographic location distribution of all wind farms in the group are collected, and the gridded numerical weather forecast data are input into the autoencoder model EM for dimensionality reduction processing. According to the collected total power data and predicted power data of the wind power group, and the gridded numerical weather forecast data with dimensionality reduction processing, the graph convolution model is trained to generate a wind power group prediction model CM1.
3. The method according to claim 2, characterized in that The data dimensions of the gridded numerical weather forecast data are: M×N×K×T, where M is the number of grids in the latitude direction, N is the number of grids in the longitude direction, K is the number of meteorological elements, and T is the number of time sections.
4. The method according to claim 2, characterized in that: The step of inputting the gridded numerical weather forecast data into the autoencoder model EM for dimensionality reduction processing includes: performing data dimensionality reduction processing on the gridded numerical weather forecast data according to time sections.
5. The method according to claim 2, characterized in that: After the data dimension reduction process, the data dimension of the gridded numerical weather forecast data is reduced from M×N×K×T to 25×K.
6. The method according to claim 1, characterized in that The method of establishing a wind power prediction model based on the data stream classification model EMC according to the initial data set includes: Based on the initial data set, establishing an initial model; Using the K-means method to classify the initial data set into multiple data sets, and extracting feature information of each data set, based on the feature information, determining cluster information; Based on the data stream classification model EMC, sample classification is performed on multiple sets of data sets according to the cluster information to obtain multiple classes of samples, and the confidence of each class of samples is determined; According to the confidence level of each type of samples, the initial model is updated to generate a wind power prediction model.
7. A system suitable for wind power forecasting, characterized in that: include: A data acquisition unit, used to establish a wind power group prediction model CM1 based on the numerical weather forecast data of the wind farm, and predict the power prediction value within a preset time in the future according to the wind power group prediction model CM1; A modeling unit, configured to use the power prediction value as an initial data set, and establish a wind power prediction model based on a data stream classification model EMC and the initial data set; The prediction unit is used to predict the wind power of the wind farm in the future period based on the wind power prediction model.
8. The system according to claim 7, characterized in that The data acquisition unit establishes a wind power group prediction model CM1 based on the numerical weather forecast data of the wind farm, including: Gridded numerical weather forecast data covering the spatial geographic location distribution of all wind farms in the group are collected, and the gridded numerical weather forecast data are input into the autoencoder model EM for dimensionality reduction processing. According to the collected total power data and predicted power data of the wind power group, and the gridded numerical weather forecast data with dimensionality reduction processing, the graph convolution model is trained to generate a wind power group prediction model CM1.
9. The system according to claim 8, characterized in that The data dimensions of the gridded numerical weather forecast data are: M×N×K×T, where M is the number of grids in the latitude direction, N is the number of grids in the longitude direction, K is the number of meteorological elements, and T is the number of time sections.
10. The system according to claim 8, characterized in that The step of inputting the gridded numerical weather forecast data into the autoencoder model EM for dimensionality reduction processing includes: performing data dimensionality reduction processing on the gridded numerical weather forecast data according to time sections.
11. The system according to claim 8, characterized in that After the data dimension reduction process, the data dimension of the gridded numerical weather forecast data is reduced from M×N×K×T to 25×K.
12. The system according to claim 7, characterized in that The method of establishing a wind power prediction model based on the data stream classification model EMC according to the initial data set includes: Based on the initial data set, establishing an initial model; Using the K-means method to classify the initial data set into multiple data sets, and extracting feature information of each data set, based on the feature information, determining cluster information; Based on the data stream classification model EMC, sample classification is performed on multiple sets of data sets according to the cluster information to obtain multiple classes of samples, and the confidence of each class of samples is determined; According to the confidence level of each type of samples, the initial model is updated to generate a wind power prediction model.
13. A computer device, characterized in that: include: one or more processors; a processor for executing one or more programs; When the one or more programs are executed by the one or more processors, the method according to any one of claims 1 to 6 is implemented.
14. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed, the method according to any one of claims 1 to 6 is implemented.