Energy storage frequency modulation instruction prediction method and system based on signal sequence regularization

Abstract

The invention discloses an energy storage frequency modulation instruction prediction method based on signal sequence regularization. An original frequency modulation instruction signal sequence is obtained; regularization characteristic parameters representing the regularity degree of the original frequency modulation instruction signal sequence are obtained according to the original frequency modulation instruction signal sequence, wherein the regularization characteristic parameters comprise a regularization index P and an average regularization degree value D; adjusting the original frequency modulation instruction signal sequence based on the regularized characteristic parameters to generate a reference frequency modulation instruction signal sequence; based on the original frequency modulation instruction signal sequence and the reference frequency modulation instruction signal sequence, a neural network is adopted for prediction to obtain an intermediate prediction result; performing mathematical operation on the intermediate prediction result to obtain a final prediction result; the prediction result of the frequency modulation instruction is more accurate, the size of the frequency modulation instruction is predicted in advance for energy storage regulation, and the response accuracy and the frequency modulation income are improved.

Description

Technical Field

[0001] This application relates to the technical field of power grid frequency modulation, and particularly to a method and system for predicting energy storage frequency modulation commands based on signal sequence regularization. Background Art

[0002] Currently, in major regional power grids in China, large-scale hydropower and thermal power units (coal-fired units / gas-fired units) are mainly used as power grid frequency modulation power sources. The output of the frequency modulation power source is adjusted to respond to the change of the system frequency. The prediction of hybrid energy storage frequency modulation commands refers to establishing a prediction model through the analysis of historical frequency modulation commands and related data to predict the future frequency modulation command requirements, so as to optimize the control strategy of the hybrid energy storage system and improve its frequency modulation performance and efficiency. It can improve system performance and respond quickly, and can quickly adjust the output power in a short time to respond rapidly to the change of the power grid frequency, effectively balance the supply-demand difference in the power system, and maintain the stability of the power grid frequency. It can improve the regulation accuracy, track the frequency modulation commands more accurately, reduce the regulation error, and improve the power supply quality and stability of the power grid. It can enhance system stability. By reasonably distributing the power and energy of different energy storage devices, the overall risk of the system can be reduced, the reliability and stability of the system can be improved, and the risk of system failure caused by the failure or performance degradation of a single energy storage device can be reduced. It can extend the service life of the device. By optimizing the charge-discharge strategy, according to the prediction results, a reasonable charge-discharge strategy can be formulated to avoid overcharging, over-discharging or frequent charge-discharging of the energy storage device, thereby extending its service life. It can balance the use of devices. Reasonably distribute the use frequency and load of different energy storage devices to make the aging degree of each device relatively balanced, reduce the maintenance cost and replacement frequency of the device. It can reduce costs and improve economic benefits. By improving the frequency modulation performance and efficiency, the hybrid energy storage system can obtain more frequency modulation benefits, and at the same time reduce the power outage losses and equipment damage costs caused by the instability of the power grid frequency. It can optimize the investment cost. According to the prediction results, the capacity and quantity of different energy storage devices can be reasonably configured to avoid over-investment and resource waste, and improve the return on investment.

[0003] The traditional prediction method is to directly put the original frequency modulation sequence into the GRU network for prediction. The original frequency modulation sequence has strong nonlinearity, which will increase the difficulty of neural network prediction. The reasons are as follows: 1. Capturing complex patterns; Nonlinear dynamic characteristics: Nonlinear time series may contain complex patterns such as mutations, multi-modal fluctuations, and chaotic behaviors (such as the Lorenz system). Traditional linear models cannot fully model them, and although neural networks can theoretically approximate these patterns, they require higher model complexity.

[0004] 2. Challenges in model training; Overfitting risk: Complex non-linear relationships require deeper networks or more parameters, but are prone to overfitting with small sample data. Optimization difficulty: The non-convexity of the loss function may lead to getting stuck in local optima. For example, when training GRUs, it is necessary to carefully select the learning rate and initialization strategy.

[0005] 3. Data requirements and quality; Data volume requirement: Learning non-linear patterns usually requires a large amount of data. For example, predicting a chaotic system may require tens of thousands of time-step samples to cover different states.

[0006] 4. Uncertainty of dynamic systems; Sensitivity of chaotic behavior: Even if the model fits historical data well, long-term predictions may still fail due to the exponential amplification of initial condition errors.

[0007] Putting the original sequence directly into the prediction will lead to inaccurate prediction results and large prediction errors. Therefore, it is necessary to preprocess the original frequency modulation sequence before prediction. To solve this problem, decomposition and noise reduction are proposed: using VMD decomposition or wavelet transform to separate trends, periods, and residuals, reducing non-linear complexity.

[0008] As Figure 2 shown, common ones such as VMD decomposition, etc., decompose the original frequency modulation sequence into a series of low-frequency subsequences and then put them into the neural network for prediction. However, when using VMD (Variational Mode Decomposition) for the decomposition and prediction of the original frequency modulation sequence, although it can improve the accuracy of non-stationary signal processing, it also has the following disadvantages: complex parameter tuning and high computational cost. Wavelet transform has strong dependence on the time-base function and limited frequency resolution. Summary of the Invention

[0009] Aiming at the deficiencies of the existing technology, the present invention proposes a prediction method that can reduce the non-linearity and non-regularity of the original sequence without using decomposition and noise reduction.

[0010] A prediction method for energy storage frequency modulation commands based on signal sequence regularization, characterized by comprising: S101. Obtain the original frequency modulation command signal sequence; S102. Obtain the regularization characteristic parameters representing the degree of regularity of the original frequency modulation command signal sequence, where the regularization characteristic parameters include a regularization index P and an average regularization degree value D; S103. Adjust and generate a reference frequency modulation command signal sequence based on the regularization characteristic parameters for the original frequency modulation command signal sequence; S104. Respectively use a neural network for prediction based on the original frequency modulation command signal sequence and the reference frequency modulation command signal sequence to obtain intermediate prediction results; S105. Perform mathematical operations on the intermediate prediction result to obtain the final prediction result.

[0011] An energy storage frequency modulation command prediction system based on signal sequence regularization, characterized by comprising: A command acquisition module for acquiring an original frequency modulation command signal sequence; A regularization characteristic parameter calculation module for obtaining regularization characteristic parameters characterizing the degree of regularity of the original frequency modulation command signal sequence, where the regularization characteristic parameters include a regularization index P and an average regularization degree value D; An intermediate sequence generation module for adjusting and generating a reference frequency modulation command signal sequence based on the regularization characteristic parameters; A prediction processing module for respectively performing predictions on the original frequency modulation command signal sequence and the reference frequency modulation command signal sequence using a neural network to obtain an intermediate prediction result; A prediction processing module for performing mathematical operations on the intermediate prediction result to obtain the final prediction result.

[0012] The beneficial effects of the present invention are: Reduce the non-linearity degree of the original signal sequence, make the corrected sequence smoother, greatly reduce the prediction difficulty, improve the accuracy of frequency modulation command prediction, and thus improve the response accuracy and frequency modulation benefit of the power plant. Description of the Drawings

[0013] Figure 1 It is a method step diagram.

[0014] Figure 2 It is a schematic diagram of a traditional algorithm. Specific Embodiments

[0015] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0016] It should be understood that the step numbers used in the text are only for convenient description and do not limit the execution order of the steps.

[0017] It should be understood that the terms used in the specification of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification of the present application and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0018] The terms "comprising" and "including" indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their combinations.

[0019] The term "and / or" refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0020] An embodiment of the present disclosure provides a method for predicting energy storage frequency modulation commands based on signal sequence regularization, as Figure 1 shown, including: S101. Obtain the original frequency modulation command signal sequence; S102. Obtain a regularization characteristic parameter characterizing the degree of regularity of the original frequency modulation command signal sequence. The regularization characteristic parameter includes a regularization index P and an average regularization degree value D; S103. Adjust and generate a reference frequency modulation command signal sequence based on the regularization characteristic parameter for the original frequency modulation command signal sequence; S104. Respectively use a neural network for prediction based on the original frequency modulation command signal sequence and the reference frequency modulation command signal sequence to obtain an intermediate prediction result; S105. Perform a mathematical operation on the intermediate prediction result to obtain a final prediction result.

[0021] The frequency modulation command prediction method based on supercapacitor energy storage provided by the embodiment of the present disclosure uses a new type of supercapacitor and a lithium battery as energy storage devices and performs frequency modulation on the power grid frequency in a power plant when it fluctuates. The state of the power plant power grid is monitored in real time. When the power supply frequency in the power plant power grid fluctuates, a corresponding frequency modulation command signal is generated. After obtaining this original signal, the power plant power grid is frequency modulated by predicting future signals.

[0022] Among them, step S101, obtaining the original frequency modulation command signal sequence, includes: The original frequency modulation command signal sequence is Pt = [X1, X2, X3,..., X i ,..., X N . The original frequency modulation command signal sequence Pt is a function of time t.

[0023] Project each numerical point in the original frequency modulation command signal sequence Pt onto a plane coordinate system. The vertical axis represents the magnitude of the value, which is the signal values X1, X2, X3,..., X i ,..., X N , and the horizontal axis is the sampling times, with one sampling per second.

[0024] In yet another embodiment provided by the disclosure of the present application, for the original frequency modulation command signal sequence, the greater the degree of non-regularity, the greater the difficulty of prediction. Generally speaking, if the original frequency modulation command signal sequence conforms to the regular undulating changes similar to the sine / cosine function, then the regularity of this sequence is relatively strong and it is relatively easy to predict. On the contrary, the prediction difficulty is relatively large. The degree of regularity of the original frequency modulation command signal sequence can be characterized by the regularization index P and the average degree of regularization D.

[0025] In step S102 above, for the original frequency modulation command signal sequence, obtain the regularization characteristic parameters characterizing its degree of regularity, and the regularization characteristic parameters include the regularization index P and the average degree of regularization value D; it includes: Step 1: Determine whether there is a regular sequence in the signal sequence within a unit length.

[0026] A certain number of consecutive signals are selected for the unit length. To reflect the regularity of the consecutive signals, 6 consecutive signals can be selected.

[0027] Select 6 consecutive signals [X q1 , X q2 , X q3 , X q4 , X q5 , X q6 . If X q2 is the minimum value of 6 consecutive signals, and X q5 is the maximum value of 6 consecutive signals, then the consecutive signals [X q1 , X q2 , X q3 , X q4 , X q5 , X q6 form a regular sequence.

[0028] Step 2: Use a traversal algorithm to find all regular sequences from the original frequency modulation command signal sequence to obtain the regularization index P.

[0029] Traverse and judge whether every 6 consecutive signals in [X1, X2, X3, …, X i , …, X N form a regular sequence, and sum up the statistics to obtain the number Q of all regular sequences.

[0030] For example, Pt = [X1, X2, X3, …, X i , …, X 12, first, determine whether [X1, X2, X3, X4, X5, X6] is a regular sequence. Next, determine whether [X2, X3, X4, X5, X6, X7] is a regular sequence. Then, determine [X3, X4, X5, X6, X7, X8], and so on until determining whether [X7, …, X 12 is a regular sequence.

[0031] Calculate the regularization index P = Q / N, where Q is the number of all regular sequences and N is the number of signals in the original frequency modulation command signal sequence.

[0032] Step 3: Calculate the regularization degree value D of each regular sequence i , and obtain the average regularization degree value D of the original frequency modulation command signal sequence.

[0033] Calculate the regularization degree value D for all regular sequences i .

[0034] Suppose one of the regular sequences is [X q1 , X q2 , X q3 , X q4 , X q5 , X q6 , calculate the average value X q1 , X q3 , X q4 , X q6 of the non-extreme points [X avg , calculate the median X q1 , X q3 , X q4 , X q6 of the non-extreme points [X Z , The mathematical meaning of the median is the value in the middle position after arranging a set of data in ascending or descending order. It can reflect the "middle level" of the data, is not sensitive to extreme values, and is one of the important indicators for describing the central tendency of the data. Specific calculation steps: First, sort the data in ascending (or descending) order. Calculate in different cases: For an odd number of data, take the number in the middle after sorting. For an even number of data, take the average of the two middle numbers.

[0035] Example: For odd cases: The data is [3, 1, 5, 2, 4], after sorting it is [1, 2, 3, 4, 5], and the median is the 3rd number 3.

[0036] For even cases: The data is [7, 2, 4, 6], after sorting it is [2, 4, 6, 7], and the median is the average 5 of the two middle numbers (4 and 6).

[0037] The regular sequence [Xq1 ,X q2 ,X q3 ,X q4 ,X q5 ,X q6 Regularization degree value D of i is: , sigmoid() is the activation function, and the double vertical bar operator represents taking the absolute value.

[0038] Calculate the regularization degree value D for all regular number sequences i to obtain the average regularization degree value D. .

[0039] The above S103, adjusting the original frequency modulation command signal sequence based on the regularization characteristic parameters to generate a reference frequency modulation command signal sequence; including: For the original frequency modulation command signal sequence Pt = [X1, X2, X3,..., X i ,..., X N , adjust the non-regular number sequences; the non-regular number sequences include continuous signals within a unit length, and 6 consecutive signals can be selected as a group for adjustment.

[0040] Let a group of continuous signals [X c1 , X c2 , X c3 , X c4 , X c5 , X c6 be the non-regular number sequence in the original frequency modulation command signal sequence Pt, where X c2 must not be the minimum value, and X c5 must not be the maximum value.

[0041] [X c1 , X c2 , X c3 , X c4 , X c5 , X c6 the minimum value is X min , [X c1 , X c2 , X c3 , X c4 , X c5 , X c6 the maximum value is X max .

[0042] First, perform a mathematical transformation on X c2 based on the regularization exponent P and the average regularization degree value D to make it the minimum value. The mathematical transformation method is , where rand() is a random function.

[0043] Similarly, for X c5 a mathematical transformation is performed based on the regularization exponent P and the average regularization degree value D to make it the maximum value. The mathematical transformation method is , where rand() is a random function.

[0044] Secondly, random values are generated for each signal in [X c1 , X c3 , X c4 , X c6 . X ci = rand ( X ci - exp( D ), X ci + exp( D ), i = 1, 3, 4, 6, and rand() is a random function.

[0045] Random values are continuously generated in a loop for a pre-specified number of times, which can be set to 200 times.

[0046] For each randomly generated [X c1 , X c3 , X c4 , X c6 and after combining it with the [X c2 , X c5 after the above mathematical transformation, calculate the regularization degree value of [X c1 , X c2 , X c3 , X c4 , X c5 , X c6 .

[0047] Take the [X c1 , X c3 , X c4 , X c6 corresponding to the calculated regularization degree value closest to the average regularization degree value D as the replacement value, and combine the replacement value [X c1 , X c3 , X c4 , X c6 with the [X c2 , X c5 after the above mathematical transformation to obtain the [X c1 , Xc2 ,X c3 ,X c4 ,X c5 ,X c6 。

[0048] Finally, after performing mathematical transformations on each non-regular sequence in the original frequency modulation command signal sequence, the reference frequency modulation command signal sequence Pt is obtained. 、 。

[0049] Among them, after performing a mathematical transformation on a non-regular sequence, continue to perform mathematical transformations on other subsequent non-regular sequences.

[0050] The above S104, based on the original frequency modulation command signal sequence and the reference frequency modulation command signal sequence, uses a neural network for prediction to obtain an intermediate prediction result; including: Put the finally obtained reference frequency modulation command signal sequence Pt 、 into the GRU network for prediction to obtain the first intermediate prediction result Y1 = [Y1, Y2, Y3, …, Y S 。

[0051] Put the difference sequence Pt - Pt of the original frequency modulation command signal sequence and the reference frequency modulation command signal sequence 、 into the GRU network for prediction to obtain the second intermediate prediction result Y2 = [Y1 、 , Y2 、 , Y3 、 , …, Y S 、 。

[0052] The above S105, perform a mathematical operation on the intermediate prediction result to obtain the final prediction result; including: The first intermediate prediction result and the second intermediate prediction result are the prediction results of the reference frequency modulation command signal sequence and the difference sequence, not the prediction result of the original frequency modulation command signal sequence Pt, and a mathematical operation is required.

[0053] The final prediction result Y3 = Y1 + Y2.

[0054] Corresponding to the method shown above Figure 1 The embodiments of the present disclosure also provide An energy storage frequency modulation command prediction system based on signal sequence regularization, characterized by including, A command acquisition module, which acquires the original frequency modulation command signal sequence; A regularization characteristic parameter calculation module, which acquires the regularization characteristic parameters representing the degree of regularity of the original frequency modulation command signal sequence. The regularization characteristic parameters include a regularization index P and an average regularization degree value D; The intermediate sequence generation module adjusts and generates a reference frequency modulation command signal sequence based on the regularized characteristic parameters from the original frequency modulation command signal sequence; The prediction processing module respectively uses a neural network to perform predictions on the original frequency modulation command signal sequence and the reference frequency modulation command signal sequence to obtain intermediate prediction results; The prediction processing module performs mathematical operations on the intermediate prediction results to obtain the final prediction results.

[0055] To further verify the advantages of the present invention, the present invention respectively uses the method of the present invention and the method of direct prediction by GRU to predict the frequency modulation sequence, and the results are as follows.

[0056] The frequency modulation sequence 1 comes from the frequency modulation data of a certain power plant in Hulunbuir from 0:00 to 20:00 on December 1, 2024.

[0057] The frequency modulation sequence 2 comes from the frequency modulation data of a certain power plant in Hulunbuir from 2:00 to 22:00 on December 2, 2024.

[0058] The performance comparison is as follows: Table 1

[0059] Among them, the evaluation index is defined as follows: Table 2

[0060] N represents the sample size, and respectively represent the actual value and the predicted value at time n.

[0061] It can be seen from the experimental results that all four evaluation indexes have decreased, indicating that the decomposition method proposed by the present invention can reduce the non-linearity degree of the original sequence compared with the traditional GRU prediction method, and further improve the prediction accuracy. It helps the power plant improve the frequency modulation response ability and further improve the power plant's benefits.

[0062] The present disclosure embodiment provides a computer device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device runs, the processor communicates with the memory through the bus. When the machine-readable instructions are executed by the processor, the steps provided in any embodiment of the present disclosure are executed.

[0063] The computer device provided by the embodiments of the present application includes a processor, a memory, and a bus. Among them, the memory is used to store execution instructions, including an internal memory and an external memory; here, the internal memory is also called the main memory, which is used to temporarily store the operation data in the processor and the data exchanged with external memories such as hard disks. The processor exchanges data with the external memory through the internal memory. When the electronic device is running, the processor communicates with the memory through the bus, so that the processor executes the following instructions: Obtain the original frequency modulation instruction signal sequence; For the original frequency modulation instruction signal sequence, obtain the regularization characteristic parameters representing its regularity degree, and the regularization characteristic parameters include a regularization index P and an average regularization degree value D; Based on the regularization characteristic parameters, adjust the original frequency modulation instruction signal sequence to generate a reference frequency modulation instruction signal sequence; Based on the original frequency modulation instruction signal sequence and the reference frequency modulation instruction signal sequence, respectively use a neural network for prediction to obtain an intermediate prediction result; Perform a mathematical operation on the intermediate prediction result to obtain a final prediction result.

[0064] The embodiments of the present disclosure provide a computer-readable storage medium, on which a computer program is stored. When the computer program is run by a processor, it executes the steps provided in any embodiment of the present disclosure. Among them, the storage medium can be a volatile or non-volatile computer-readable storage medium.

[0065] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments of the present disclosure can be implemented through hardware, or can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solutions of the embodiments of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a read-only optical disc, a USB flash drive, a mobile hard disk, etc.), including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present disclosure.

[0066] Those skilled in the art can understand that the drawings are only schematic diagrams of a preferred embodiment, and the modules or processes in the drawings are not necessarily essential for implementing the present disclosure.

[0067] Those skilled in the art can understand that the modules in the device in the embodiments can be distributed in the device in the embodiments according to the description of the embodiments, or can be correspondingly changed and located in one or more devices different from the present embodiment. The modules in the above embodiments can be combined into one module, or can be further split into multiple sub-modules.

[0068] The serial numbers of the above-mentioned embodiments of the present disclosure are only for description and do not represent the advantages or disadvantages of the embodiments.

[0069] Obviously, those skilled in the art can make various changes and modifications to the present disclosure without departing from the spirit and scope of the present disclosure. Thus, if these modifications and variations of the present disclosure fall within the scope of the claims of the present disclosure and their equivalent technologies, the present disclosure is also intended to include these modifications and variations.

[0070] Finally, it should be noted that the above is only an explanation of the present invention and is not used to limit the present invention. Although the present invention has been described in detail, those skilled in the art can still modify the aforementioned technical solutions or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for predicting energy storage frequency modulation instructions based on signal sequence regularization, characterized in that: include: S101, obtaining an original frequency modulation instruction signal sequence; S102, obtaining regularity characteristic parameters representing the regularity degree of the original frequency modulation instruction signal sequence, wherein the regularity characteristic parameters include a regularity index P and an average regularity degree value D; S103, adjusting the original frequency modulation instruction signal sequence based on the regularized characteristic parameters to generate a reference frequency modulation instruction signal sequence; S104, using a neural network to perform prediction based on the original frequency modulation instruction signal sequence and the reference frequency modulation instruction signal sequence to obtain an intermediate prediction result; S105: Perform mathematical operations on the intermediate prediction results to obtain the final prediction results.

2. According to claim 1, the energy storage frequency modulation instruction prediction method based on signal sequence regularization is characterized in that: In step S101, the original frequency modulation instruction signal sequence Pt is [X1, X2, X3, ..., X i ,…,X N ], Project each numerical point in the original frequency modulation command signal sequence Pt onto the plane coordinate system, the vertical axis represents the magnitude of the numerical value, and the horizontal axis represents the number of samplings, with one sampling per second.

3. The energy storage frequency modulation instruction prediction method based on signal sequence regularization according to claim 1 is characterized in that: Step S102 includes: Step 1: Determine whether there is a regular sequence in the signal sequence within the unit length; Step 2: Use the traversal algorithm to extract the original frequency modulation command signal sequence [X1, X2, X3, …, X i ,…,X N ] to find all regular number sequences, sum them up to get the number Q of all regular number sequences, and calculate the regularization index P; Step 3: Calculate the regularity value D of each regular series i , and obtain the average regularity degree value D of the original frequency modulation command signal sequence.

4. The energy storage frequency modulation instruction prediction method based on signal sequence regularization according to claim 3 is characterized in that: In step one, Six continuous signals are selected as the unit length; Select 6 consecutive signals [X q1 ,X q2 ,X q3 ,X q4 ,X q5 ,X q6 ], if X is satisfied q2 is the minimum value of 6 consecutive signals, and X q5 is the maximum value of 6 consecutive signals, then the consecutive signal [X q1 ,X q2 ,X q3 ,X q4 ,X q5 ,X q6 ] form a regular number sequence.

5. The energy storage frequency modulation instruction prediction method based on signal sequence regularization according to claim 4 is characterized in that: In step 2, Traverse and judge the original frequency modulation command signal sequence [X1,X2,X3,…,X i ,…,X N ] to determine whether every 6 consecutive signals form a regular sequence, and to obtain the number Q of all regular sequence by summing them up; Calculate the regularization index P = Q / N, where Q is the number of all regular series and N is the number of signals in the original frequency modulation command signal sequence.

6. The energy storage frequency modulation instruction prediction method based on signal sequence regularization according to claim 5 is characterized in that: In step three, Calculate the regularity value D for all regular series i ; One of the regularized sequences is [X q1 ,X q2 ,X q3 ,X q4 ,X q5 ,X q6 ], calculate the non-extreme point [X q1 ,X q3 ,X q4 ,X q6 The average value of avg , calculate the non-extreme point [X q1 ,X q3 ,X q4 ,X q6 ] Z , Regularized sequence [X q1 ,X q2 ,X q3 ,X q4 ,X q5 ,X q6 The regularity value D i yes ， sigmoid() is the activation function, and the two vertical line operators represent the absolute value; The regularity value D of all regular series i Calculate and obtain the average regularization degree value D, 。 7. The energy storage frequency modulation instruction prediction method based on signal sequence regularization according to claim 1 is characterized in that: Step S103 includes: The irregular number series in the original frequency modulation instruction signal sequence Pt is adjusted; the irregular number series includes continuous signals within a unit length, and 6 continuous signals are selected as a group for adjustment; Continuous signal [X c1 ,X c2 ,X c3 ,X c4 ,X c5 ,X c6 ] is the irregular sequence in the original frequency modulation command signal sequence Pt, where X c2 It must not be the minimum value, and X c5 It must not be a maximum value; [X c1 ,X c2 ,X c3 ,X c4 ,X c5 ,X c6 ] is the minimum value of X min ,[X c1 ,X c2 ,X c3 ,X c4 ,X c5 ,X c6 ] is the maximum value X max ; First, for X c2 Based on the regularization index P and the average regularization degree value D, mathematical transformation is performed to make it the minimum value. The mathematical transformation method is: ， Among them, rand() is a random function; Similarly, for X c5 Based on the regularization index P and the average regularization degree value D, mathematical transformation is performed to make it the maximum value. The mathematical transformation method is: ， Where rand() is a random function.

8. The energy storage frequency modulation instruction prediction method based on signal sequence regularization according to claim 7 is characterized in that: Step S103 also includes: Yes [X c1 ,X c3 ,X c4 ,X c6 ] generates a random value for each signal, X ci = rand ( X ci -exp( D ), X ci +exp( D )),i=1,3,4,6,rand() is a random function; Generate random values ​​in a continuous loop for a predetermined number of times; The calculated regularity value closest to the average regularity value D corresponds to [X c1 ,X c3 ,X c4 ,X c6 ] as the replacement value, and replace the value [X c1 ,X c3 ,X c4 ,X c6 ] and after mathematical transformation [X c2 ,X c5 ] After combining, we get [X c1 ,X c2 ,X c3 ,X c4 ,X c5 ,X c6 ]; After mathematical transformation of each irregular number series in the original frequency modulation command signal sequence, the reference frequency modulation command signal sequence Pt is obtained. 、 .

9. The energy storage frequency modulation instruction prediction method based on signal sequence regularization according to claim 8 is characterized in that: Step S104 includes: The final reference frequency modulation command signal sequence Pt 、 Put it into the GRU network for prediction and get the first intermediate prediction result Y1; The original frequency modulation command signal sequence Pt and the reference frequency modulation command signal sequence Pt 、 The difference sequence is put into the GRU network for prediction to obtain the second intermediate prediction result Y2; Step S105 includes: The final prediction result is Y3=Y1+Y2.

10. A system for predicting energy storage frequency modulation instructions based on signal sequence regularization, characterized in that: include, An instruction acquisition module, which acquires the original frequency modulation instruction signal sequence; A regularization characteristic parameter calculation module obtains regularization characteristic parameters representing the regularity degree of the original frequency modulation instruction signal sequence, wherein the regularization characteristic parameters include a regularization index P and an average regularity degree value D; An intermediate sequence generation module adjusts the original frequency modulation command signal sequence based on the regularized characteristic parameters to generate a reference frequency modulation command signal sequence; A prediction processing module, which uses a neural network to predict the original frequency modulation instruction signal sequence and the reference frequency modulation instruction signal sequence to obtain an intermediate prediction result; The prediction processing module performs mathematical operations on the intermediate prediction results to obtain the final prediction results.

Images