A method and system for predicting energy storage frequency modulation instructions based on VMD decomposition and exchange method

Abstract

The present invention discloses a method and system for predicting energy storage frequency modulation instructions based on the VMD decomposition and exchange method, which relates to the field of time series prediction technology. The method comprises the following steps: decomposing the frequency modulation instruction Pt through VMD to obtain a subsequence IMF; calculating the resistance value for each data point of each subsequence IMF to obtain a resistance sequence; and exchanging data points of different subsequence IMFs, predicting the results after the exchange, and obtaining the final frequency modulation instruction prediction value. The present invention adjusts data points across subsequences through the exchange method, optimizes the local feature distribution of the subsequences while satisfying the combined resistance conditions, reduces the interference of high resistance points on the prediction model, and improves the model's adaptability to complex patterns.

Description

Technical Field

[0001] The present invention relates to the technical field of time series prediction, and in particular to a method and system for predicting energy storage frequency modulation instructions based on a VMD decomposition and exchange method. Background Art

[0002] The traditional hybrid energy storage (supercapacitor + lithium battery) method for assisting thermal power generation units in frequency regulation involves transmitting the difference between the frequency regulation command and the thermal power generation unit to the hybrid energy storage, with the battery handling the low-frequency portion and the supercapacitor handling the high-frequency portion. However, the signal transmission (from the frequency regulation command to the supercapacitor / lithium battery) takes time, as does the response time of the supercapacitor or lithium battery itself. This results in a certain response time difference, further affecting profitability and, consequently, the power plant's revenue. To address this, the present invention proposes a prediction method that anticipates the magnitude of the frequency regulation command, enabling the supercapacitor / battery to take action in advance, thereby improving profitability. Traditional prediction methods make the resulting subsequences more difficult to predict, with some subsequences exhibiting even greater nonlinearity than the original sequence, making prediction even more challenging. Summary of the Invention

[0003] In view of the problems existing in the existing energy storage frequency modulation instruction prediction and system based on VMD decomposition and exchange method, the present invention is proposed.

[0004] Therefore, the problem to be solved by the present invention is that in traditional prediction methods, the decomposed subsequences are more difficult to predict, and the nonlinearity of some subsequences is even higher than that of the original sequence, which makes prediction more difficult.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] In a first aspect, an embodiment of the present invention provides a method for predicting energy storage frequency modulation instructions based on VMD decomposition and exchange method, which includes the following steps:

[0007] Decompose the frequency modulation instruction Pt through VMD to obtain the subsequence IMF;

[0008] Calculate the resistance value for each data point of each subsequence IMF to obtain the resistance sequence;

[0009] The data points of different subsequence IMFs are exchanged, and the results after the exchange are predicted to obtain the final frequency modulation instruction prediction value.

[0010] As a preferred solution of the energy storage frequency modulation instruction prediction method based on the VMD decomposition and exchange method of the present invention, the calculation of the resistance value includes the following steps:

[0011] Project the subsequence IMF to the YX coordinate system, where the Y axis is the value and the X axis is the data point subscript;

[0012] Generate reference curve , where Z is the subsequence median, J is the mean, [•] indicates rounding, and N indicates the subsequence length;

[0013] For each data point X ii , draw a perpendicular line downward and intersect the reference curve at point O, calculate the intersection of the tangent line at point O and the X ii The area of ​​the triangle formed by the vertical line is used as the resistance value.

[0014] As a preferred solution of the energy storage frequency modulation instruction prediction method based on VMD decomposition and exchange method described in the present invention, when data points are exchanged, the following conditions are met at the same time:

[0015] The exchanged data points come from different subsequences;

[0016] The combined resistance value satisfies: and ,in, is the resistance value of the mth data point in the ith subsequence, is the resistance value of the m-1th data point in the ith subsequence, is the resistance value of the nth data point in the jth subsequence, is the resistance value of the n-1th data point in the jth subsequence.

[0017] As a preferred solution of the energy storage frequency modulation instruction prediction method based on the VMD decomposition and exchange method of the present invention, when predicting the result after the exchange, the specific steps are as follows:

[0018] The swapped subsequences are input into the GRU network for independent prediction, and the prediction results of each subsequence are added together to obtain the final FM instruction prediction value.

[0019] As a preferred solution of the energy storage frequency modulation instruction prediction method based on the VMD decomposition and exchange method of the present invention, when the data points are exchanged, if there are multiple data points that meet the conditions in the same subsequence, the following judgment is performed:

[0020] Get the absolute value of all data points that meet the conditions;

[0021] Determine the maximum absolute value of all data points and place the maximum value in the first tier.

[0022] As a preferred solution of the energy storage frequency modulation instruction prediction method based on the VMD decomposition and exchange method described in the present invention, when the reference curve is generated, when the subsequence median Z or the average value J is zero, it is replaced by the subsequence standard deviation.

[0023] As a preferred solution of the energy storage frequency modulation instruction prediction method based on the VMD decomposition and exchange method described in the present invention, when decomposing and obtaining the subsequence IMF through VMD, the subsequence IMF is normalized.

[0024] In a second aspect, an embodiment of the present invention provides an energy storage frequency modulation instruction prediction system based on the VMD decomposition and exchange method, which includes a VMD decomposition module, a resistance calculation module, a data exchange module, and a prediction module;

[0025] The VMD decomposition module is responsible for decomposing the frequency modulation instruction Pt into multiple subsequences IMF;

[0026] The resistance calculation module performs calculations on the data points in each subsequence IMF;

[0027] The data exchange module determines whether the data points between different subsequence IMFs meet the exchange conditions based on the calculation results of the resistance value;

[0028] The prediction module inputs the subsequence after the exchange process into the prediction model to obtain the final frequency modulation instruction prediction value.

[0029] In a third aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the processor executes the computer program, any step of the above-mentioned energy storage frequency modulation instruction prediction method based on the VMD decomposition and exchange method is implemented.

[0030] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the above-mentioned energy storage frequency modulation instruction prediction method based on the VMD decomposition and exchange method is implemented.

[0031] The beneficial effects of the present invention are:

[0032] Constrained variational optimization is used to ensure the mathematical rigor of the decomposition and effectively avoid modal aliasing. The resistance method is combined to dynamically evaluate the prediction difficulty (resistance value) of each subsequence data point, providing a quantitative basis for subsequent optimization.

[0033] By adjusting data points across subsequences through the exchange method, the local feature distribution of the subsequences is optimized while satisfying the combined resistance conditions, reducing the interference of high resistance points on the prediction model and improving the model's adaptability to complex patterns.

[0034] The optimized subsequences are input into the GRU network for independent prediction, making full use of its gating mechanism to capture long-term dependencies, and the prediction advantages of each subsequence are fused by superimposing the results, ultimately achieving error dispersion and accuracy improvement.

[0035] Fault-tolerant mechanism and normalization preprocessing are introduced to ensure the stability of resistance value calculation and enhance the applicability of the method in different noise environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 This is a flow chart of the energy storage frequency modulation instruction prediction method based on VMD decomposition and exchange method.

[0038] Figure 2 Schematic diagram of the YX coordinate axis of the energy storage frequency modulation instruction prediction method based on VMD decomposition and exchange method.

[0039] Figure 3 Schematic diagram of the reference curve of the energy storage frequency modulation instruction prediction method based on VMD decomposition and exchange method.

[0040] Figure 4 This is the VMD decomposition flow chart of the energy storage frequency modulation instruction prediction method based on VMD decomposition and exchange method. DETAILED DESCRIPTION

[0041] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0042] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0043] What you need to know is: IMF is the intrinsic mode function, VMD is the variational mode decomposition, IMF (Intrinsic Mode Function) and VMD (Variational Mode Decomposition) are two important concepts in the field of signal processing.

[0044] Example 1, reference Figures 1 to 4 , which is the first embodiment of the present invention, provides a method for predicting energy storage frequency modulation instructions based on VMD decomposition and exchange method, comprising the following steps:

[0045] S1. Decompose the frequency modulation instruction Pt through VMD to obtain the subsequence IMF.

[0046] Reference Figure 4 As shown, it should be known that when decomposing and obtaining the subsequence IMF through VMD, the subsequence IMF is normalized.

[0047] In this embodiment, the input signal is: frequency modulation instruction Pt (time series data with a length of N).

[0048] In addition, the decomposition parameter settings include the mode number K: selected according to the signal characteristics (for example, set to K=5 through spectrum analysis or experience); the penalty factor α: the default value is 2000, which is used to balance the bandwidth constraint and data fidelity; the convergence tolerance: set to 1e-6 to ensure the decomposition accuracy; the maximum number of iterations: 1000 times to avoid infinite loops.

[0049] Call the VMD algorithm library (such as Python's vmdpy or MATLAB toolkit), input Pt and the above parameters, and obtain the decomposed K subsequences IMF1, IMF2, ..., IMF K ; During output, the set output requirement is that the length of each IMF is consistent with the original signal (N points) and satisfies the orthogonality and frequency band separation characteristics.

[0050] Assume that Pt becomes IMF1, IMF2, IMF3, IMF after VMD decomposition. i , ..., IMF K , their subsequences are:

[0051] .

[0052] like Figure 2 As shown, project this subsequence onto the YX coordinate axis, where the Y axis represents the size of the value and the horizontal axis is the subscript.

[0053] Each IMF subsequence is independently normalized to ensure the numerical stability of the subsequence, providing reliable input for subsequent resistance value calculation and exchange optimization, while retaining the original signal characteristics and avoiding information loss.

[0054] S2. Calculate the resistance value for each data point of each subsequence IMF to obtain a resistance sequence.

[0055] The corresponding resistance sequence is:

[0056] ;

[0057] K represents the number of subsequences, i represents the number of the i-th subsequence, N is the subsequence length, and S represents the resistance value.

[0058] Reference Figure 3 As shown,

[0059] Project the subsequence IMF to the YX coordinate system, where the Y axis is the value and the X axis is the data point subscript;

[0060] Generate reference curve , where Z is the subsequence median, J is the mean, [•] indicates rounding, and N indicates the subsequence length;

[0061] For each data point X ii , draw a perpendicular line downward and intersect the reference curve at point O, calculate the intersection of the tangent line at point O and the X ii The area of ​​the triangle formed by the vertical line is used as the resistance value, and the positive and negative signs are determined by X ii The point is determined by whether it is above or below the reference curve. If it is above the reference curve, it is positive, and if it is below the reference curve, it is negative. When the reference curve is generated, if the subsequence median Z or mean J is zero, it is replaced by the subsequence standard deviation.

[0062] Assume that the IMF i The parameters of a subsequence are: Z=0.5, J=0.3, N=10, the current calculation point i=2, X ii =1.2;

[0063] Reference curve Y value: ;

[0064] Tangent slope: ;

[0065] Length of vertical line: ;

[0066] Resistance value: ;

[0067] because , take a positive value.

[0068] When the data points are exchanged, the following conditions are met at the same time:

[0069] The exchanged data points come from different subsequences;

[0070] The combined resistance value satisfies: and ,in, is the resistance value of the mth data point in the ith subsequence, is the resistance value of the m-1th data point in the ith subsequence, is the resistance value of the nth data point in the jth subsequence, is the resistance value of the n-1th data point in the jth subsequence.

[0071] S3. Exchange the data points of different subsequence IMFs, predict the results after the exchange, and obtain the final frequency modulation instruction prediction value.

[0072] When predicting the results after the exchange, the specific steps are as follows:

[0073] The swapped subsequences are input into the GRU network for independent prediction, and the prediction results of each subsequence are added together to obtain the final FM instruction prediction value.

[0074] When exchanging data points, if there are multiple data points that meet the conditions in the same subsequence, the following judgment is made:

[0075] Get the absolute value of all data points that meet the conditions;

[0076] Determine the maximum absolute value of all data points and place the maximum value in the first tier.

[0077] In this embodiment, the GRU network structure includes an input layer: a univariate time series (each subsequence is input independently); a hidden layer: two layers of GRU units, each with 64 neurons; an output layer: a fully connected layer, which outputs the predicted value for the next T steps;

[0078] Training parameters: Loss function: mean squared error (MSE); Optimizer: Adam (learning rate = 0.001); Batch size: 32; Early stopping mechanism: If the validation loss does not decrease for 5 consecutive rounds, terminate the training.

[0079] Each subsequence is input into an independent GRU model to obtain the prediction result. The prediction results of all subsequences are summed to obtain the final FM instruction prediction value:

[0080] ;

[0081] in, To predict the results, is the predicted value of the final frequency modulation instruction.

[0082] If a subsequence does not have a satisfying exchange point, the original subsequence is retained and predicted directly.

[0083] Swap conflict: If multiple subsequences compete for the same external point, the subsequence with the greater absolute resistance value will be prioritized. When verifying the prediction results, calculate the prediction error (such as MAE, RMSE); compare the errors before and after the swap to verify the optimization effect.

[0084] This embodiment also provides an example calculation: original subsequence: IMF1 = [0.2, -0.5, 1.1, -0.3], IMF2 = [0.8, -1.2, 0.4, 0.6];

[0085] Resistance sequence: S1= [0.3, -0.7, 1.3, -0.4], S2= [0.9, -1.5, 0.5, 0.7];

[0086] Candidate swap pairs: (IMF1[2], IMF2[1]): Comprehensive resistance value: 1.3 + (-1.5) - 1 = -1.2 (does not meet the conditions, discard);

[0087] (IMF1[3], IMF2[4]): ; Comprehensive resistance value: -0.4+0.7-1=-0.7 (meets the conditions).

[0088] Swap IMF1[3] and IMF2[4] to obtain: IMF1' = [0.2, -0.5, 1.1, 0.6], IMF2' = [0.8,-1.2, 0.4, -0.3]; input IMF1' and IMF2', and add the prediction results.

[0089] This embodiment prioritizes the absolute value of resistance to ensure the priority exchange of high resistance points, effectively reducing the local interference of the prediction model; combined with GRU independent prediction and result fusion, it significantly improves the accuracy and stability of frequency modulation instruction prediction.

[0090] In summary, constrained variational optimization ensures the mathematical rigor of the decomposition and effectively avoids modal aliasing. The resistance method is combined to dynamically evaluate the prediction difficulty (resistance value) of each subsequence data point, providing a quantitative basis for subsequent optimization. The exchange method adjusts data points across subsequences. While satisfying the combined resistance condition, the local feature distribution of the subsequences is optimized, reducing the interference of high-resistance points on the prediction model and improving the model's adaptability to complex patterns. The optimized subsequences are independently predicted by the GRU network, leveraging its gating mechanism to capture long-term dependencies. The predicted advantages of each subsequence are then combined and integrated, ultimately achieving error dispersion and improved accuracy. Fault-tolerance mechanisms and normalization preprocessing are introduced to ensure the stability of resistance value calculation and enhance the method's applicability in various noise environments.

[0091] Example 2: Based on the first example, this example further provides a system for predicting energy storage frequency modulation instructions based on the VMD decomposition and exchange method, including a VMD decomposition module, a resistance calculation module, a data exchange module, and a prediction module.

[0092] The VMD decomposition module is responsible for decomposing the frequency modulation instruction Pt into multiple subsequences IMF;

[0093] The resistance calculation module performs calculations on the data points in each subsequence IMF;

[0094] The data exchange module determines whether the data points between different subsequence IMFs meet the exchange conditions based on the calculation results of the resistance value;

[0095] The prediction module inputs the subsequence after the exchange process into the prediction model to obtain the final frequency modulation instruction prediction value.

[0096] This embodiment also provides a computer device, which is applicable to the case of the energy storage frequency modulation instruction prediction method based on the VMD decomposition and exchange method, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the energy storage frequency modulation instruction prediction method based on the VMD decomposition and exchange method proposed in the above embodiment.

[0097] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0098] This embodiment further provides a storage medium storing a computer program, which, when executed by a processor, implements the energy storage frequency modulation instruction prediction method based on the VMD decomposition and exchange method proposed in the above embodiment.

[0099] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiment belong to the same inventive concept. Technical details not fully described in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0100] Example 3: Based on the previous two examples, this example provides a method for predicting energy storage frequency modulation instructions based on VMD decomposition and exchange method.

[0101] In order to further verify the advantages of the present invention, the present invention uses the method of the present invention and the GRU prediction method to predict the FM sequence, and the results are shown in Table 1.

[0102] Table 1: Schematic diagram of FM sequence prediction results

[0103] ,

[0104] There are four other evaluation indicators, and the results are shown in Table 2.

[0105] Table 2: Evaluation index table

[0106] ,

[0107] In this embodiment, N represents the sample size, and They represent the actual value and predicted value at time n respectively. From the experimental results, we can see that the four evaluation indicators are reduced, which shows that the proposed model can well improve the prediction accuracy.

[0108] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for predicting energy storage frequency modulation instructions based on VMD decomposition and exchange method, characterized by: The following steps are included: Decompose the frequency modulation instruction Pt through VMD to obtain the subsequence IMF; Calculate the resistance value for each data point of each subsequence IMF to obtain the resistance sequence; Exchange the data points of different subsequence IMFs, predict the results after the exchange, and obtain the final frequency modulation instruction prediction value; The calculation of the resistance value includes the following steps: Project the subsequence IMF to the YX coordinate system, where the Y axis is the value and the X axis is the data point subscript; Generate reference curve Y=e [ZJ] + sin(NX), where Z is the subsequence median, J is the mean, [·] indicates rounding, and N is the subsequence length; For each data point X ii , draw a perpendicular line downward and intersect the reference curve at point O, calculate the intersection of the tangent line at point O and the X ii The area of ​​the triangle formed by the vertical line is used as the resistance value, where the positive and negative signs are determined by X ii Determined by being above or below the reference curve; When the data points are exchanged, the following conditions are met at the same time: The exchanged data points come from different subsequences; Resistance value meets: S im +S jn-1 ∈[-1,1] and S jn +S im-1 ∈[-1,1], where S im is the resistance value of the mth data point in the ith subsequence, S im-1 is the resistance value of the m-1th data point in the ith subsequence, S jn is the resistance value of the nth data point in the jth subsequence, S jn-1 is the resistance value of the n-1th data point in the jth subsequence.

2. The method for predicting energy storage frequency modulation instructions based on VMD decomposition and exchange method according to claim 1, wherein: When predicting the results after the exchange, the specific steps are as follows: The swapped subsequences are input into the GRU network for independent prediction, and the prediction results of each subsequence are added together to obtain the final FM instruction prediction value.

3. The energy storage frequency modulation instruction prediction method based on the VMD decomposition and exchange method according to claim 2, characterized in that: When exchanging data points, if there are multiple data points that meet the conditions in the same subsequence, the following judgment is made: Get the absolute value of all data points that meet the conditions; Determine the maximum absolute value of all data points and place the maximum value in the first tier.

4. The method for predicting energy storage frequency modulation instructions based on VMD decomposition and exchange method according to claim 3, characterized in that: When the reference curve is generated, if the subsequence median Z or mean J is zero, it is replaced by the subsequence standard deviation.

5. The energy storage frequency modulation instruction prediction method based on VMD decomposition and exchange method according to claim 4, characterized in that: When decomposing and obtaining the subsequence IMF through VMD, the subsequence IMF is normalized.

6. A system for predicting energy storage frequency modulation instructions based on VMD decomposition and exchange method, based on the energy storage frequency modulation instruction prediction method based on VMD decomposition and exchange method according to any one of claims 1 to 5, characterized in that: Includes VMD decomposition module, resistance calculation module, data exchange module, and prediction module; The VMD decomposition module is responsible for decomposing the frequency modulation instruction Pt into multiple subsequences IMF; The resistance calculation module performs calculations on the data points in each subsequence IMF; The data exchange module determines whether the data points between different subsequence IMFs meet the exchange conditions based on the calculation results of the resistance value; The prediction module inputs the subsequence after the exchange process into the prediction model to obtain the final frequency modulation instruction prediction value.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the energy storage frequency modulation instruction prediction method based on the VMD decomposition and exchange method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the energy storage frequency modulation instruction prediction method based on the VMD decomposition and exchange method described in any one of claims 1 to 5 are implemented.

Images