A method and system for predicting energy storage frequency modulation instructions based on signal sequence folding rate

Abstract

A method and system for predicting energy storage frequency modulation instructions based on a signal sequence folding rate, which obtain an original frequency modulation instruction signal sequence; obtain optimization parameters for each signal in the original frequency modulation instruction signal sequence, wherein the optimization parameters include a first folding rate Z1, a first average value Q1, and a first variance C1; generate an auxiliary signal sequence based on the optimization parameters, and obtain a combined signal sequence; use a neural network to perform prediction based on the auxiliary signal sequence and the combined signal sequence to obtain an intermediate prediction result; perform a decoupling operation on the intermediate prediction result to obtain a final prediction result; make the prediction result of the frequency modulation instruction more accurate, predict the size of the frequency modulation instruction in advance for energy storage control, and improve response accuracy and frequency modulation benefits.

Description

Technical Field

[0001] The present application relates to the technical field of power grid frequency regulation, and in particular to a method and system for predicting energy storage frequency regulation instructions based on signal sequence folding rate. Background Art

[0002] Currently, large hydropower and thermal power units (coal-fired and gas-fired) are the primary frequency regulation sources in my country's major regional power grids. Frequency regulation power output is adjusted to respond to system frequency changes. Hybrid energy storage frequency regulation command prediction involves analyzing historical frequency regulation commands and related data to develop a prediction model to predict future frequency regulation command requirements. This optimizes the control strategy of the hybrid energy storage system and improves its frequency regulation performance and efficiency. This system can enhance system performance and provide rapid response. Output power can be adjusted quickly to respond to grid frequency changes, effectively balancing supply and demand within the power system and maintaining grid frequency stability. Improved regulation accuracy allows for more precise tracking of frequency regulation commands, reducing regulation errors and improving power quality and stability. Enhanced system stability can be achieved by rationally allocating power and energy across different energy storage devices, reducing overall system risk, improving reliability and stability, and mitigating the risk of system failure due to failure or performance degradation of a single energy storage device. Prediction can also extend device life and optimize charging and discharging strategies. Based on the prediction results, a reasonable charging and discharging strategy can be formulated to avoid excessive or frequent charging and discharging of energy storage devices, thereby extending their service life. Balanced equipment usage: Rationally allocating the usage frequency and load of different energy storage devices ensures relatively even aging across all devices, reducing maintenance costs and replacement frequency. This reduces costs and improves economic efficiency. By improving frequency regulation performance and efficiency, hybrid energy storage systems can achieve greater frequency regulation benefits while reducing power outages and equipment damage costs caused by unstable grid frequency. Investment costs can also be optimized. Based on forecast results, the capacity and quantity of different energy storage devices can be rationally allocated, avoiding overinvestment and waste of resources, and improving return on investment.

[0003] The traditional prediction method is to directly put the original FM sequence into the GRU network for prediction. The original FM sequence has strong nonlinearity, which increases the difficulty of neural network prediction. The reasons are as follows:

[0004] 1. Capturing complex patterns;

[0005] Nonlinear dynamic characteristics: Nonlinear time series may contain complex patterns such as mutations, multimodal fluctuations, chaotic behavior (such as Lorentz systems), etc. Traditional linear models cannot fully model these patterns. Although neural networks can theoretically approximate these patterns, they require higher model complexity.

[0006] 2. Challenges in model training;

[0007] Overfitting risk: Complex nonlinear relationships require deeper networks and more parameters, but are prone to overfitting with small sample sizes. Optimization difficulty: The non-convexity of the loss function can lead to falling into local optima. For example, when training GRUs, careful selection of learning rates and initialization strategies is crucial.

[0008] 3. Data requirements and quality;

[0009] Data requirements: Learning nonlinear patterns often requires large amounts of data. For example, predicting chaotic systems may require tens of thousands of time steps of samples to cover different states.

[0010] 4. Uncertainty in dynamic systems;

[0011] Sensitivity to chaotic behavior: Even if the model fits historical data well, long-term forecasts may still fail due to exponentially amplified errors in initial conditions.

[0012] Directly applying the original sequence to prediction results in inaccurate predictions and large prediction errors. Therefore, the original FM sequence must be preprocessed before prediction. To address this issue, decomposition and denoising are proposed. VMD decomposition or wavelet transforms are used to separate trends, cycles, and residuals, reducing nonlinear complexity.

[0013] like Figure 2 As shown, common methods, such as VMD decomposition, decompose the original FM sequence into a series of low-frequency subsequences, which are then fed into a neural network for prediction. However, while VMD (Variational Mode Decomposition) improves the accuracy of non-stationary signal processing, it also suffers from the following drawbacks: complex parameter tuning and high computational cost. Wavelet transforms are highly dependent on the time basis function and have limited frequency resolution. Summary of the Invention

[0014] In view of the shortcomings of the existing technology, the present invention proposes a prediction method that can reduce the nonlinearity and irregularity of the original sequence without using decomposition and noise reduction.

[0015] A method for predicting energy storage frequency modulation instructions based on signal sequence folding rate, comprising:

[0016] S101, obtaining an original frequency modulation instruction signal sequence;

[0017] S102, obtaining optimization parameters for each signal in the original frequency modulation instruction signal sequence, wherein the optimization parameters include a first folding rate Z1, a first average value Q1, and a first variance C1;

[0018] S103, generating an auxiliary signal sequence based on the optimized parameters, and obtaining a combined signal sequence;

[0019] S104, using a neural network to perform prediction based on the auxiliary signal sequence and the combined signal sequence to obtain an intermediate prediction result;

[0020] S105: Perform decoupling operations on the intermediate prediction results to obtain the final prediction results.

[0021] A system for predicting energy storage frequency modulation instructions based on signal sequence folding rate, comprising:

[0022] The command acquisition module acquires the original frequency modulation command signal sequence;

[0023] An optimization parameter calculation module obtains optimization parameters for each signal in the original frequency modulation instruction signal sequence, wherein the optimization parameters include a first folding rate Z1, a first average value Q1, and a first variance C1;

[0024] An intermediate sequence generation module generates an auxiliary signal sequence based on the optimized parameters and obtains a combined signal sequence;

[0025] The prediction processing module uses a neural network to predict based on the auxiliary signal sequence and the combined signal sequence to obtain an intermediate prediction result;

[0026] The decoupling and restoration module performs decoupling operations on the intermediate prediction results to obtain the final prediction results.

[0027] The beneficial effects of the present invention are:

[0028] Reducing the nonlinearity of the original signal sequence makes the corrected sequence smoother, greatly reduces the difficulty of prediction, and improves the accuracy of frequency regulation instruction prediction, thereby improving the response accuracy of the power plant and the frequency regulation benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 A diagram of the method steps.

[0030] Figure 2 This is an illustration of the traditional algorithm. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0032] It should be understood that the step numbers used herein are only for convenience of description and are not intended to limit the order in which the steps are to be executed.

[0033] It should be understood that the terms used in this specification are only for the purpose of describing specific embodiments and are not intended to limit this application. As used in this specification and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0034] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0035] The term "and / or" refers to and includes any and all possible combinations of one or more of the associated listed items.

[0036] The embodiment of the present disclosure provides a method for predicting energy storage frequency modulation instructions based on the signal sequence folding rate, such as Figure 1 Shown, including:

[0037] S101, obtaining an original frequency modulation instruction signal sequence;

[0038] S102, obtaining optimization parameters for each signal in the original frequency modulation instruction signal sequence, wherein the optimization parameters include a first folding rate Z1, a first average value Q1, and a first variance C1;

[0039] S103, generating an auxiliary signal sequence based on the optimized parameters, and obtaining a combined signal sequence;

[0040] S104, using a neural network to perform prediction based on the auxiliary signal sequence and the combined signal sequence to obtain an intermediate prediction result;

[0041] S105: Perform decoupling operations on the intermediate prediction results to obtain the final prediction results.

[0042] The frequency modulation command prediction method based on supercapacitor energy storage, provided in the embodiments of the present disclosure, uses novel supercapacitors and lithium batteries as energy storage devices and modulates the frequency of a power plant's power grid when its frequency fluctuates. The method monitors the power plant's power grid in real time and generates a corresponding frequency modulation command signal when the power supply frequency fluctuates. After acquiring this original signal, the method then predicts future signals to implement frequency modulation of the power plant's power grid.

[0043] The above-mentioned S101, obtaining the original frequency modulation instruction signal sequence, includes:

[0044] The original frequency modulation command signal sequence is Pt=[ X1,X2,X3,…,X i ,…,X N ]. The original frequency modulation instruction signal sequence Pt is a function of time t.

[0045] Project each numerical point in the original frequency modulation command signal sequence Pt onto the plane coordinate system, and the vertical axis represents the magnitude of the numerical value, which is the signal value X1, X2, X3, ..., X i ,…,X N , the horizontal axis is the number of sampling times, one sample per second.

[0046] In another embodiment disclosed in the present application, for the original frequency modulation instruction signal sequence, the stronger the degree of nonlinearity, the greater the difficulty of prediction. The degree of nonlinearity of the original frequency modulation instruction signal sequence can be characterized by the folding rate.

[0047] The above S102, obtaining optimization parameters for each signal in the original frequency modulation instruction signal sequence, wherein the optimization parameters include the first folding rate Z1, the first average value Q1, and the first variance C1; includes:

[0048] Step 1: Determine whether there is effective signal folding in the signal sequence within the unit length.

[0049] A certain number of continuous sampling points are selected per unit length, such as 3 continuous sampling points X i 、X i+1 、X i+2 ,

[0050] When X is satisfied i+1 It's X i 、X i+1 、X i+2 The maximum or minimum of the three,

[0051] and meet the conditions

[0052]

[0053] , max() is for finding the maximum value, min() is for finding the minimum value, exp() is the natural exponential function,

[0054] Then the sampling point X i 、X i+1 、X i+2 This constitutes an effective signal folding.

[0055] Step 2: Use the traversal algorithm to find all valid signal folds from the original FM instruction signal sequence and obtain the fold number M.

[0056] Traverse and judge [X1,X2,X3,…,X i ,…,X N ] Every 3 consecutive sampling points X i 、X i+1 、X i+2, such as (X1, X2, X3), (X2, X3, X4), ..., (X N-2 、X N-1 、X N ), and the statistical summation is used to obtain the number M of folds of all valid signals.

[0057] Step 3: Calculate the first folding rate Z1, the first average value Q1, and the first variance C1 of the original frequency modulation command signal sequence;

[0058] The first folding rate Z1=M / N; N is the number of signals in the original FM command signal sequence;

[0059] The first average value Q1=avg[X1,X2,X3,…,X i ,…,X N ]; avg() is the average function;

[0060] First variance C1=Var[X1,X2,X3,…,X i ,…,X N ]; Var() is the variance function;

[0061] The above-mentioned S103, generating an auxiliary signal sequence based on the optimized parameters and obtaining a combined signal sequence, includes:

[0062] Assume that the auxiliary signal sequence is Pt 、 =[X1 、 ,X2 、 ,X3 、 ,…,X i 、 ,…,X N 、 ].

[0063] First, an auxiliary signal sequence is randomly generated;

[0064] X i 、 Generate an example, X i 、 is a randomly generated value, X i 、 The value is located at

[0065] between.

[0066] [X1 、 ,X2 、 ,X3 、 ,…,X i 、 ,…,X N 、 ] is generated accordingly.

[0067] Secondly, the auxiliary signal sequence is verified based on the optimized parameters, and the auxiliary signal sequence that meets the conditions is determined;

[0068] The auxiliary signal sequence Pt 、 [X1 、 ,X2 、 ,X3 、 ,…,X i 、 ,…,X N 、 ] and the original FM command signal sequence Pt [X1, X2, X3, ..., X i ,…,X N ] Add each item one by one (that is, add each item) to obtain the combined signal sequence Pt 、、 =[X1 、、 ,X2 、、 ,X3 、、 ,…,X i 、、 ,…,X N 、、 ].

[0069] Calculate the combined signal sequence Pt 、、 [X1 、、 ,X2 、、 ,X3 、、 ,…,X i 、、 ,…,X N 、、 ]'s second folding rate Z2, second average value Q2, and second variance C2, and performing a verification operation on them with the first folding rate Z1, first average value Q1, and first variance C1;

[0070] Repeatedly randomly generate auxiliary signal sequence Pt 、 [X1 、 ,X2 、 ,X3 、 ,…,X i 、 ,…,X N 、 ] until the ratio of the second folding rate Z2, the second average value Q2, the second variance C2 to the first folding rate Z1, the first average value Q1, the first variance C1 meets the set threshold condition in the verification operation.

[0071] Set the threshold conditions as follows:

[0072] Z2 / Z1<0.7, and Q2 / Q1∈【0.8 , 1.2】, and C2 / C1<0.7.

[0073] Finally, determine the auxiliary signal sequence Pt that meets the threshold condition 、 and the combined sequence Pt 、、 , used in subsequent predictions.

[0074] The above-mentioned S104, using a neural network to perform prediction based on the auxiliary signal sequence and the combined signal sequence to obtain an intermediate prediction result, includes:

[0075] The final combined sequence Pt 、、 Put it into the GRU network for prediction and get the first intermediate prediction result Y1=[Y1,Y2,Y3,…,Y S ].

[0076] The final auxiliary sequence Pt 、 Put it into the GRU network for prediction and get the second intermediate prediction result Y2=[Y1 、 ,Y2 、 ,Y3 、 ,…,Y S 、 ].

[0077] The above S105, performing decoupling operations on the intermediate prediction results to obtain the final prediction results, includes:

[0078] The first intermediate prediction result and the second intermediate prediction result are the prediction results of the combined signal sequence and the auxiliary signal sequence, but are not the prediction results of the original frequency modulation instruction signal sequence Pt, and therefore require decoupling operations.

[0079] The final prediction result of the decoupling operation is Y3=[Y1 、、 ,Y2 、、 ,Y3 、、 ,…,Y S 、、 ],

[0080] Y S 、、 For example,

[0081] , the denominator is the number of folds M.

[0082] Y3=[Y1 、、 ,Y2 、、 ,Y3 、、 ,…,Y S 、、 ] is the final prediction result.

[0083] With the above Figure 1 The method shown corresponds to the embodiment of the present disclosure.

[0084] A system for predicting energy storage frequency modulation instructions based on signal sequence folding rate, characterized by comprising:

[0085] The command acquisition module acquires the original frequency modulation command signal sequence;

[0086] An optimization parameter calculation module obtains optimization parameters for each signal in the original frequency modulation instruction signal sequence, wherein the optimization parameters include a first folding rate Z1, a first average value Q1, and a first variance C1;

[0087] An intermediate sequence generation module generates an auxiliary signal sequence based on the optimized parameters and obtains a combined signal sequence;

[0088] The prediction processing module uses a neural network to predict based on the auxiliary signal sequence and the combined signal sequence to obtain an intermediate prediction result;

[0089] The decoupling and restoration module performs decoupling operations on the intermediate prediction results to obtain the final prediction results.

[0090] In order to further verify the advantages of the present invention, the present invention uses the method of the present invention and the GRU prediction method to predict the frequency modulation sequence, and the results are as follows.

[0091] The experimental frequency modulation instructions were obtained from a power plant in Inner Mongolia. The acquisition time for frequency modulation sequence 1 was from 4:00 to 18:00 on a certain day in November 2024, with one data point collected every 1 second. The acquisition time for frequency modulation sequence 2 was from 0:00 to 22:00 on a certain day in February 2025, with one data point collected every 1 second.

[0092]

[0093] The evaluation indicators are defined as follows:

[0094]

[0095] N represents the sample size, and Represent the actual value and predicted value at time n respectively.

[0096] From the experimental results, we can see that all four evaluation indicators have been reduced, which shows that the proposed model can well improve the prediction accuracy.

[0097] An embodiment of the present disclosure provides a computer device, comprising: a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps provided in any embodiment of the present disclosure are performed.

[0098] The computer device provided in the embodiment of the present application includes a processor, a memory, and a bus. The memory is used to store and execute instructions, and includes internal memory and external memory. The internal memory here is also called internal memory, which is used to temporarily store the calculation data in the processor and the data exchanged with the external memory such as the hard disk. The processor exchanges data with the external memory through the internal memory. When the electronic device is running, the processor and the memory communicate through the bus, so that the processor executes the following instructions:

[0099] Obtaining the original frequency modulation command signal sequence;

[0100] Obtaining optimization parameters for each signal in the original frequency modulation command signal sequence, the optimization parameters including a first folding rate Z1, a first average value Q1, and a first variance C1;

[0101] generating an auxiliary signal sequence based on the optimized parameters and obtaining a combined signal sequence;

[0102] Based on the auxiliary signal sequence and the combined signal sequence, a neural network is used to predict and obtain an intermediate prediction result;

[0103] Perform decoupling operations on the intermediate prediction results to obtain the final prediction results.

[0104] The present disclosure provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, executes the steps provided in any embodiment of the present disclosure. The storage medium may be a volatile or non-volatile computer-readable storage medium.

[0105] Through the above description of the embodiments, those skilled in the art will clearly understand that the embodiments of the present disclosure can be implemented through hardware or through software plus the necessary general hardware platform. Based on this 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 (such as a read-only optical disk, a USB flash drive, a mobile hard disk, etc.) and includes a number of instructions for causing a computer device (such as a personal computer, a server, or a network device) to execute the methods described in the various embodiments of the present disclosure.

[0106] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes in the accompanying drawings are not necessarily required for implementing the present disclosure.

[0107] Those skilled in the art will appreciate that the modules in the devices of the embodiments may be distributed in the devices of the embodiments as described in the embodiments, or may be located in one or more devices different from the embodiments with corresponding changes. The modules of the above embodiments may be combined into one module or further split into multiple submodules.

[0108] The serial numbers of the above-mentioned embodiments of the present disclosure are for description only and do not represent the advantages or disadvantages of the embodiments.

[0109] Obviously, those skilled in the art may 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 equivalents, the present disclosure is intended to include these modifications and variations.

[0110] Finally, it should be noted that the foregoing description is merely an explanation of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail, those skilled in the art will be able to modify the aforementioned technical solutions or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for predicting energy storage frequency modulation instructions based on signal sequence folding rate, characterized in that: include: S101, obtaining an original frequency modulation instruction signal sequence; S102, obtaining optimization parameters for each signal in the original frequency modulation instruction signal sequence, wherein the optimization parameters include a first folding rate Z1, a first average value Q1, and a first variance C1; S103, generating an auxiliary signal sequence based on the optimized parameters, and obtaining a combined signal sequence; S104, using a neural network to perform prediction based on the auxiliary signal sequence and the combined signal sequence to obtain an intermediate prediction result; S105: Perform decoupling operations on the intermediate prediction results to obtain the final prediction results; Step S102 includes: Step 1: determine whether there is effective signal folding in the signal sequence within the unit length; Step 2: Use the traversal algorithm to find all valid signal folds from the original FM command signal sequence Pt and obtain the fold number M; Step 3: Calculate the first folding rate Z1, the first average value Q1, and the first variance C1 of the original frequency modulation command signal sequence Pt; In step one; The unit length is selected as 3 consecutive sampling points X i 、X i+1 、X i+2 , When X is satisfied i+1 It's X i 、X i+1 、X i+2 The maximum or minimum value among the three, and meeting the conditions , max() is for maximum value, min() is for minimum value, exp() is for natural exponential function, then the sampling point X i 、X i+1 、X i+2 Constitute an effective signal fold; In step 2; Traverse and judge the original frequency modulation instruction signal sequence Pt [X1,X2,X3,…,X i ,…,X N ] Every 3 consecutive sampling points X i 、X i+1 、X i+2 , statistically sum to obtain the number M of folds of all valid signals; In step three; The first folding rate Z1=M / N; N is the number of signals in the original FM command signal sequence; The first average value Q1=avg[X1,X2,X3,…,X i ,…,X N ]; avg() is the average function; First variance C1=Var[X1,X2,X3,…,X i ,…,X N ]; Var() is the variance function.

2. The energy storage frequency modulation instruction prediction method based on the signal sequence folding rate according to claim 1 is characterized in that: In step S101, the original frequency modulation instruction signal sequence is Pt = [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 folding rate according to claim 1 is characterized in that: In step S103; First, an auxiliary signal sequence Pt is randomly generated 、 =[X1 、 ,X2 、 ,X3 、 ,…,X i 、 ,…,X N 、 ]; where X i 、 is a randomly generated value, X i 、 The value is located at between; Secondly, the auxiliary signal sequence is verified based on the optimized parameters, and the auxiliary signal sequence that meets the conditions is determined; The auxiliary signal sequence Pt 、 [X1 、 ,X2 、 ,X3 、 ,…,X i 、 ,…,X N 、 ] and the original FM command signal sequence Pt [X1, X2, X3, ..., X i ,…,X N ] Add them one by one to get the combined signal sequence Pt 、、 =[X1 、、 ,X2 、、 ,X3 、、 ,…,X i 、、 ,…,X N 、、 ]; Calculate the combined signal sequence Pt 、、 [X1 、、 ,X2 、、 ,X3 、、 ,…,X i 、、 ,…,X N 、、 ]'s second folding rate Z2, second average value Q2, and second variance C2, and performing a verification operation on them with the first folding rate Z1, first average value Q1, and first variance C1; Repeatedly randomly generate auxiliary signal sequence Pt 、 [X1 、 ,X2 、 ,X3 、 ,…,X i 、 ,…,X N 、 ] until the ratio of the second folding rate Z2, the second average value Q2, the second variance C2 to the first folding rate Z1, the first average value Q1, and the first variance C1 meets the set threshold condition in the verification operation; Finally, determine the auxiliary signal sequence Pt that meets the threshold condition 、 and combined signal sequence Pt 、、 , used in subsequent predictions.

4. The energy storage frequency modulation instruction prediction method based on the signal sequence folding rate according to claim 3 is characterized in that: The threshold conditions are as follows: Z2 / Z1<0.7, and Q2 / Q1∈【0.8 , 1.2】, and C2 / C1<0.

7.

5. The energy storage frequency modulation instruction prediction method based on signal sequence folding rate according to claim 3 is characterized in that: Step S104 includes: The final combined signal sequence Pt 、、 Put it into the GRU network for prediction and get the first intermediate prediction result Y1=[Y1,Y2,Y3,…,Y S ]; The final auxiliary signal sequence Pt 、 Put it into the GRU network for prediction and get the second intermediate prediction result Y2=[Y1 、 ,Y2 、 ,Y3 、 ,…,Y S 、 ].

6. The energy storage frequency modulation instruction prediction method based on signal sequence folding rate according to claim 5 is characterized in that: Step S105 includes: The final prediction result of the decoupling operation is Y3=[Y1 、、 ,Y2 、、 ,Y3 、、 ,…,Y S 、、 ], in, , the denominator is the number of folds M.

7. A system for predicting energy storage frequency modulation instructions based on signal sequence folding rate, characterized in that: include, The command acquisition module acquires the original frequency modulation command signal sequence; An optimization parameter calculation module obtains optimization parameters for each signal in the original frequency modulation instruction signal sequence, wherein the optimization parameters include a first folding rate Z1, a first average value Q1, and a first variance C1; An intermediate sequence generation module generates an auxiliary signal sequence based on the optimized parameters and obtains a combined signal sequence; The prediction processing module uses a neural network to predict based on the auxiliary signal sequence and the combined signal sequence to obtain an intermediate prediction result; The decoupling and restoration module performs decoupling operations on the intermediate prediction results to obtain the final prediction results.

Images