Method and system for controlling output active power of wind power plant
By extracting the characteristic vectors of the power grid and wind farm and performing classification processing, the output active power of the wind farm is adaptively adjusted, which solves the problem of poor energy management of wind turbines in high permeability power grids, and achieves the optimization of energy management and the stability of grid frequency.
Patent Information
- Application Number
- CN202311540609.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-11-17
- Publication Date
- 2025-05-20
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In high permeability regional power grids, individual differences in wind turbines lead to the inability to optimize energy management when frequency regulation, resulting in waste of energy and may cause fluctuations in the grid frequency.
By obtaining the target energy value sent by the power grid and the output active power value of the wind farm, extracting the frequency adjustment requirement feature vector and the active power feature vector to form a classification feature matrix, and inputting a pre-trained classifier model to obtain instructions to control the output active power of the wind farm.
Adaptive adjustments are achieved based on the target energy value of the power grid and the individual differences of the wind farm, energy management is optimized, energy management is avoided, and energy waste is stabilized to a certain extent.
Smart Images
Figure CN120021125A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of wind power technology, and particularly to a method and system for controlling the active power output of a wind farm. Background Art
[0002] With the continuous increase in the penetration rate of new energy generating units, the safety and stability of wind turbines have attracted extensive attention in the power grid in high-penetration areas. In the actual operation of the power grid, when the power consumption does not match the power supply, it may cause fluctuations in the power grid frequency. Moreover, during the process of wind turbines participating in the system frequency control, due to the individual differences of each wind farm, it is impossible to achieve the optimal energy management during frequency modulation, resulting in energy waste. Summary of the Invention
[0003] This application aims to solve at least one of the technical problems in the related art to some extent.
[0004] To this end, a first aspect of this application proposes a method for controlling the active power output of a wind farm, including:
[0005] Obtaining the target energy values issued by the power grid at multiple time points within a preset time period and the multiple active power output values of multiple wind farms at multiple time points within the preset time period;
[0006] Obtaining a frequency adjustment demand feature vector according to the target energy values issued by the power grid at multiple time points within the preset time period and a first multi-scale neighborhood feature extraction module;
[0007] Obtaining multiple active power feature vectors corresponding to multiple wind farms according to the multiple active power output values of multiple wind farms at multiple time points within the preset time period and a second multi-scale neighborhood feature extraction module;
[0008] Obtaining multiple classification feature matrices corresponding to multiple wind farms according to the frequency adjustment demand feature vector and the multiple active power feature vectors;
[0009] Inputting the multiple classification feature matrices into a pre-trained classifier model to obtain multiple classification results; the classification results are used to indicate whether the active power output of the corresponding wind farm should be increased or decreased; wherein, the classifier model has learned the mapping relationship between the classification feature matrix and the classification result;
[0010] Controlling the active power output of the corresponding wind farm according to the multiple classification results.
[0011] A second aspect of this application proposes a control system for the active power output of a wind farm, including:
[0012] The first acquisition module is used to acquire the target energy values issued by the power grid at multiple time points within a preset time period and the multiple output active power values of multiple wind farms at multiple points within the preset time period;
[0013] The second acquisition module is used to obtain a frequency adjustment demand feature vector according to the target energy values issued by the power grid at multiple time points within the preset time period and the first multi-scale neighborhood feature extraction module;
[0014] The third acquisition module is used to obtain multiple active power feature vectors corresponding to multiple wind farms according to the multiple output active power values of the multiple wind farms at multiple time points within the preset time period and the second multi-scale neighborhood feature extraction module;
[0015] The fourth acquisition module is used to obtain multiple classification feature matrices corresponding to multiple wind farms according to the frequency adjustment demand feature vector and the multiple active power feature vectors;
[0016] The classification module inputs the multiple classification feature matrices into a pre-trained classifier model to obtain multiple classification results; the classification results are used to indicate whether the output active power of the corresponding wind farm should be increased or decreased; wherein, the classifier model has learned the mapping relationship between the classification feature matrix and the classification result;
[0017] The control module is used to control the output active power of the corresponding wind farm according to the multiple classification results.
[0018] The third aspect of the present application proposes a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the program, the method described in the first aspect above is implemented.
[0019] The fourth aspect of the present application proposes a computer-readable storage medium, including a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the program, the method described in the first aspect above is implemented.
[0020] According to the method for controlling the output active power of a wind farm according to the embodiments of the present application, according to the target energy value issued by the power grid and the output active power values of multiple wind farms, the output active power of each wind farm is adaptively adjusted, which not only considers the individual differences of each wind farm, optimizes energy management, avoids energy waste, but also can, to a certain extent, avoid the power grid fluctuations caused by the mismatch between power consumption and power supply.
[0021] Some of the additional aspects and advantages of the present application will be given in the following description, some will become obvious from the following description, or be understood through the practice of the present application. Brief Description of the Drawings
[0022] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description of embodiments in conjunction with the accompanying drawings, wherein:
[0023] Figure 1 It is a schematic flowchart of a method for controlling the active power output of a wind farm provided by an embodiment of the present application;
[0024] Figure 2 It is a schematic flowchart of another method for controlling the active power output of a wind farm provided by an embodiment of the present application;
[0025] Figure 3 It is a schematic diagram of a control system for the active power output of a wind farm provided by an embodiment of the present application;
[0026] Figure 4 It is a structural block diagram of a computer device provided by an embodiment of the present application. Detailed Embodiments
[0027] The embodiments of the present application will be described in detail below. The examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present application, and should not be construed as a limitation of the present application.
[0028] The present application proposes a method and system for controlling the active power output of a wind farm. Specifically, the method and system for controlling the active power output of a wind farm according to the embodiments of the present application will be described below with reference to the accompanying drawings.
[0029] Figure 1 It is a schematic flowchart of a method for controlling the active power output of a wind farm provided by an embodiment of the present application. As Figure 1 shown, the method for controlling the active power output of the wind farm includes the following steps:
[0030] Step 101, obtain the target energy values issued by the power grid at multiple time points within a preset time period and the multiple active power output values of multiple wind farms at multiple time points within the preset time period.
[0031] It should be noted that the target energy value issued by the power grid is the difference between the power consumption and the power supply, and one active power output value corresponds to one time point within the preset time period for one wind farm.
[0032] Step 102, obtain a frequency adjustment demand feature vector according to the target energy values issued by the power grid at multiple time points within the preset time period and the first multi-scale neighborhood feature extraction module.
[0033] In some embodiments of the present application, the target energy values issued by the power grid at multiple time points can be arranged in a time dimension to form an energy value input vector. Then, the energy value input vector is input into the first multi-scale neighborhood feature extraction module to obtain a frequency adjustment requirement feature vector, that is, to extract the dynamic multi-scale neighborhood correlation features of the target energy values issued by the power grid under different time spans.
[0034] As an example, the energy value input vector is input into the first convolutional layer of the first multi-scale neighborhood feature extraction module, and the first neighborhood scale frequency adjustment requirement feature vector is obtained through formula (1). Among them, the first convolutional layer has a first one-dimensional convolutional kernel with a first length. The energy value input vector is input into the second convolutional layer of the first multi-scale neighborhood feature extraction module, and the second neighborhood scale frequency adjustment requirement feature vector is obtained through formula (2). Among them, the second convolutional layer has a second one-dimensional convolutional kernel with a second length, and the first length is different from the second length. The first neighborhood scale frequency adjustment requirement feature vector and the second neighborhood scale frequency adjustment requirement feature vector are cascaded to obtain the frequency adjustment requirement feature vector.
[0035]
[0036] Among them, a is the width of the first convolutional kernel in the x direction, F(a) is the first convolutional kernel parameter vector, G(x - a) is the local vector matrix for operation with the convolutional kernel function, w is the size of the first convolutional kernel, and X is the energy value input vector.
[0037]
[0038] Among them, b is the width of the second convolutional kernel in the x direction, F(b) is the second convolutional kernel parameter vector, G(x - b) is the local vector matrix for operation with the convolutional kernel function, m is the size of the second convolutional kernel, and X is the energy value input vector.
[0039] Step 103, according to the multiple output active power values of multiple wind farms at multiple time points within a preset time period and the second multi-scale neighborhood feature extraction module, obtain multiple active power feature vectors corresponding to the multiple wind farms.
[0040] In some embodiments of the present application, the multiple output active power values of multiple wind farms at multiple time points within a preset time period can be respectively arranged in a time dimension to form multiple active power input vectors corresponding to the multiple wind farms. Then, the multiple active power input vectors are respectively input into the second multi-scale neighborhood feature extraction module to obtain multiple active power feature vectors corresponding to the multiple wind farms. Among them, one wind farm corresponds to one active power feature vector, that is, to extract the dynamic multi-scale neighborhood correlation features of the output active power values of each wind farm under different time spans.
[0041] Step 104: According to the frequency adjustment demand feature vector and multiple active power feature vectors, obtain multiple classification feature matrices corresponding to multiple wind farms.
[0042] As a possible implementation, the transfer matrix of each active power feature vector relative to the frequency adjustment demand feature vector can be calculated respectively, so as to obtain multiple classification feature matrices.
[0043] It should be noted that during the actual operation of the fan, the output active power values of each wind farm are interrelated, and there are hidden correlation pattern features in the power generation capacity and working state of each wind farm. Therefore, in some embodiments of the present application, multiple global context active power feature vectors can be obtained by passing multiple active power feature vectors through a context encoder. And calculate the transfer matrix of each global context active power feature vector relative to the frequency adjustment demand feature vector respectively to obtain multiple responsiveness feature matrices. Determine the multiple responsiveness feature matrices as multiple classification feature matrices corresponding to multiple wind farms.
[0044] As an example, multiple active power feature vectors can be arranged in one dimension to obtain a global active power feature vector. Calculate the product between the global active power feature vector and the transposed vector of each active power feature vector respectively to obtain multiple self-attention correlation matrices. Standardize each self-attention correlation matrix respectively, and obtain multiple probability values by passing each standardized self-attention correlation matrix through the Softmax classification function. Use each probability value in the multiple probability values as a weight to weight each active power feature vector respectively, so as to obtain multiple global context active power feature vectors. Calculate the transfer matrix of each global context active power feature vector relative to the frequency adjustment demand feature vector respectively through formula (3) to obtain multiple responsiveness feature matrices, so as to represent the adaptability feature information between the working state features of each wind farm and the target energy features issued by the power grid.
[0045]
[0046] Among them, V 1 is the global context active power feature vector, M 2 is the transfer matrix of the frequency adjustment demand feature vector, M is the responsiveness feature matrix, represents vector multiplication.
[0047] Step 105: Input the multiple classification feature matrices into a pre-trained classifier model to obtain multiple classification results. The classification results are used to indicate whether the output active power of the corresponding wind farm should be increased or decreased. Among them, the classifier model has learned the mapping relationship between the classification feature matrix and the classification result.
[0048] Step 106, control the active power output of the corresponding wind farm according to multiple classification results.
[0049] According to the method for controlling the active power output of a wind farm according to an embodiment of the present application, the active power output of each wind farm is adaptively adjusted according to the target energy value issued by the power grid and the active power output values of multiple wind farms. It not only considers the individual differences of each wind farm, optimizes energy management, and avoids energy waste, but also can, to a certain extent, avoid power grid fluctuations caused by the mismatch between power consumption and power supply.
[0050] It should be noted that in some embodiments of the present application, after multiple active power feature vectors pass through a context encoder based on a converter to obtain multiple global context active power feature vectors, the context correlation relationship between the multiple global context active power feature vectors is strengthened. However, correspondingly, the feature distribution of the global context active power feature vectors may deviate from the multi-scale temporal correlation distribution of the features extracted by the multi-scale neighborhood feature extraction module, resulting in inconsistent feature distributions between the global context active power feature vectors and the frequency adjustment demand feature vectors, and causing the distribution convergence of the responsiveness feature matrix to be abnormal relative to the overall distribution. As a result, the high-dimensional feature distribution represented by the responsiveness feature matrix may have inductive divergence when transferred to the target domain of the classification problem, affecting the accuracy of the classification results.
[0051] Therefore, the present application also proposes a method for controlling the active power output of a wind farm to optimize the responsiveness feature matrix to improve the accuracy of the classification results. Figure 2 It is a schematic flowchart of another method for controlling the active power output of a wind farm provided by an embodiment of the present application. As Figure 2 shown, the method for controlling the active power output of the wind farm includes the following steps:
[0052] Step 201, obtain the target energy value issued by the power grid at multiple time points within a preset time period and the multiple active power output values of multiple wind farms at multiple time points within the preset time period.
[0053] Step 202, obtain a frequency adjustment demand feature vector according to the target energy value issued by the power grid at multiple time points within the preset time period and the first multi-scale neighborhood feature extraction module.
[0054] Step 203, obtain multiple active power feature vectors corresponding to multiple wind farms according to the multiple active power output values of multiple wind farms at multiple time points within the preset time period and the second multi-scale neighborhood feature extraction module.
[0055] Step 204, pass the multiple active power feature vectors through a context encoder to obtain multiple global context active power feature vectors.
[0056] Step 205: Calculate the transfer matrix of each of the global context active power feature vectors relative to the frequency adjustment demand feature vector, and obtain a plurality of responsiveness feature matrices.
[0057] Step 206: Optimize the high-dimensional data manifolds of the plurality of responsiveness feature matrices respectively to obtain a plurality of classification feature matrices.
[0058] As an example, the high-dimensional data manifolds of each of the responsiveness feature matrices can be optimized through formula (4) to obtain the corresponding classification feature matrices.
[0059]
[0060] where m i，j ′ is the eigenvalue at each position in the classification feature matrix, m i，j is the eigenvalue at each position in the responsiveness feature matrix, W is the width of the responsiveness feature matrix, H is the height of the responsiveness feature matrix, and log represents the logarithmic function value with base 2.
[0061] Step 207: Input the plurality of classification feature matrices into a pre-trained classifier model to obtain a plurality of classification results. The classification results are used to indicate whether the output active power of the corresponding wind farm should be increased or decreased. Among them, the classifier model has learned the mapping relationship between the classification feature matrix and the classification results.
[0062] As an example, the classifier model can obtain the classification results through formula (5) according to the classification feature matrix.
[0063] O = softmax{(W n , B n ):...:(W 1 , B 1 )|Project(F)} (5)
[0064] where Project(F) represents projecting the classification feature matrix into a vector, W 1 to W n are the weight matrices of each fully connected layer, and B 1 to B n represent the bias vectors of each fully connected layer.
[0065] Step 208: Control the output active power of the corresponding wind farm according to the plurality of classification results.
[0066] In the embodiments of the present application, steps 201 - 204 can be implemented in any one of the embodiments of the present application respectively. The present application does not make specific limitations in this regard and will not be elaborated further.
[0067] According to the control method for the active power output of a wind farm according to an embodiment of the present application, based on the target energy value issued by the power grid and the active power output values of multiple wind farms, the active power output of each wind farm is adaptively adjusted. And the responsiveness feature matrix is optimized so that the feature distribution of the responsiveness feature matrix is transferred to the range with a stable and structured boundary under the target domain to improve the accuracy of the classification result. When controlling the active power output of the wind farm, the present application not only considers the individual differences of each wind farm, optimizes the energy management, avoids energy waste, but also can, to a certain extent, avoid the power grid fluctuations caused by the mismatch between power consumption and power supply.
[0068] Figure 3 It is a schematic diagram of a control system for the active power output of a wind farm provided by an embodiment of the present application. As Figure 3 shown, the control system for the active power output of the wind farm includes: a first acquisition module 301, a second acquisition module 302, a third acquisition module 303, a fourth acquisition module 304, a classification module 305, and a control module 306. Among them,
[0069] The first acquisition module 301 is used to acquire the target energy value issued by the power grid at multiple time points within a preset time period and the active power output values of multiple wind farms at multiple points within the preset time period.
[0070] The second acquisition module 302 is used to obtain a frequency adjustment demand feature vector according to the target energy value issued by the power grid at multiple time points within a preset time period and the first multi-scale neighborhood feature extraction module.
[0071] The third acquisition module 303 is used to obtain multiple active power feature vectors corresponding to multiple wind farms according to the active power output values of multiple wind farms at multiple time points within a preset time period and the second multi-scale neighborhood feature extraction module.
[0072] The fourth acquisition module 304 is used to obtain multiple classification feature matrices corresponding to multiple wind farms according to the frequency adjustment demand feature vector and multiple active power feature vectors.
[0073] In some embodiments of the present application, the fourth acquisition module 304 is further used to: pass the multiple active power feature vectors through a context encoder to obtain multiple global context active power feature vectors. Calculate the transfer matrix of each global context active power feature vector relative to the frequency adjustment demand feature vector respectively to obtain multiple responsiveness feature matrices. Determine multiple classification feature matrices according to the multiple responsiveness feature matrices.
[0074] In some embodiments of the present application, multiple classification feature matrices are determined according to multiple responsive feature matrices, including: respectively optimizing the high-dimensional data manifolds of the multiple responsive feature matrices to obtain multiple classification feature matrices.
[0075] The classification module 305 inputs the multiple classification feature matrices into a pre-trained classifier model to obtain multiple classification results. The classification results are used to indicate that the output active power of the corresponding wind farm should be increased or decreased. Among them, the classifier model has learned the mapping relationship between the classification feature matrix and the classification result.
[0076] The control module 306 is configured to control the output active power of the corresponding wind farm according to the multiple classification results.
[0077] Regarding the system in the above embodiments, the specific manners in which each module performs operations have been described in detail in the embodiments related to the method, and will not be elaborated herein.
[0078] According to the control system for the output active power of a wind farm in an embodiment of the present application, according to the target energy value issued by the power grid and the output active power values of multiple wind farms, the output active power of each wind farm is adaptively adjusted. It not only considers the individual differences of each wind farm, optimizes energy management, and avoids energy waste, but also can, to a certain extent, avoid power grid fluctuations caused by the mismatch between power consumption and power supply.
[0079] To implement the above embodiments, the present application also provides a computer device. Figure 4 The following is a block diagram of a computer device provided in an embodiment of the present application. As Figure 4 shown, the computer device 400 may include a memory 401, a processor 402, and a computer program 403 stored on the memory 401 and executable on the processor 402. When the processor 402 executes the computer program 403, it executes the method for controlling the output active power of a wind farm described in any of the above embodiments of the present application.
[0080] To implement the above embodiments, the present application also proposes a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for controlling the output active power of a wind farm described in any of the above embodiments of the present application.
[0081] In the description of this specification, the descriptions referring to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of this application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.
[0082] Any process or method description shown in the flowchart or described in other ways herein can be understood to represent a module, segment, or portion of code including one or more executable instructions for implementing a customized logic function or process, and the scope of the preferred embodiments of this application includes additional implementations, where the functions can be executed in a substantially simultaneous manner or in a reverse order according to the involved functions, rather than in the order shown or discussed, which should be understood by those skilled in the art to which the embodiments of this application pertain.
[0083] It should be understood that each part of this application can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following techniques well known in the art or a combination of them can be used: discrete logic circuits having logic gate circuits for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.
[0084] Those of ordinary skill in the art in this technical field can understand that all or part of the steps carried by the method for implementing the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.
[0085] In addition, each functional unit in various embodiments of this application can be integrated in a processing module, or each unit can exist physically alone, or two or more units can be integrated in one module. The above integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.
[0086] The storage medium mentioned above may be a read-only memory, a magnetic disk, an optical disc, or the like. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.
Claims
1. A method for controlling active power output of a wind farm, characterized in that: The following steps are involved: Acquire target energy values issued by the power grid at multiple time points within a preset time period and multiple output active power values of multiple wind farms at multiple time points within the preset time period; Obtaining a frequency adjustment demand feature vector according to the target energy values issued by the power grid at multiple time points within the preset time period and the first multi-scale neighborhood feature extraction module; According to a plurality of output active power values of the plurality of wind farms at a plurality of time points within the preset time period and a second multi-scale neighborhood feature extraction module, a plurality of active power feature vectors corresponding to the plurality of wind farms are obtained; Obtaining a plurality of classification feature matrices corresponding to a plurality of wind farms according to the frequency adjustment demand feature vector and the plurality of active power feature vectors; Inputting the multiple classification feature matrices into a pre-trained classifier model to obtain multiple classification results; the classification results are used to indicate whether the output functional power of the corresponding wind farm should be increased or decreased; wherein the classifier model has learned a mapping relationship between the classification feature matrix and the classification results; According to the multiple classification results, the output active power of the corresponding wind farm is controlled.
2. The method according to claim 1, characterized in that The step of obtaining a frequency adjustment demand feature vector according to the target energy values issued by the power grid at multiple time points within the preset time period and the first multi-scale neighborhood feature extraction module includes: Arranging the target energy values issued by the power grid at the multiple time points into an energy value input vector according to the time dimension; The energy value input vector is input into the first multi-scale neighborhood feature extraction module to obtain the frequency adjustment requirement feature vector.
3. The method according to claim 1, characterized in that The step of obtaining a plurality of active power feature vectors corresponding to the plurality of wind farms according to the plurality of output active power values of the plurality of wind farms at a plurality of time points within the preset time period and the second multi-scale neighborhood feature extraction module comprises: Arranging a plurality of output active power values of the plurality of wind farms at a plurality of time points within the preset time period into a plurality of active power input vectors corresponding to the plurality of wind farms according to the time dimension; The multiple active power input vectors are respectively input into the second multi-scale neighborhood feature extraction module to obtain multiple active power feature vectors corresponding to the multiple wind farms.
4. The method according to claim 1, characterized in that: The step of obtaining a plurality of classification feature matrices corresponding to a plurality of wind farms according to the frequency adjustment demand feature vector and the plurality of active power feature vectors includes: Passing the multiple active power feature vectors through a context encoder to obtain multiple global context active power feature vectors; Respectively calculating the transfer matrix of each of the global context active power feature vectors relative to the frequency adjustment demand feature vector to obtain a plurality of responsiveness feature matrices; A plurality of classification feature matrices corresponding to the plurality of wind farms are determined according to the plurality of responsiveness feature matrices.
5. The method according to claim 4, characterized in that Determining the plurality of classification feature matrices according to the plurality of responsiveness feature matrices comprises: The high-dimensional data manifolds of the multiple responsiveness feature matrices are optimized respectively to obtain the multiple classification feature matrices.
6. The method according to claim 5, characterized in that The high-dimensional data manifold of each of the responsiveness feature matrices is optimized by the following formula to obtain the corresponding classification feature matrix. Among them, m i,j ′ is the eigenvalue of each position in the classification feature matrix, m i,j ′ is the eigenvalue of each position in the responsiveness characteristic matrix, W is the width of the responsiveness characteristic matrix, h is the height of the responsiveness characteristic matrix, and log represents the logarithmic function value with base 2.
7. A control system for outputting active power from a wind farm, characterized in that: include: A first acquisition module is used to acquire target energy values issued by the power grid at multiple time points within a preset time period and multiple output active power values of multiple wind farms at multiple points within the preset time period; A second acquisition module, configured to obtain a frequency adjustment demand feature vector according to the target energy values issued by the power grid at multiple time points within the preset time period and the first multi-scale neighborhood feature extraction module; A third acquisition module, configured to obtain a plurality of active power feature vectors corresponding to the plurality of wind farms according to a plurality of output active power values of the plurality of wind farms at a plurality of time points within the preset time period and a second multi-scale neighborhood feature extraction module; A fourth acquisition module, configured to obtain a plurality of classification feature matrices corresponding to a plurality of wind farms according to the frequency adjustment demand feature vector and the plurality of active power feature vectors; A classification module, inputting the plurality of classification feature matrices into a pre-trained classifier model to obtain a plurality of classification results; the classification results are used to indicate whether the output functional power of the corresponding wind farm should be increased or decreased; wherein the classifier model has learned a mapping relationship between the classification feature matrix and the classification results; A control module is used to control the output active power of the corresponding wind farm according to the multiple classification results.
8. The device according to claim 7, characterized in that The fourth acquisition module is also used for: Passing the multiple active power feature vectors through a context encoder to obtain multiple global context active power feature vectors; Respectively calculating the transfer matrix of each of the global context active power feature vectors relative to the frequency adjustment demand feature vector to obtain a plurality of responsiveness feature matrices; The plurality of classification feature matrices are determined based on the plurality of responsiveness feature matrices.
9. The device according to claim 8, characterized in that Determining the plurality of classification feature matrices according to the plurality of responsiveness feature matrices comprises: The high-dimensional data manifolds of the multiple responsiveness feature matrices are optimized respectively to obtain the multiple classification feature matrices.
10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 6 is implemented.
11. A computer-readable storage medium, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 6 is implemented.