Wind power prediction method and system based on frequency domain adaptive hyperparameter optimization

Abstract

The application discloses a wind power prediction method and system based on frequency domain adaptive super parameter optimization, which comprises the following steps: firstly, collecting historical wind power and meteorological data; secondly, proposing a brand-new frequency domain coverage overlap coefficient as a fitness function, combining the grey wolf optimization algorithm to adaptively optimize the super parameter of the variational mode decomposition, and using the optimized parameter to decompose the power sequence; thirdly, calculating the sample entropy of each component and reconstructing it into three categories of randomness, fluctuation and trend according to the entropy value to reduce the complexity; then, screening the strongly correlated meteorological features for each component based on the Pearson correlation coefficient, and inputting them into a prediction model for prediction; finally, superimposing the predicted values of each component to obtain the final power prediction result. The application solves the problem of artificial setting of the VMD super parameter, realizes the balance between the decomposition quality and the prediction efficiency, and significantly improves the accuracy and practicability of the ultra-short-term wind power prediction.

Description

Technical Field

[0001] This invention belongs to the field of power automation technology, and in particular relates to a wind power prediction method and system based on frequency domain adaptive hyperparameter optimization. Background Technology

[0002] Currently, wind power prediction methods can be mainly divided into physical methods, statistical methods, and artificial intelligence methods. Among them, deep learning-based artificial intelligence methods, such as Long Short-Term Memory networks and their variants, are widely used due to their powerful nonlinear time series modeling capabilities. However, the original wind power sequence has high non-stationarity and complexity, and a single deep learning model is insufficient to fully capture its multi-timescale fluctuation characteristics, resulting in a bottleneck in prediction accuracy.

[0003] To improve the processing capability of non-stationary sequences, the "decomposition-prediction-reconstruction" strategy is widely adopted. This strategy first uses signal decomposition techniques to decompose the original power sequence into several relatively stationary subsequences (intrinsic mode functions, IMFs), then models and predicts each subsequence separately, and finally aggregates the prediction results. Variational mode decomposition, as a non-recursive, adaptive signal decomposition method, constructs and solves constrained variational problems, and can adaptively decompose signals into modal components with specific sparsity in the frequency domain. Compared with methods such as empirical mode decomposition, it effectively avoids mode aliasing problems and has stronger decomposition stability.

[0004] However, the decomposition effect of VMD heavily relies on two key hyperparameters: the number of modes K and the penalty factor α. K determines the granularity of the decomposition; too small a value leads to under-decomposition, while too large a value introduces noise and increases computational burden. α controls the bandwidth of each modal component, directly affecting the frequency domain concentration and independence of the components. In practical applications, these parameters are usually manually preset using expert experience or grid search, lacking adaptability and making it difficult to guarantee the optimality of the decomposition results in the frequency domain, thus limiting the performance ceiling of subsequent prediction models. Furthermore, establishing an independent and complex prediction model for each decomposed IMF component would lead to a sharp increase in computational costs, making it difficult to meet the efficiency requirements of engineering applications.

[0005] Therefore, how to adaptively determine the optimal hyperparameters of VMD and reduce model complexity while ensuring prediction accuracy has become a key technical problem that needs to be solved to improve the performance of ultra-short-term wind power prediction. Summary of the Invention

[0006] This invention aims to overcome the shortcomings of existing technologies and provide a wind power prediction method and system based on frequency domain adaptive hyperparameter optimization. This invention designs a novel frequency domain evaluation index to adaptively optimize VMD parameters and performs intelligent reconstruction and differentiated modeling based on component complexity, thereby improving prediction accuracy while effectively controlling computational complexity.

[0007] In a first aspect, the present invention provides a wind power prediction method based on frequency domain adaptive hyperparameter optimization, comprising:

[0008] Historical wind power data of the target area and historical meteorological data associated with the historical wind power data are obtained and divided into training set and test set according to a preset ratio;

[0009] Define a frequency domain coverage overlap coefficient based on the frequency domain, and use the Grey Wolf optimization algorithm to adaptively optimize the total number of modes and penalty factor of variational mode decomposition with the frequency domain coverage overlap coefficient as the optimization objective, so as to obtain the optimal hyperparameter combination.

[0010] Using the variational mode decomposition model configured with the optimal hyperparameter combination, the wind power sequences in the training set and the test set are decomposed to obtain several intrinsic mode function components.

[0011] Calculate the sample entropy values ​​of the aforementioned intrinsic mode function components, and reconstruct all intrinsic mode function components into random components, fluctuation components, and trend components according to the range of sample entropy values.

[0012] Based on the Pearson correlation coefficient, a first feature subset corresponding to the random component, a second feature subset corresponding to the fluctuation component, and a third feature subset corresponding to the trend component are selected from the historical meteorological data, respectively.

[0013] The first feature subset, the second feature subset, and the third feature subset are input into a preset prediction model, and the prediction model outputs the corresponding random component prediction value, fluctuation component prediction value, and trend component prediction value.

[0014] The predicted values ​​of the random component, the fluctuation component, and the trend component are superimposed to obtain the final wind power prediction result.

[0015] Secondly, the present invention provides a wind power prediction system based on frequency domain adaptive hyperparameter optimization, comprising:

[0016] The acquisition module is configured to acquire historical wind power data of the target area and historical meteorological data associated with the historical wind power data, and divide them into training set and test set according to a preset ratio;

[0017] The optimization module is configured to define a frequency domain coverage overlap coefficient based on the frequency domain, and use the Grey Wolf optimization algorithm to adaptively optimize the total number of modes and the penalty factor of the variational mode decomposition with the frequency domain coverage overlap coefficient as the optimization target, so as to obtain the optimal hyperparameter combination.

[0018] The decomposition module is configured to use the variational mode decomposition model configured by the optimal hyperparameter combination to decompose the wind power sequences in the training set and the test set respectively, and obtain several intrinsic mode function components.

[0019] The reconstruction module is configured to calculate the sample entropy values ​​of the plurality of intrinsic mode function components, and reconstruct all intrinsic mode function components into random components, fluctuation components and trend components according to the range of sample entropy values.

[0020] The filtering module is configured to filter out a first feature subset corresponding to the random component, a second feature subset corresponding to the fluctuation component, and a third feature subset corresponding to the trend component from the historical meteorological data based on the Pearson correlation coefficient.

[0021] The prediction module is configured to input the first feature subset, the second feature subset, and the third feature subset into a preset prediction model, and the prediction model outputs the corresponding random component prediction value, fluctuation component prediction value, and trend component prediction value.

[0022] The overlay module is configured to overlay the predicted values ​​of the random component, the predicted values ​​of the fluctuation component, and the predicted values ​​of the trend component to obtain the final wind power prediction result.

[0023] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the wind power prediction method based on frequency domain adaptive hyperparameter optimization according to any embodiment of the present invention.

[0024] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the steps of the wind power prediction method based on frequency domain adaptive hyperparameter optimization according to any embodiment of the present invention.

[0025] The wind power prediction method and system based on frequency domain adaptive hyperparameter optimization proposed in this application have the following specific advantages:

[0026] A frequency domain coverage overlap coefficient is proposed as a fitness function for VMD hyperparameter optimization. This function can simultaneously evaluate the completeness and modal independence of the decomposition, guide the Grey Wolf optimization algorithm to find the globally optimal combination of K and α in the frequency domain, overcome the blindness of manual parameter tuning, and significantly improve the quality of VMD decomposition.

[0027] By calculating the sample entropy of the IMF and reconstructing it into three components—random, fluctuating, and trend—the number of subsequences that need to be modeled independently is significantly reduced while preserving the multi-scale features of the sequence. This effectively reduces the overall computational complexity and improves prediction efficiency.

[0028] Different prediction models are selected in a differentiated manner to address the fluctuation characteristics of different reconstructed components, and highly relevant specific meteorological features are screened for them. This achieves synergistic optimization of "data characteristics-feature input-model structure", thereby comprehensively improving the final prediction accuracy of ultra-short-term wind power and providing more reliable data support for the precise scheduling of the power grid. Attached Figure Description

[0029] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 A flowchart illustrating a wind power prediction method based on frequency domain adaptive hyperparameter optimization, provided in an embodiment of the present invention;

[0031] Figure 2 This is a structural block diagram of a wind power prediction system based on frequency domain adaptive hyperparameter optimization, provided in an embodiment of the present invention.

[0032] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Please see Figure 1 The diagram shows a flowchart of a wind power prediction method based on frequency domain adaptive hyperparameter optimization according to this application.

[0035] like Figure 1 As shown, the wind power prediction method based on frequency domain adaptive hyperparameter optimization specifically includes the following steps:

[0036] Step S101: Obtain historical wind power data of the target area and historical meteorological data associated with the historical wind power data, and divide them into training set and test set according to a preset ratio;

[0037] Step S102: Define the frequency domain coverage overlap coefficient based on the frequency domain, and use the Grey Wolf optimization algorithm to adaptively optimize the total number of modes and the penalty factor of variational mode decomposition with the frequency domain coverage overlap coefficient as the optimization target, so as to obtain the optimal hyperparameter combination.

[0038] In this step, the formula for calculating the frequency domain coverage overlap coefficient is:

[0039] ,

[0040] In the formula, This represents the maximum value in the frequency range after the Fast Fourier Transform. This represents the total number of modes in the VMD decomposition. Let be the amplitude of the k-th order IMF component at frequency ω. Let be the amplitude of the (k+1)th order IMF component at frequency ω. This represents the amplitude of wind power at frequency ω.

[0041] Specifically, the gray wolf population is initialized, and the optimization interval between the total number of modes and the penalty factor is set. In each iteration, the frequency domain coverage overlap coefficient under the hyperparameter combination corresponding to the current gray wolf position is calculated. The gray wolf population is updated according to the position update mechanism of the gray wolf optimization algorithm, and the historical best solution is recorded. When the preset maximum number of iterations is reached, the hyperparameter combination that minimizes the value of the frequency domain coverage overlap coefficient is output as the optimal hyperparameter combination.

[0042] Step S103: Using the variational mode decomposition model configured with the optimal hyperparameter combination, the wind power sequences in the training set and the test set are decomposed to obtain several intrinsic mode function components.

[0043] In this step, a constrained variational problem is constructed. The objective of this constrained variational problem is to minimize the sum of the estimated bandwidths of all intrinsic mode function components. The constraint condition of the constrained variational problem is that the sum of all components equals the original power sequence, expressed as:

[0044] ,

[0045] In the formula, For the decomposed IMF set, To decompose the center frequency set of the IMF, The number of IMFs in the decomposition. For time Partial derivative operations, Pi The imaginary unit, For the first IMF, For the first The center frequency of the first-order IMF, This is the original wind power sequence;

[0046] By introducing a quadratic penalty factor and Lagrange multipliers, the constrained variational problem is transformed into an augmented Lagrangian function, expressed as:

[0047] ,

[0048] In the formula, For Lagrange multipliers, The penalty coefficient is... Let be the Lagrange multiplier function in the time domain. It is the Dirac function;

[0049] The augmented Lagrange function is solved using the alternating direction multiplier method. The center frequency and amplitude of each eigenmode function component, as well as the Lagrange multipliers, are alternately updated in the frequency domain until the convergence condition is met, yielding the decomposed eigenmode function components, expressed as:

[0050] ,

[0051] In the formula, For the first During the nth iteration Fourier transform of the first-order IMF The Fourier transform of the original wind power sequence For the first During the nth iteration Fourier transform of the first-order IMF For the first During the nth iteration Fourier transform of the first-order IMF For the first Fourier transform of the Lagrange multipliers in the next iteration The penalty coefficient is... For the first During the nth iteration The center frequency of the first-order IMF, For the first Fourier transform of Lagrange multipliers during iteration For the first Fourier transform of the Lagrange multipliers in the next iteration Noise tolerance.

[0052] Step S104: Calculate the sample entropy values ​​of the plurality of intrinsic mode function components, and reconstruct all intrinsic mode function components into random components, fluctuation components and trend components according to the range of sample entropy values.

[0053] In this step, a first entropy threshold and a second entropy threshold are set, wherein the first entropy threshold is less than the second entropy threshold;

[0054] The intrinsic mode function components whose sample entropy value is greater than the second entropy threshold are classified as random components.

[0055] The intrinsic mode function components whose sample entropy values ​​are between the first entropy threshold and the second entropy threshold are classified as fluctuation components.

[0056] The intrinsic mode function components whose sample entropy value is less than the first entropy threshold are classified as trend components.

[0057] Step S105: Based on the Pearson correlation coefficient, select a first feature subset corresponding to the random component, a second feature subset corresponding to the fluctuation component, and a third feature subset corresponding to the trend component from the historical meteorological data.

[0058] In this step, the Pearson correlation coefficient between each meteorological feature in the historical meteorological data and the current component is calculated, wherein the current component is the random component, the fluctuation component, or the trend component;

[0059] All meteorological features whose absolute Pearson correlation coefficient is greater than a preset threshold are selected to form a feature subset corresponding to the current component.

[0060] Step S106: Input the first feature subset, the second feature subset, and the third feature subset into a preset prediction model. The prediction model outputs the corresponding predicted values ​​of random components, fluctuation components, and trend components.

[0061] In this step, the preset prediction models include: a long short-term memory network model for predicting the trend component, a bidirectional long short-term memory network model for predicting the fluctuation component, and a bidirectional long short-term memory network model with an attention mechanism for predicting the random component.

[0062] Step S107: The predicted values ​​of the random component, the predicted values ​​of the fluctuation component, and the predicted values ​​of the trend component are superimposed to obtain the final wind power prediction result.

[0063] In summary, the method of this application, by defining a novel frequency domain coverage overlap coefficient as the fitness function and coupling it with the Grey Wolf optimization algorithm, achieves adaptive, globally optimal search for the key hyperparameters of variational mode decomposition (total number of modes K and penalty factor α). The optimized variational mode decomposition can decompose the original non-stationary power sequence into several intrinsic mode function components with sufficient frequency domain feature expression and strong inter-mode independence. Furthermore, by calculating the sample entropy value of each component and intelligently reconstructing it into three types of components—random, fluctuating, and trend—according to their magnitude, while preserving the multi-scale physical characteristics of the sequence, the number of required prediction models is significantly reduced, effectively lowering the overall computational complexity. Regarding the distinctly different fluctuation characteristics of the reconstructed components, this invention employs... A differentiated modeling strategy is employed, matching the trend component, fluctuation component, and random component with LSTM, BiLSTM, and BiLSTM-A models incorporating an attention mechanism, respectively. Based on the Pearson correlation coefficient, a subset of strongly correlated specific meteorological features is selected as input for each component, achieving precise collaborative optimization of "sequence characteristics - input features - model structure." Ultimately, this method significantly improves the accuracy and robustness of ultra-short-term wind power forecasting through a complete technical chain of "frequency domain-guided parameter optimization - entropy-driven intelligent reconstruction - feature-matched collaborative prediction," while ensuring the model's engineering practicality and providing more reliable data support for grid optimization scheduling, reserve arrangements, and safe and stable operation.

[0064] Please see Figure 2 The diagram shows a block diagram of a wind power prediction system based on frequency domain adaptive hyperparameter optimization according to this application.

[0065] like Figure 2 As shown, the wind power prediction system 200 includes an acquisition module 210, an optimization module 220, a decomposition module 230, a reconstruction module 240, a screening module 250, a prediction module 260, and an overlay module 270.

[0066] The acquisition module 210 is configured to acquire historical wind power data of the target area and historical meteorological data associated with the historical wind power data, and divide them into training set and test set according to a preset ratio; the optimization module 220 is configured to define a frequency domain coverage overlap coefficient based on the frequency domain, and use the Grey Wolf optimization algorithm to adaptively optimize the total number of modes and penalty factor of variational mode decomposition with the frequency domain coverage overlap coefficient as the optimization objective to obtain the optimal hyperparameter combination; the decomposition module 230 is configured to use the variational mode decomposition model configured by the optimal hyperparameter combination to decompose the wind power sequences in the training set and test set respectively to obtain several intrinsic mode function components; the reconstruction module 240 is configured to calculate the sample entropy value of the several intrinsic mode function components, and according to the size range of the sample entropy value... The system reconstructs all intrinsic mode function components into random components, fluctuation components, and trend components. A filtering module 250 is configured to, based on the Pearson correlation coefficient, filter out a first feature subset corresponding to the random component, a second feature subset corresponding to the fluctuation component, and a third feature subset corresponding to the trend component from the historical meteorological data. A prediction module 260 is configured to input the first, second, and third feature subsets into a preset prediction model, and the prediction model outputs corresponding predicted values ​​for the random component, fluctuation component, and trend component. A superposition module 270 is configured to superimpose the predicted values ​​for the random component, fluctuation component, and trend component to obtain the final wind power prediction result.

[0067] It should be understood that Figure 2 The modules and references described in the document Figure 1 The steps described in the text correspond to those in the method described above. Therefore, the operations, features, and corresponding technical effects described above also apply to the method described in the text. Figure 2 The various modules in the document will not be described in detail here.

[0068] In other embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the wind power prediction method based on frequency domain adaptive hyperparameter optimization in any of the above method embodiments.

[0069] In one embodiment, the computer-readable storage medium of the present invention stores computer-executable instructions, which are configured as follows:

[0070] Historical wind power data of the target area and historical meteorological data associated with the historical wind power data are obtained and divided into training set and test set according to a preset ratio;

[0071] Define a frequency domain coverage overlap coefficient based on the frequency domain, and use the Grey Wolf optimization algorithm to adaptively optimize the total number of modes and penalty factor of variational mode decomposition with the frequency domain coverage overlap coefficient as the optimization objective, so as to obtain the optimal hyperparameter combination.

[0072] Using the variational mode decomposition model configured with the optimal hyperparameter combination, the wind power sequences in the training set and the test set are decomposed to obtain several intrinsic mode function components.

[0073] Calculate the sample entropy values ​​of the aforementioned intrinsic mode function components, and reconstruct all intrinsic mode function components into random components, fluctuation components, and trend components according to the range of sample entropy values.

[0074] Based on the Pearson correlation coefficient, a first feature subset corresponding to the random component, a second feature subset corresponding to the fluctuation component, and a third feature subset corresponding to the trend component are selected from the historical meteorological data, respectively.

[0075] The first feature subset, the second feature subset, and the third feature subset are input into a preset prediction model, and the prediction model outputs the corresponding random component prediction value, fluctuation component prediction value, and trend component prediction value.

[0076] The predicted values ​​of the random component, the fluctuation component, and the trend component are superimposed to obtain the final wind power prediction result.

[0077] Computer-readable storage media may include a stored program area and a stored data area, wherein the stored program area may store an operating system and an application program required for at least one function; the stored data area may store data created based on the use of the frequency-domain adaptive hyperparameter optimization-based wind power prediction system, etc. Furthermore, the computer-readable storage medium may include high-speed random access memory, and may also include memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include memory remotely configured relative to a processor, which can be connected to the frequency-domain adaptive hyperparameter optimization-based wind power prediction system via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0078] Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present invention, such as... Figure 3As shown, the device includes a processor 310 and a memory 320. The electronic device may also include an input device 330 and an output device 340. The processor 310, memory 320, input device 330, and output device 340 can be connected via a bus or other means. Figure 3 Taking a bus connection as an example, the memory 320 is the computer-readable storage medium described above. The processor 310 executes various server functions and data processing by running non-volatile software programs, instructions, and modules stored in the memory 320, thereby implementing the wind power prediction method based on frequency domain adaptive hyperparameter optimization described in the above embodiment. The input device 330 can receive input digital or character information and generate key signal inputs related to user settings and function control of the wind power prediction system based on frequency domain adaptive hyperparameter optimization. The output device 340 may include a display screen or other display device.

[0079] The aforementioned electronic device can execute the method provided in the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in the embodiments of the present invention.

[0080] In one implementation, the above-described electronic device is applied to a wind power prediction system based on frequency domain adaptive hyperparameter optimization, for a client, and includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0081] Historical wind power data of the target area and historical meteorological data associated with the historical wind power data are obtained and divided into training set and test set according to a preset ratio;

[0082] Define a frequency domain coverage overlap coefficient based on the frequency domain, and use the Grey Wolf optimization algorithm to adaptively optimize the total number of modes and penalty factor of variational mode decomposition with the frequency domain coverage overlap coefficient as the optimization objective, so as to obtain the optimal hyperparameter combination.

[0083] Using the variational mode decomposition model configured with the optimal hyperparameter combination, the wind power sequences in the training set and the test set are decomposed to obtain several intrinsic mode function components.

[0084] Calculate the sample entropy values ​​of the aforementioned intrinsic mode function components, and reconstruct all intrinsic mode function components into random components, fluctuation components, and trend components according to the range of sample entropy values.

[0085] Based on the Pearson correlation coefficient, a first feature subset corresponding to the random component, a second feature subset corresponding to the fluctuation component, and a third feature subset corresponding to the trend component are selected from the historical meteorological data, respectively.

[0086] The first feature subset, the second feature subset, and the third feature subset are input into a preset prediction model, and the prediction model outputs the corresponding random component prediction value, fluctuation component prediction value, and trend component prediction value.

[0087] The predicted values ​​of the random component, the fluctuation component, and the trend component are superimposed to obtain the final wind power prediction result.

[0088] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0089] 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 foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A wind power prediction method based on frequency domain adaptive hyperparameter optimization, characterized in that, include: Historical wind power data of the target area and historical meteorological data associated with the historical wind power data are obtained and divided into training set and test set according to a preset ratio; A frequency domain coverage overlap coefficient is defined, and the Grey Wolf optimization algorithm is used to adaptively optimize the total number of modes and the penalty factor in variational mode decomposition, with the frequency domain coverage overlap coefficient as the optimization objective, to obtain the optimal hyperparameter combination. The formula for calculating the frequency domain coverage overlap coefficient is as follows: ， In the formula, This represents the maximum value in the frequency range after the Fast Fourier Transform. This represents the total number of modes in the VMD decomposition. Let be the amplitude of the k-th order IMF component at frequency ω. Let be the amplitude of the (k+1)th order IMF component at frequency ω. This represents the amplitude of wind power at frequency ω; Using the variational mode decomposition model configured with the optimal hyperparameter combination, the wind power sequences in the training set and the test set are decomposed to obtain several intrinsic mode function components. Calculate the sample entropy values ​​of the aforementioned intrinsic mode function components, and reconstruct all intrinsic mode function components into random components, fluctuation components, and trend components according to the range of sample entropy values. Based on the Pearson correlation coefficient, a first feature subset corresponding to the random component, a second feature subset corresponding to the fluctuation component, and a third feature subset corresponding to the trend component are selected from the historical meteorological data, respectively. The first feature subset, the second feature subset, and the third feature subset are input into a preset prediction model, and the prediction model outputs the corresponding random component prediction value, fluctuation component prediction value, and trend component prediction value. The predicted values ​​of the random component, the fluctuation component, and the trend component are superimposed to obtain the final wind power prediction result.

2. The wind power prediction method based on frequency domain adaptive hyperparameter optimization according to claim 1, characterized in that, The gray wolf optimization algorithm, using the frequency domain coverage overlap coefficient as the optimization objective, adaptively optimizes the total number of modes and the penalty factor of the variational mode decomposition to obtain the optimal hyperparameter combination, including: Initialize the gray wolf population and set the optimization interval between the total number of modalities and the penalty factor; In each iteration, the frequency domain coverage overlap coefficient under the hyperparameter combination corresponding to the current gray wolf position is calculated; The gray wolf population is updated according to the position update mechanism of the gray wolf optimization algorithm, and the historical best solution is recorded; When the preset maximum number of iterations is reached, the hyperparameter combination that minimizes the frequency domain coverage overlap coefficient is output as the optimal hyperparameter combination.

3. The wind power prediction method based on frequency domain adaptive hyperparameter optimization according to claim 1, characterized in that, The variational mode decomposition model configured using the optimal hyperparameter combination decomposes the wind power sequences in the training and test sets respectively, obtaining several intrinsic mode function components, including: A constrained variational problem is constructed, the objective of which is to minimize the sum of the estimated bandwidths of all eigenmode function components. The constraint condition of the constrained variational problem is that the sum of all components equals the original power sequence, expressed as: ， In the formula, For the decomposed IMF set, To decompose the center frequency set of the IMF, This represents the total number of modes in the VMD decomposition. For time Partial derivative operations, Pi The imaginary unit, For the first IMF, For the first The center frequency of the first-order IMF, This is the original wind power sequence; By introducing a quadratic penalty factor and Lagrange multipliers, the constrained variational problem is transformed into an augmented Lagrangian function, expressed as: ， In the formula, For Lagrange multipliers, The penalty coefficient is... Let be the Lagrange multiplier function in the time domain. It is the Dirac function; The augmented Lagrange function is solved using the alternating direction multiplier method. The center frequency and amplitude of each eigenmode function component, as well as the Lagrange multipliers, are alternately updated in the frequency domain until the convergence condition is met, yielding the decomposed eigenmode function components, expressed as: ， In the formula, For the first During the nth iteration Fourier transform of the first-order IMF The Fourier transform of the original wind power sequence For the first During the nth iteration Fourier transform of the first-order IMF For the first During the nth iteration Fourier transform of the first-order IMF For the first Fourier transform of the Lagrange multipliers in the next iteration The penalty coefficient is... For the first During the nth iteration The center frequency of the first-order IMF, For the first Fourier transform of Lagrange multipliers during iteration For the first Fourier transform of the Lagrange multipliers in the next iteration Noise tolerance.

4. The wind power prediction method based on frequency domain adaptive hyperparameter optimization according to claim 1, characterized in that, The process of reconstructing all intrinsic mode function components into random components, fluctuation components, and trend components based on the range of sample entropy values ​​includes: A first entropy threshold and a second entropy threshold are set, wherein the first entropy threshold is less than the second entropy threshold; The intrinsic mode function components whose sample entropy value is greater than the second entropy threshold are classified as random components. The intrinsic mode function components whose sample entropy values ​​are between the first entropy threshold and the second entropy threshold are classified as fluctuation components. The intrinsic mode function components whose sample entropy value is less than the first entropy threshold are classified as trend components.

5. The wind power prediction method based on frequency domain adaptive hyperparameter optimization according to claim 1, characterized in that, The preset prediction models include: a long short-term memory network model for predicting the trend component, a bidirectional long short-term memory network model for predicting the fluctuation component, and a bidirectional long short-term memory network model with an attention mechanism for predicting the random component.

6. The wind power prediction method based on frequency domain adaptive hyperparameter optimization according to claim 1, characterized in that, The step of selecting a first feature subset corresponding to the random component, a second feature subset corresponding to the fluctuation component, and a third feature subset corresponding to the trend component from the historical meteorological data based on the Pearson correlation coefficient includes: Calculate the Pearson correlation coefficient between each meteorological feature in the historical meteorological data and the current component, wherein the current component is the random component, the fluctuation component, or the trend component; All meteorological features whose absolute Pearson correlation coefficient is greater than a preset threshold are selected to form a feature subset corresponding to the current component.

7. A wind power prediction system based on frequency domain adaptive hyperparameter optimization, characterized in that, include: The acquisition module is configured to acquire historical wind power data of the target area and historical meteorological data associated with the historical wind power data, and divide them into training set and test set according to a preset ratio; The optimization module is configured to define a frequency domain coverage overlap coefficient based on the frequency domain, and use the Grey Wolf optimization algorithm to adaptively optimize the total number of modes and the penalty factor of the variational mode decomposition with the frequency domain coverage overlap coefficient as the optimization objective, to obtain the optimal hyperparameter combination. The formula for calculating the frequency domain coverage overlap coefficient is: ， In the formula, This represents the maximum value in the frequency range after the Fast Fourier Transform. This represents the total number of modes in the VMD decomposition. Let be the amplitude of the k-th order IMF component at frequency ω. Let be the amplitude of the (k+1)th order IMF component at frequency ω. This represents the amplitude of wind power at frequency ω; The decomposition module is configured to use the variational mode decomposition model configured by the optimal hyperparameter combination to decompose the wind power sequences in the training set and the test set respectively, and obtain several intrinsic mode function components. The reconstruction module is configured to calculate the sample entropy values ​​of the plurality of intrinsic mode function components, and reconstruct all intrinsic mode function components into random components, fluctuation components and trend components according to the range of sample entropy values. The filtering module is configured to filter out a first feature subset corresponding to the random component, a second feature subset corresponding to the fluctuation component, and a third feature subset corresponding to the trend component from the historical meteorological data based on the Pearson correlation coefficient. The prediction module is configured to input the first feature subset, the second feature subset, and the third feature subset into a preset prediction model, and the prediction model outputs the corresponding random component prediction value, fluctuation component prediction value, and trend component prediction value. The overlay module is configured to overlay the predicted values ​​of the random component, the predicted values ​​of the fluctuation component, and the predicted values ​​of the trend component to obtain the final wind power prediction result.

8. An electronic device, characterized in that, include: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the method described in any one of claims 1 to 6.

Images