A method and system for predicting energy storage frequency modulation commands

Abstract

The present invention relates to the technical field of energy storage frequency modulation, in particular to a method and system for predicting energy storage frequency modulation commands, which includes decomposing a frequency modulation command sequence into multiple subsequences; determining the subsequence with the largest frequency change among all subsequences as the target subsequence, and generating a target virtual sequence to replace the target subsequence; using a GRU network to predict the target virtual sequence and the remaining subsequences to obtain corresponding multiple prediction results; and superimposing all the prediction results to obtain a target frequency modulation command prediction result. The beneficial effects of the present invention are that by using the variational mode decomposition algorithm to decompose the frequency modulation command sequence into multiple subsequences, and then processing and predicting the subsequences by the frequency method and the virtual sequence method, the frequency modulation commands can be predicted in advance, enabling the energy storage system to respond more quickly, reducing the response time difference, improving the frequency modulation performance index, and creating higher economic benefits for the power plant.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage frequency modulation, and particularly to a method and system for predicting energy storage frequency modulation commands. Background Art

[0002] With the continuous increase in the penetration rate of renewable energy in the power grid, the frequency modulation demand of the power grid has also increased accordingly. Energy storage systems, especially hybrid energy storage systems (such as the combination of supercapacitors and lithium-ion batteries), play an important role in assisting thermal power units in frequency modulation. However, there are some technical limitations in existing hybrid energy storage assisted frequency modulation methods.

[0003] In traditional frequency modulation methods, the difference between the frequency modulation command and the output of the thermal power unit is transmitted to the hybrid energy storage system, where the low-frequency part is borne by the battery and the high-frequency part is borne by the supercapacitor. A major problem with this method is that both signal transmission and the response of the energy storage system require time, resulting in a response time delay, which in turn affects the benefit value of frequency modulation and the overall economic efficiency of the power plant.

[0004] In addition, existing prediction methods, such as using variational mode decomposition (VMD) to decompose the original sequence into multiple subsequences, then making predictions separately, and finally superimposing the prediction results, have some defects. In particular, some of the subsequences after VMD decomposition have extremely high frequencies, and these high-frequency subsequences have strong randomness and non-linearity, resulting in difficult prediction and low prediction accuracy. Summary of the Invention

[0005] To solve the above technical problems, the present invention provides the following technical solutions:

[0006] In a first aspect, the present invention provides a method for predicting energy storage frequency modulation commands, including decomposing a frequency modulation command sequence into multiple subsequences;

[0007] Determining the subsequence with the largest frequency change among all subsequences as the target subsequence, and generating a target virtual sequence to replace the target subsequence;

[0008] Using a GRU network to predict the target virtual sequence and the remaining subsequences to obtain corresponding multiple prediction results;

[0009] Superimposing all the prediction results to obtain a target frequency modulation command prediction result;

[0010] Among them, determining the subsequence with the largest frequency change among all subsequences as the target subsequence includes:

[0011] Calculating the frequency change function corresponding to each subsequence;

[0012] Calculating the corresponding integral value according to each frequency change function;

[0013] Determine the target subsequence according to the magnitudes of all integral values.

[0014] As a preferred embodiment of the energy storage frequency modulation command prediction method of the present invention, wherein: decomposing the frequency modulation command sequence into a plurality of subsequences includes:

[0015] Use the variational mode decomposition algorithm to decompose the frequency modulation command sequence to obtain a plurality of intrinsic mode function subsequences.

[0016] As a preferred embodiment of the energy storage frequency modulation command prediction method of the present invention, wherein: determining the target subsequence according to the magnitudes of all integral values includes:

[0017] Calculate the absolute value corresponding to each integral value;

[0018] Determine the subsequence corresponding to the maximum value among all the absolute values as the target subsequence.

[0019] As a preferred embodiment of the energy storage frequency modulation command prediction method of the present invention, wherein: calculating the frequency change function corresponding to each subsequence includes:

[0020] Perform a difference operation on each intrinsic mode function subsequence to generate a difference sequence;

[0021] Perform an exponential calculation on each value of the difference sequence to generate a difference exponential sequence.

[0022] As a preferred embodiment of the energy storage frequency modulation command prediction method of the present invention, wherein: generating a target virtual sequence to replace the target subsequence includes:

[0023] In the coordinate system, perform a symmetric transformation on the target subsequence based on the horizontal line of Y = value M to generate a target virtual sequence;

[0024] The value M is the value of M corresponding to the minimum sum of the target functions;

[0025] The target function is:

[0026] ,

[0027] wherein, X i represents the i-th data point in the target subsequence, N is the total number of data points in the target subsequence, and M is an optimal value used to generate the target virtual sequence.

[0028] As a preferred embodiment of the energy storage frequency modulation command prediction method of the present invention, wherein: using a GRU network to predict the target virtual sequence and the remaining subsequences to obtain corresponding multiple prediction results further includes: correcting the prediction result of the target virtual sequence.

[0029] Second aspect, the present invention provides an energy storage frequency modulation command prediction system, including: a decomposition module, configured to decompose a frequency modulation command sequence into multiple subsequences;

[0030] a generation module, configured to determine the subsequence with the largest frequency change among all subsequences as the target subsequence, and generate a target virtual sequence to replace the target subsequence;

[0031] a prediction module, configured to use a GRU network to predict the target virtual sequence and the remaining subsequences to obtain corresponding multiple prediction results;

[0032] a superposition module, configured to superpose all prediction results to obtain a target frequency modulation command prediction result.

[0033] Third aspect, the present invention provides a computer device, including a memory and a processor, the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of an energy storage frequency modulation command prediction method are implemented.

[0034] Fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and is characterized in that when the computer program is executed by a processor, the steps of an energy storage frequency modulation command prediction method are implemented.

[0035] Advantages of the present invention: effectively solve the problems of long response time delay and low prediction accuracy of the hybrid energy storage assisting thermal power units in frequency modulation in the prior art. By using the variational mode decomposition (VMD) algorithm to decompose the frequency modulation command sequence into multiple subsequences, and then processing and predicting the subsequences by the frequency method and the virtual sequence method, the frequency modulation command can be predicted in advance, enabling the energy storage system to respond more quickly, reducing the response time difference, improving the frequency modulation performance index, and creating higher economic benefits for the power plant. In addition, by optimizing the algorithm to find the optimal value to generate a symmetric virtual sequence, the volatility of the high-frequency sequence is further reduced, and the prediction accuracy is significantly improved. Description of the Drawings

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0037] Figure 1 It is a flow schematic diagram of an energy storage frequency modulation command prediction method.

[0038] Figure 2 It is a flow schematic diagram of a traditional prediction method.

[0039] Figure 3Schematic diagram before the change of the target subsequence.

[0040] Figure 4 Schematic diagram after the change of the target subsequence. Specific implementation manner

[0041] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the specific implementation manner of the present invention in detail with reference to the accompanying drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0042] Example 1, refer to Figures 1 to 4 , which is the first embodiment of the present invention. This embodiment provides a method for predicting energy storage frequency modulation commands, including:

[0043] S1. Decompose the frequency modulation command sequence into multiple subsequences.

[0044] Further, decomposing the frequency modulation command sequence into multiple subsequences includes:

[0045] Use the variational mode decomposition algorithm to decompose the frequency modulation command sequence to obtain multiple intrinsic mode function subsequences.

[0046] It should be noted that assuming the frequency modulation command is Pt, after VMD decomposition, it becomes IMF1, IMF2, IMF3,..., IMF k (where: IMF k is the k-th subsequence), and further assume that a subsequence IMF i = [X i1 , X i2 , X i3 ,..., X iN (where: X iN is the value of the i-th IMF subsequence at the N-th time point).

[0047] S2. Determine the subsequence with the largest frequency change among all subsequences as the target subsequence, and generate a target virtual sequence to replace the target subsequence.

[0048] Further, determining the subsequence with the largest frequency change among all subsequences as the target subsequence includes:

[0049] Calculate the frequency change function corresponding to each subsequence.

[0050] Further, calculating the frequency change function corresponding to each subsequence includes:

[0051] Differential processing is performed on each subsequence of intrinsic mode functions to generate a difference sequence, where the difference sequence is obtained by calculating the differences between pairwise elements;

[0052] Exponential calculations are performed on each value of the difference sequence to generate a difference exponential sequence, where generating the difference numerical sequence is to quantify the relative importance of the differences and enhance the influence of larger differences through the exponential function;

[0053] Fitting or simulation is performed on the difference exponential sequence to generate a frequency variation function.

[0054] It should be noted that the difference sequence obtained after differential processing of the i-th subsequence is:

[0055] IMF ci =[X i1 -X i2 ,X i1 -X i3 ,X i1 -X i4 ,...,X i1 -X iN ,X i2 -X i3， X i2 -X i4 ,...,X i(N-1) -X iN (with a total of N(N - 1) / 2 values) (where: X iN is the value of the current subsequence at the N-th time point, and X i(N-1) is the value of the current subsequence at the N - 1-th time point)

[0056] Furthermore, each of the above difference sequences is multiplied by the exponential function e () The resulting difference exponential sequence is:

[0057] =[(X i1 -X i2 )(X i1 -X i3 ),(X i1 -X i4 ),...,(X i1 -X iN ),(X i2 -X i3 ),(X i2 -X i4 ),...,(X i(N-1) -X iN )]

[0058] Furthermore, a function f(t) is constructed through fitting or simulation software i, this function takes different values at different time points t, and these values are determined by the difference index sequence. Specifically:

[0059] When t = 1, f(t) i =(X i1 -X i2 );

[0060] When t = 2, f(t) i =(X i1 -X i3 );

[0061] When t = 3, f(t) i =(X i1 -X i4 );

[0062] ……

[0063] When t = N(N - 1) / 2, f(t) i =(X i(N-1) -X iN ).

[0064] Furthermore, the frequency change function is:

[0065] ,

[0066] where F(t) is the frequency change function;

[0067] A is the augmentation coefficient, specifically the natural logarithm of the ratio of the maximum value to the average value of the difference index sequence;

[0068] is the exponential decay function, used to calculate the integral value of the frequency change function F(t).

[0069] Calculate the corresponding integral value according to each frequency change function.

[0070] It should be noted that the integral value represents the area enclosed by f(t) and the X-axis, and is used to quantify the frequency change of the subsequence. The specific calculation formula is:

[0071] ,

[0072] where T is the total length of the time series;

[0073] A i represents the f(t) i area enclosed by the i-th subsequence and the X-axis;

[0074] f(t) i represents the frequency change function of the i-th subsequence.

[0075] Determine the target subsequence according to the magnitudes of all integral values.

[0076] Further, determining the target subsequence according to the magnitudes of all integral values includes:

[0077] Calculate the absolute value corresponding to each integral value;

[0078] Determine the subsequence corresponding to the maximum value among all the absolute values as the target subsequence.

[0079] Further, generating a target virtual sequence to replace the target subsequence includes:

[0080] Perform a symmetric transformation on the target subsequence based on the horizontal line of Y = value M in the coordinate system to generate a target virtual sequence;

[0081] The value M is the value of M corresponding to the minimum sum of the objective function;

[0082] The objective function is:

[0083] ,

[0084] where X i represents the i-th data point in the target subsequence, N is the total number of data points in the target subsequence, M is an optimal value used to generate the target virtual sequence, sin(X i ) is the sine function, and its input is the data point X i .

[0085] It should be noted that the solution method for the value M can use iterative loops or optimization algorithms such as (WOA, PSO), etc. Assume the target subsequence is IMF i = [X i1 , X i2 , X i3 , ……, X iN (where X iN is the value of the current subsequence at the N-th time point), establish a coordinate system, the abscissa of the coordinate system represents the subscript of the data points in the target subsequence, and the ordinate represents the specific values of the data points in the target subsequence, attached Figure 4 is the illustration after the symmetric transformation through the horizontal line of Y = value M, and predict by replacing the target subsequence with the target virtual sequence after the transformation. Figure 3

[0086] S3. Use a GRU network to predict the target virtual sequence and the remaining subsequences to obtain corresponding multiple prediction results.

[0087] Further, using a GRU network to predict the target virtual sequence and its subsequences, obtaining a plurality of corresponding prediction results further includes: correcting the prediction result of the target virtual sequence.

[0088] It should be noted that let the prediction result be [X i(N+1) , X i(N+2) , ……, X i(N+0.1N) , where N is the length of the sequence, i is the prediction step, N + 0.1N represents adding 10% of the sequence length to the original length N of the sequence as the prediction time point, and the correction formula is:

[0089] ,

[0090] In the formula, sigmoid(N + i) is an activation function used to control the amplitude of the correction;

[0091] Q is a constant used to adjust the output of the activation function to ensure that the corrected predicted value is within a reasonable range;

[0092] X i(N+1) : represents the predicted value of the i-th subsequence at the N + 1 time point;

[0093] X i(N+1) (the second one on the right side of the equation): represents the predicted value after correction.

[0094] S4. Superimpose all the prediction results to obtain the predicted result of the target frequency modulation command.

[0095] It should be noted that superimpose all the separately predicted prediction results (using the above prediction method) (this is a prior art and will not be elaborated here) to obtain the final predicted result of the target frequency modulation command.

[0096] In summary, a method for predicting energy storage frequency modulation commands of the present invention effectively solves the problems of long response time delay and low prediction accuracy in the prior art for hybrid energy storage to assist thermal power units in frequency modulation. By using the variational mode decomposition (VMD) algorithm to decompose the frequency modulation command sequence into multiple subsequences, and then processing and predicting the subsequences through the frequency method and the virtual sequence method, the frequency modulation command can be predicted in advance, enabling the energy storage system to respond more quickly, reducing the response time difference, improving the frequency modulation performance index, and creating higher economic benefits for the power plant. In addition, by optimizing the algorithm to find the optimal value to generate a symmetric virtual sequence, the volatility of the high-frequency sequence is further reduced, and the prediction accuracy is significantly improved.

[0097] Example 2 is the second example of the present invention. To verify the advantages of the present invention, the frequency modulation sequence is predicted by using the method of the present invention and the prediction method of VMD-GRU (a hybrid model combining variational mode decomposition and gated recurrent unit), and the results are shown in Table 1:

[0098] Table 1 Comparison table of prediction results

[0099] ,

[0100] It should be noted that the calculation formulas and definitions of the above four evaluation indexes are shown in Table 2:

[0101] Table 2 Operation formulas and definitions of each index

[0102] ,

[0103] Among them, N represents the sample size, and and respectively represent the actual value and the predicted value at time n.

[0104] To sum up, according to the experimental data, the errors of the prediction using this method are all smaller than the results of the VMD-GRU prediction, which fully shows that the prediction effect of using the present invention is better.

[0105] Example 3 is the third example of the present invention. The difference from the previous two examples is:

[0106] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0107] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, which can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or used in conjunction with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0108] More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which a program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or other appropriate processing as necessary, and then stored in a computer memory.

[0109] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0110] Example 4 is the fourth embodiment of the present invention, and this embodiment provides an energy storage frequency modulation instruction prediction system, including a decomposition module: for decomposing a frequency modulation instruction sequence into multiple subsequences;

[0111] a generation module: for determining the subsequence with the largest frequency change among all subsequences as the target subsequence, and generating a target virtual sequence to replace the target subsequence;

[0112] a prediction module: for using a GRU network to predict the target virtual sequence and the remaining subsequences to obtain corresponding multiple prediction results;

[0113] Superposition module: used to superpose all prediction results to obtain the predicted result of the target frequency modulation command.

[0114] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for predicting energy storage frequency modulation instructions, characterized in that: including decompose the frequency modulation instruction sequence into multiple subsequences; determine the subsequence with the largest frequency change among all subsequences as the target subsequence, and generate a target virtual sequence to replace the target subsequence; use a GRU network to predict the target virtual sequence and the remaining subsequences to obtain corresponding multiple prediction results; superimpose all the prediction results to obtain a target frequency modulation instruction prediction result; wherein, the determining the subsequence with the largest frequency change among all subsequences as the target subsequence includes: calculate the frequency change function corresponding to each subsequence; calculate the corresponding integral value according to each frequency change function; determine the target subsequence according to the magnitudes of all integral values; wherein, the determining the target subsequence according to the magnitudes of all integral values includes: calculate the absolute value corresponding to each integral value; determine the subsequence corresponding to the maximum value among all absolute values as the target subsequence; wherein, the generating a target virtual sequence to replace the target subsequence includes: perform a symmetry transformation on the target subsequence based on the horizontal line of Y = value M in the coordinate system to generate a target virtual sequence; the value M is the value of M corresponding to the minimum sum of the target functions.

2. The energy storage frequency modulation command prediction method according to claim 1, wherein: the decomposing the frequency modulation instruction sequence into multiple subsequences includes: use the variational mode decomposition algorithm to decompose the frequency modulation instruction sequence to obtain multiple intrinsic mode function subsequences.

3. The energy storage frequency modulation command prediction method according to claim 2, wherein: the calculating the frequency change function corresponding to each subsequence includes: perform a difference processing on each intrinsic mode function subsequence to generate a difference sequence; perform an exponential calculation on each value of the difference sequence to generate a difference exponential sequence; fit or simulate the difference exponential sequence to generate a frequency change function.

4. The energy storage frequency modulation command prediction method according to claim 3, wherein: the generating a target virtual sequence to replace the target subsequence includes: perform a symmetry transformation on the target subsequence based on the horizontal line of Y = value M in the coordinate system to generate a target virtual sequence; the value M is the value of M corresponding to the minimum sum of the target functions; the target function is: Among them, X i represents the i-th data point in the target subsequence, N is the total number of data points in the target subsequence, and M is an optimal value used to generate the target virtual sequence.

5. A method for predicting energy storage frequency modulation instructions according to claim 4, characterized in that: the using a GRU network to predict the target virtual sequence and the remaining subsequences to obtain corresponding multiple prediction results further includes: correcting the prediction result of the target virtual sequence.

6. A system adopting a method for predicting energy storage frequency modulation instructions as described in any one of claims 1 to 5, characterized in that: including a decomposition module: used to decompose the frequency modulation instruction sequence into multiple subsequences; a generation module: used to determine the subsequence with the largest frequency change among all subsequences as the target subsequence, and generate a target virtual sequence to replace the target subsequence; a prediction module: used to use a GRU network to predict the target virtual sequence and the remaining subsequences to obtain corresponding multiple prediction results; a superimposing module: used to superimpose all the prediction results to obtain a target frequency modulation instruction prediction result.

7. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of a method for predicting energy storage frequency modulation instructions according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of a method for predicting energy storage frequency modulation instructions according to any one of claims 1 to 5.

Images