A supercapacitor coupled lithium battery energy storage frequency modulation control method and system

Abstract

The present invention discloses a method and system for energy storage and frequency modulation control of a supercapacitor-coupled lithium battery, belonging to the field of energy storage and frequency modulation control technology. The method comprises collecting data from an original frequency modulation instruction sequence, generating an alternative sequence using a search method, and generating a regular optimal alternative sequence through positive, negative, and comprehensive optimization; training a GRU model using the optimal alternative sequence, and optimizing the hyperparameters of the GRU neural network using an improved whale optimization algorithm; inputting the optimal alternative sequence into the optimized GRU model to generate a preliminary prediction result, and correcting the prediction result in combination with the deviation between the prediction residual and the original data to obtain a final prediction result. The present invention significantly reduces prediction error, improves the prediction accuracy of the frequency modulation control system, effectively avoids local optimal problems, enhances the globality of the optimization process, and ensures that the generated optimal sequence has higher quality; the improved whale optimization algorithm significantly improves the algorithm convergence efficiency and avoids falling into a local optimal solution.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage and frequency modulation control, and in particular to a method and system for energy storage and frequency modulation control of a supercapacitor-coupled lithium battery. Background Art

[0002] The traditional hybrid energy storage (supercapacitor + lithium battery) method of assisting thermal power units in frequency regulation is to transmit the difference between the frequency regulation command and the thermal power unit to the hybrid energy storage. However, the signal transmission (frequency regulation command to supercapacitor / lithium battery) takes time, and the supercapacitor or lithium battery itself also takes time to respond. This will cause a certain response time difference, which not only affects the real-time performance of frequency regulation control, but also further affects the profit value and the revenue of the power plant. Therefore, the present invention proposes a prediction method to predict the size of the frequency regulation command in advance, enabling the supercapacitor / battery to take action in advance, thereby improving the profit value.

[0003] Traditional prediction methods often directly put the original sequence into a neural network for prediction and directly obtain the prediction results. However, due to the nonlinearity and irregularity of the original sequence, such prediction methods have large prediction errors and cannot meet the load demand in real time. The prediction result has large errors and it is difficult to ensure the economic benefits of the power plant. Therefore, the present invention proposes a prediction method using a search method. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: the existing frequency regulation control method of the hybrid energy storage system has response delay, low prediction accuracy, and reduced system benefits, and how to achieve more accurate frequency regulation prediction and faster response by optimizing the prediction method and improving the algorithm.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for controlling energy storage frequency modulation of a supercapacitor coupled lithium battery, comprising the following steps:

[0007] Data is collected from the original frequency modulation instruction sequence, and a search method is used to generate an alternative sequence. Through positive, negative and comprehensive optimization, a regular optimal alternative sequence is generated. The optimal alternative sequence is used to train the GRU (gated recurrent unit, a variant of the recurrent neural network) model. Based on the training results, the hyperparameters of the GRU neural network are optimized through an improved whale optimization algorithm. The optimal alternative sequence is input into the optimized GRU model to generate preliminary prediction results. The prediction results are corrected in combination with the deviation between the prediction residual and the original data to obtain the final prediction results.

[0008] As a preferred solution of the energy storage frequency modulation control method of a supercapacitor coupled lithium battery described in the present invention, wherein: the collected data includes:

[0009] Set the original FM sequence to , the front length is sequence;

[0010] Take the sequence of the previous length as input and predict the test group The value and prediction of unknown group The value of

[0011] Among them, all errors caused by the actual and predicted values of the test group are collected. is the sequence length, and i is the variable index.

[0012] As a preferred solution of the energy storage frequency modulation control method of a supercapacitor coupled lithium battery described in the present invention, wherein: the generating regular optimal replacement sequence includes: Generate regular optimal replacement sequences through search method;

[0013] Randomly generate 90 sets of replacement sequences ,set up Perform a forward search on any sequence of the first 30 groups and substitute it into the forward search formula to obtain the random range of the i-th value of the current sequence. The forward search formula is as follows:

[0014] ,

[0015] arrive Set to the middle sequence, with both positive search and negative search. is located in arrive Substitute any sequence into the mixed search formula to obtain the random range of the i-th value of the sequence. The mixed search formula is as follows:

[0016] ,

[0017] set up Perform a negative search on any sequence of the last 30 groups and substitute it into the forward search formula to get the random range of the i-th value of the current sequence. The negative search formula is as follows:

[0018] ,

[0019] Where W is The experimental group error is generated, is the i-th value of the original frequency modulation instruction sequence, is the random range of the i-th value of the j-th replacement sequence, i and j are variable indices, The baseline adjustment value, is a dynamic amplification factor that adjusts the upper bound of the search range through the linear combination of group number j and error W; w is the weight parameter corresponding to the error, is a natural constant, The dynamic shrinkage factor is used to control the lower bound of the negative search range.

[0020] As a preferred solution of the energy storage frequency modulation control method of a supercapacitor coupled lithium battery according to the present invention, wherein: generating the regular optimal replacement sequence further includes re-optimizing the generated sequence;

[0021] Set Alternative Sequence The error caused by the experimental group is ;

[0022] When the error generated during the forward search reaches the error generated by the experimental group, the corresponding sequence is , the corresponding error is ;

[0023] When the error generated during mixed search reaches the error generated by the experimental group, the corresponding sequence is The corresponding error is ;

[0024] When the error generated in the negative search reaches the error generated in the experimental group, the corresponding sequence is The corresponding error is ;

[0025] Will get , , Substitute the following optimization formula to obtain the optimal replacement sequence , the optimization formula is as follows:

[0026] ,

[0027] in, 、 、 are the minimum error values under the positive, mixed and negative search strategies respectively, 、 、 are the sequence values generated for different search strategies, is the error correction term, is the error value corresponding to the i-th position.

[0028] As a preferred solution of the energy storage frequency modulation control method of a supercapacitor coupled lithium battery described in the present invention, wherein: the improved whale optimization algorithm includes a first optimization and a second optimization;

[0029] The first optimization is specifically to optimize the hyperparameters of the GRU neural network, using a nonlinear decreasing strategy to improve the convergence factor;

[0030] Balancing the global and local search capabilities of the high whale optimization algorithm, the convergence factor The improved formula is as follows:

[0031] ,

[0032] in, It is expressed as the maximum number of iterations of the whale algorithm, t is the current number of iterations, Represented as activation function;

[0033] The improved whale optimization algorithm is used to optimize the hyperparameters of the GRU neural network. The optimization formula is as follows:

[0034] ,

[0035] in, represents a new hyperparameter combination, represents the best hyperparameter combination, represents the front particle position, Represents a random number.

[0036] As a preferred solution of the energy storage frequency modulation control method of a supercapacitor-coupled lithium battery described in the present invention, the second optimization includes optimizing the hyperparameters of the GRU neural network, setting the learning rate and the number of hidden layer neurons of the GRU neural network as the properties of the search particles in the space, and minimizing the output error as the objective function.

[0037] As a preferred solution of the energy storage frequency modulation control method of a supercapacitor coupled lithium battery of the present invention, wherein: the final prediction result includes inputting the optimal replacement sequence into the optimized GRU model;

[0038] The improved whale optimization algorithm is used to predict the optimal replacement sequence, and the prediction result is ;

[0039] The improved whale optimization algorithm is used to predict the difference sequence between the optimal replacement sequence and the original sequence. The prediction result is ;

[0040] The two prediction results are calculated based on the correlation to obtain the final prediction result, where the correlation calculation is:

[0041] ,

[0042] The final prediction result is ;

[0043] in, represents the optimal replacement sequence generated, The initial forecast value obtained for the forecast is It represents the difference prediction value obtained by using the improved whale optimization algorithm to predict the difference sequence between the optimal replacement sequence and the original sequence. It is a sine function with input π(N+i) and is used to introduce periodic correction.

[0044] Another object of the present invention is to provide a supercapacitor coupled lithium battery energy storage frequency modulation control system.

[0045] In order to solve the above technical problems, the present invention provides the following technical solutions: a supercapacitor coupled lithium battery energy storage frequency modulation control system, comprising an acquisition and adjustment module, an algorithm optimization module, and a prediction and correction module;

[0046] The acquisition and adjustment module acquires data from the original frequency modulation instruction sequence, generates a replacement sequence using a search method, and generates a regular optimal replacement sequence through positive, negative and comprehensive optimization;

[0047] The algorithm optimization module uses the optimal substitution sequence to train the GRU model, and based on the training results, optimizes the hyperparameters of the GRU neural network using the improved whale optimization algorithm;

[0048] The prediction correction module inputs the optimal replacement sequence into the optimized GRU model to generate a preliminary prediction result, and corrects the prediction result in combination with the deviation between the prediction residual and the original data to obtain the final prediction result.

[0049] The present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of a method for controlling energy storage and frequency modulation of a supercapacitor-coupled lithium battery when executing the computer program.

[0050] The present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the energy storage frequency modulation control method of a supercapacitor-coupled lithium battery are implemented.

[0051] Beneficial effects of the present invention: The energy storage frequency modulation method of the supercapacitor-coupled lithium battery provided by the present invention adopts a search method to generate the optimal alternative sequence, which significantly reduces the prediction error and improves the prediction accuracy of the frequency modulation control system. In addition, by optimizing the sequence generation method, the system can more effectively respond to changes in the frequency modulation instructions in advance, thereby improving the real-time performance of the frequency modulation response and maximizing the economic benefits of the power plant.

[0052] Through positive, negative and comprehensive optimization searches of random sequences, local optimal problems are effectively avoided, the globality of the optimization process is improved, the generated optimal sequence is ensured to have higher quality, the regularity of the sequence is enhanced, the nonlinear interference of the prediction algorithm is reduced, and higher prediction accuracy is ensured.

[0053] The improved Whale Optimization Algorithm significantly improves algorithm convergence efficiency and model training effectiveness, avoiding regression into local optimal solutions. The GRU neural network optimized using the Integral Whale Optimization Algorithm (IWOA) enhances the performance of energy storage frequency regulation control. The prediction error correction method reduces error accumulation, lowers the risk of regression into local optimal solutions, and improves execution stability. This invention achieves improved frequency regulation prediction accuracy, response efficiency, and economic benefits for power plants. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] 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.

[0055] Figure 1 This is an overall flow chart of a supercapacitor-coupled lithium battery energy storage and frequency modulation method provided by the first embodiment of the present invention. DETAILED DESCRIPTION

[0056] 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.

[0057] Example 1, with reference to Figure 1 , which is the first embodiment of the present invention, provides a method for controlling energy storage frequency modulation of a supercapacitor-coupled lithium battery, comprising:

[0058] S1: Collect data from the original FM instruction sequence, use the search method to generate an alternative sequence, and generate a regular optimal alternative sequence through positive, negative and comprehensive optimization.

[0059] Specifically, the original frequency modulation sequence is set to , the front length is sequence.

[0060] Take the sequence of the previous length as input and predict the test group The value and prediction of unknown group value.

[0061] The actual value and predicted value of the test group are used to generate all the errors (MAPE) in the present invention. What the present invention actually predicts is the value of the unknown group.

[0062] in, is the sequence length, and i is the variable index.

[0063] It should be noted that the original FM sequence Because of the strong non-periodicity and irregularity, directly putting it into the upgraded network will result in a large prediction error. Therefore, the present invention generates a regular optimal replacement sequence through a search method (this sequence has a strong periodicity and is more suitable for the prediction of the neural network).

[0064] Randomly generate 90 sets of replacement sequences ,set up Perform a forward search on any sequence of the first 30 groups and substitute it into the forward search formula to obtain the random range of the i-th value of the current sequence. The forward search formula is as follows:

[0065] ,

[0066] arrive Set to the middle sequence, with both positive search and negative search. is located in arrive Substitute any sequence into the mixed search formula to obtain the random range of the i-th value of the sequence. The mixed search formula is as follows:

[0067] ,

[0068] set up Perform a negative search on any sequence of the last 30 groups and substitute it into the forward search formula to obtain the random range of the i-th value of the current sequence. The negative search formula is as follows:

[0069] ,

[0070] Where W is The experimental group error is generated, is the i-th value of the original frequency modulation instruction sequence, is the random range of the i-th value of the j-th replacement sequence, i and j are variable indices, The baseline adjustment value, is a dynamic amplification factor that adjusts the upper bound of the search range through the linear combination of group number j and error W; w is the weight parameter corresponding to the error, is a natural constant, The dynamic shrinkage factor is used to control the lower bound of the negative search range.

[0071] Furthermore, the generated sequence is re-optimized;

[0072] Set Alternative Sequence The error caused by the experimental group is ;

[0073] When the error generated during the forward search reaches the error generated by the experimental group, the corresponding sequence is , the corresponding error is ;

[0074] When the error generated during mixed search reaches the error generated by the experimental group, the corresponding sequence is The corresponding error is ;

[0075] When the error generated in the negative search reaches the error generated in the experimental group, the corresponding sequence is The corresponding error is .

[0076] Will get , , Substitute the following optimization formula to obtain the optimal replacement sequence , the optimization formula is as follows:

[0077] ,

[0078] in, 、 、 are the minimum error values under the positive, mixed and negative search strategies respectively, 、 、 are the sequence values generated for different search strategies, is the error correction term, is the error value corresponding to the i-th position.

[0079] S2: Use the optimal replacement sequence to train the GRU model. Based on the training results, the hyperparameters of the GRU neural network are optimized using the improved whale optimization algorithm.

[0080] Specifically, the Whale Algorithm suffers from low solution accuracy and slow convergence when solving complex global optimization problems. This paper proposes a nonlinear decreasing strategy that better balances the global and local search capabilities of the Whale Algorithm. This prevents the whale population from prematurely switching from global to local search for prey, which could cause the algorithm to become trapped in a local optimum.

[0081] Therefore, the improved whale optimization algorithm includes a first optimization (improvement of the convergence factor) and a second optimization.

[0082] The first optimization is specifically to optimize the hyperparameters of the GRU neural network, using a nonlinear decreasing strategy to improve the convergence factor;

[0083] Balancing the global and local search capabilities of the high whale optimization algorithm, the convergence factor The improved formula is as follows:

[0084] ,

[0085] in, It is expressed as the maximum number of iterations of the whale algorithm, t is the current number of iterations, Represented as activation function;

[0086] The improved whale optimization algorithm is used to optimize the hyperparameters of the GRU neural network. The optimization formula is as follows:

[0087] ,

[0088] in, represents a new hyperparameter combination, represents the best hyperparameter combination, represents the front particle position, Represents a random number.

[0089] Furthermore, the second optimization includes optimizing the hyperparameters of the GRU neural network, setting the learning rate and the number of hidden neurons of the GRU neural network as the properties of the search particles in the space, and minimizing the output error as the objective function.

[0090] S3: Input the optimal replacement sequence into the optimized GRU model to generate a preliminary prediction result. The prediction result is corrected based on the deviation between the prediction residual and the original data to obtain the final prediction result.

[0091] It should be noted that the optimal replacement sequence is input into the optimized GRU model;

[0092] The improved whale optimization algorithm is used to predict the optimal replacement sequence, and the prediction result is ;

[0093] The improved whale optimization algorithm is used to predict the difference sequence between the optimal replacement sequence and the original sequence. The prediction result is ;

[0094] The two prediction results are calculated based on the correlation to obtain the final prediction result, where the correlation calculation is:

[0095] ,

[0096] The final prediction result is ;

[0097] in, represents the optimal replacement sequence generated, The initial forecast value obtained for the forecast is It represents the difference prediction value obtained by using the improved whale optimization algorithm to predict the difference sequence between the optimal replacement sequence and the original sequence. It is a sine function with input π(N+i) and is used to introduce periodic correction.

[0098] Example 2 is an embodiment of the present invention, which provides a supercapacitor-coupled lithium battery energy storage frequency modulation method. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0099] First, the effectiveness of the energy storage frequency regulation predictive control method combining the search method of the present invention with the decomposition optimization algorithm and the improved whale optimization algorithm (IWOA-GRU) in improving the prediction accuracy and response speed is verified.

[0100] In order to further verify the advantages of the present invention, the present invention uses the method of the present invention and the VMD-WOA-GRU prediction method to predict the frequency modulation sequence, and the results are as follows.

[0101] The prediction performance of the following methods is compared: VMD-WOA-ELM model (whale optimization algorithm for variational mode decomposition) and the search method of the present invention combined with IWOA-GRU (improved whale optimization algorithm) model. The prediction data are shown in Table 1.

[0102] Table 1 Prediction data table

[0103] ,

[0104] Prediction performance was evaluated using the following four metrics: MAE, SSE, RMSE, and MAPE. The model's adaptability and robustness were verified across various scenarios. The proposed search method, combined with the improved whale optimization algorithm (IWOA-GRU) model, outperformed other methods in terms of prediction accuracy and response speed. Across multiple experimental scenarios, the proposed method demonstrated low prediction errors (MAE, SSE, RMSE, and MAPE) and good adaptability. Table 2 demonstrates the practicality and effectiveness of the proposed method in energy storage frequency regulation predictive control.

[0105] Table 2 Evaluation indicators

[0106] ,

[0107] Where M represents the sample size, and Represent the actual value and predicted value at time n respectively.

[0108] Compared with traditional methods, this invention offers significant advantages in the following aspects. It optimizes the whale algorithm through a nonlinear decreasing strategy, balancing global and local search capabilities, resulting in faster convergence and more reasonable prediction model parameters. Incorporating the long-short-term memory characteristics of the GRU, it enhances the learning ability of complex time series data, demonstrating greater robustness, particularly with multi-dimensional feature inputs. It also dynamically corrects errors in prediction results to ensure their practical application value, making them more practical in frequency modulation control scenarios.

[0109] Example 3 is the third embodiment of the present invention, which differs from the first two embodiments in that:

[0110] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0111] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For 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.

[0112] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, and then editing, interpreting, or processing in another suitable manner as necessary, and then storing it in a computer memory.

[0113] It should be understood that various components of the present invention may be implemented using hardware, software, firmware, or a combination thereof. In the aforementioned embodiments, multiple steps or methods may be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any of the following technologies known in the art, or a combination thereof, may be used: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field-programmable gate array (FPGA), etc.

[0114] Example 4 is the fourth embodiment of the present invention, which provides a supercapacitor-coupled lithium battery energy storage frequency modulation control system, including an acquisition and adjustment module, an algorithm optimization module, and a prediction and correction module;

[0115] The acquisition and adjustment module acquires data from the original frequency modulation instruction sequence, generates a replacement sequence using a search method, and generates a regular optimal replacement sequence through positive, negative and comprehensive optimization;

[0116] The algorithm optimization module uses the optimal substitution sequence to train the GRU model, and based on the training results, optimizes the hyperparameters of the GRU neural network using the improved whale optimization algorithm;

[0117] The prediction correction module inputs the optimal replacement sequence into the optimized GRU model to generate a preliminary prediction result, and corrects the prediction result in combination with the deviation between the prediction residual and the original data to obtain the final prediction result.

[0118] 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 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 controlling energy storage frequency modulation of a supercapacitor coupled lithium battery, characterized in that: include, Collect data from the original frequency modulation instruction sequence, use the search method to generate alternative sequences, and generate the regular optimal alternative sequence through positive, negative and comprehensive optimization; The GRU model is trained using the optimal substitution sequence. Based on the training results, the hyperparameters of the GRU neural network are optimized using the improved whale optimization algorithm. The optimal replacement sequence is input into the optimized GRU model to generate a preliminary prediction result. The prediction result is corrected by combining the deviation between the prediction residual and the original data to obtain the final prediction result. The collected data includes: Set the original FM sequence to Pt = [X1, X2, X3, ... X i …,X N ], the front length is 0.9N[X1,X2,X3,…,X 0.9N ] sequence; Take the sequence of the previous length as input and predict the test group [X 0.9N+1 ,X 0.9N+2 ,X 0.9N+3 ,…,X N ] and predict the value of the unknown group [X N+1 ,X N+2 ,X N+3 ,…X N+0.1N ]; Among them, all errors generated by the actual and predicted values of the test group are collected, N is the sequence length, and i is the variable index; The generating regular optimal replacement sequence includes converting the original frequency modulation sequence Pt=[X1, X2, X3, ... X i …,X N ] Generate regular optimal replacement sequences through search method; Randomly generate 90 groups of replacement sequences Pt1, Pt2…Pt 90 , let Pt j Perform a forward search on any sequence of the first 30 groups and substitute it into the forward search formula to obtain the random range of the i-th value of the current sequence. The forward search formula is as follows: Pt 31 To Pt 60 Set as the middle sequence, there are both positive search and negative search, let Pt j It is located in Pt 31 To Pt 60 Substitute any sequence into the mixed search formula to obtain the random range of the i-th value of the sequence. The mixed search formula is as follows: Assume Pt j Perform a negative search on any sequence of the last 30 groups and substitute it into the forward search formula to get the random range of the i-th value of the current sequence. The negative search formula is as follows: Where W is the experimental group error caused by Pt, X i is the i-th value of the original frequency modulation instruction sequence, x ji is the random range of the i-th value of the j-th replacement sequence, i and j are variable indices, X i * The baseline adjustment value, e 1+W！ is a dynamic amplification factor that adjusts the upper bound of the search range through the linear combination of group number j and error W; w is the weight parameter corresponding to the error, e is a natural constant, and e w*i The dynamic shrinkage factor is used to control the lower bound of the negative search range; Generating the regular optimal replacement sequence further includes re-optimizing the generated sequence; Set the replacement sequence Pt1, Pt2...Pt 90 The errors generated by the experimental group are W1, W2…W 90 ; When the error generated during the forward search reaches the error generated by the test group, the corresponding sequence is Pt Q , the corresponding error is W 1min ; When the error generated during mixed search reaches the error generated by the test group, the corresponding sequence is Pt W The corresponding error is W 2min ; When the error generated during the negative search reaches the error generated by the test group, the corresponding sequence is Pt E The corresponding error is W 3min ; The obtained W 1min , W 2min , W 3min Substitute the following optimization formula to obtain the optimal replacement sequence x Ri , the optimization formula is as follows: Among them, W 1min 、W 2min 、W 3min are the minimum error values under the positive, mixed and negative search strategies respectively, x Qi 、x Wi 、x Ei Sequence values generated by different search strategies, sigmoid(e i ) is the error correction term, e i is the error value corresponding to the i-th position; The improved whale optimization algorithm includes a first optimization and a second optimization; The first optimization is specifically to optimize the hyperparameters of the GRU neural network, using a nonlinear decreasing strategy to improve the convergence factor; To balance the global search and local search capabilities of the high whale optimization algorithm, the convergence factor a is improved as follows: Among them, T max It represents the maximum number of iterations of the whale algorithm, t is the current number of iterations, and tanh is the activation function; The improved whale optimization algorithm is used to optimize the hyperparameters of the GRU neural network. The optimization formula is as follows: X(t+1)=X * (t)-a·|C·X * (t)-X(t)| Among them, X(t+1) represents the new hyperparameter combination, X * (t) represents the optimal hyperparameter combination, X(t) represents the front particle position, and C represents a random number; The second optimization includes optimizing the hyperparameters of the GRU neural network, setting the learning rate and the number of hidden layer neurons of the GRU neural network as the properties of the search particles in the space, and minimizing the output error as the objective function; The final prediction result includes inputting the optimal replacement sequence into the optimized GRU model; The improved whale optimization algorithm is used to predict the optimal replacement sequence, and the prediction result is [X′ N+1 ,X′ N+2 ,X′ N+3 ,…X′ N+i …X′ N+0.1N ]; The improved whale optimization algorithm is used to predict the difference sequence between the optimal replacement sequence and the original sequence, and the prediction result is [X” N+1 ,X” N+2 ,X” N+3 ,…X” N+i …X” N+0.1N ]; The two prediction results are calculated based on the correlation to obtain the final prediction result, where the correlation calculation is: X N+i =X' N+i -X” N+i +e sinπ(N+i) The final prediction result is [X N+1 ,X N+2 ,X N+3 ,…X N+i …X N+0.1N ]; Among them, Pt R represents the optimal replacement sequence generated, X' N+0.1N The initial predicted value obtained for prediction, X” N+i It represents the difference prediction value obtained by using the improved whale optimization algorithm to predict the difference sequence between the optimal replacement sequence and the original sequence. sinπ(N+i) is a sine function with input π(N+i) used to introduce periodic correction.

2. A supercapacitor coupled lithium battery energy storage frequency modulation control system, applying a supercapacitor coupled lithium battery energy storage frequency modulation control method according to any one of claims 1, characterized in that: Including acquisition adjustment module, algorithm optimization module, and prediction correction module; The acquisition and adjustment module acquires data from the original frequency modulation instruction sequence, generates a replacement sequence using a search method, and generates a regular optimal replacement sequence through positive, negative and comprehensive optimization; The algorithm optimization module uses the optimal substitution sequence to train the GRU model, and based on the training results, optimizes the hyperparameters of the GRU neural network using the improved whale optimization algorithm; The prediction correction module inputs the optimal replacement sequence into the optimized GRU model to generate a preliminary prediction result, and corrects the prediction result in combination with the deviation between the prediction residual and the original data to obtain the final prediction result.

3. 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 control method of a supercapacitor-coupled lithium battery according to claim 1 are implemented.

4. 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 control method of a supercapacitor-coupled lithium battery according to claim 1 are implemented.

Images