Short-term wind power prediction method, system and equipment based on signal decomposition and model optimization and medium

Abstract

The invention discloses a short-term wind power prediction method, system and equipment based on signal decomposition and model optimization, and a medium, and the method comprises the steps: carrying out the decomposition of normalized wind power data through N different signal decomposition algorithms, and obtaining N groups of wind power sequence sets after one-time decomposition; the method comprises the steps that N different signal decomposition methods are permutated and combined to obtain N * (N-1) combinations, then secondary decomposition is conducted on a wind power sequence to obtain N * (N-1) data sets, and weather forecast data are configured for the N * (N-1) data sets and N wind power sequence sets obtained after primary decomposition to form N * N model input data sets; the N * N groups of model input data sets are input into the optimized improved TimeMixer model, N * N prediction results are obtained, a final short-term wind power prediction result is calculated according to the N * N prediction results, and the short-term wind power can be accurately predicted through the method, the system, the equipment and the medium.

Description

Technical Field

[0001] The present invention belongs to the field of wind power prediction, and relates to a short-term wind power prediction method, system, equipment and medium based on signal decomposition and model optimization. Background Art

[0002] With the increasing maturity of wind power technology and the continuous increase in wind power installed capacity, the proportion of wind power in the total power generation of the power system has also increased year by year. Wind power generation depends on weather factors such as wind speed and wind direction. These factors are naturally highly random and volatile, which makes wind power output difficult to predict and has an increasing impact on the power grid. In order to help power system managers cope with the instability of wind power, improve wind power utilization efficiency, balance power grid load, and enhance the safety, stability and economy of the power system, wind power forecasting technology, especially short-term forecasting, is particularly important and has been widely studied and applied. At present, the main methods for wind power forecasting are physical methods, statistical methods and artificial intelligence methods, among which artificial intelligence methods are the hotspots of research. The widely used artificial intelligence models mainly include support vector machines, random forests, gradient boosting decision trees, various neural networks, etc.

[0003] At present, most models are based on machine learning models combined with other algorithms to obtain predicted power, such as empirical mode decomposition or wavelet decomposition combined with neural networks to predict wind power, but the prediction effect still needs to be improved. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a short-term wind power prediction method, system, device and medium based on signal decomposition and model optimization, which can accurately predict short-term wind power.

[0005] In order to achieve the above object, the present invention adopts the following technical scheme:

[0006] In one aspect, the present invention provides a short-term wind power prediction method based on signal decomposition and model optimization, comprising:

[0007] Collect weather forecast data and wind power data of wind farms, remove outliers and normalize the collected wind power data;

[0008] The normalized wind power data are decomposed using N different signal decomposition algorithms to obtain N sets of wind power sequence sets after one decomposition;

[0009] Arrange and combine N different signal decomposition methods to obtain N*(N-1) combinations, then perform secondary decomposition on the wind power sequence to obtain N*(N-1) sets of data sets, and configure weather forecast data for the N*(N-1) sets of data sets and the N sets of wind power sequence sets after the primary decomposition, respectively, to form N*N sets of model input data sets;

[0010] Input N*N groups of model input data sets into the optimized improved TimeMixer model to obtain N*N prediction results, and calculate the final short-term wind power prediction result according to the N*N prediction results.

[0011] The short-term wind power prediction method based on signal decomposition and model optimization described in the present invention is further improved in that:

[0012] Furthermore, the improved TimeMixer model is optimized by a hybrid solar wind optimization algorithm.

[0013] Furthermore, the process of calculating the final short-term wind power forecast result according to the N*N forecast results is as follows:

[0014] A stack denoising autoencoder is used to configure nonlinear weights for the N*N prediction results, and then the N*N prediction results are weighted summed to obtain a final short-term wind power prediction result.

[0015] Furthermore, N different signal decomposition algorithms include but are not limited to: wavelet packet decomposition (WPD), empirical mode decomposition (EMD), empirical wavelet transform (EWT), variational mode decomposition (VMD), characteristic mode decomposition (FMD), local mean decomposition (LMD), symplectic geometric mode decomposition (SGMD), extreme symmetric mode decomposition (ESMD), singular spectrum analysis (SSA) and group decomposition (SWD).

[0016] In a second aspect of the present invention, the present invention provides a short-term wind power prediction system based on signal decomposition and model optimization, comprising:

[0017] The acquisition module is used to collect the weather forecast data and wind power data of the wind farm, and to remove abnormal values ​​and normalize the collected wind power data;

[0018] A primary decomposition module is used to decompose the normalized wind power data using N different signal decomposition algorithms to obtain N sets of wind power sequence sets after primary decomposition;

[0019] A secondary decomposition module, which arranges and combines N different signal decomposition methods to obtain N*(N-1) combinations, and then performs secondary decomposition on the wind power sequence to obtain N*(N-1) sets of data sets, and configures weather forecast data for the N*(N-1) sets of data sets and the N sets of wind power sequence sets after the primary decomposition, respectively, to form N*N sets of model input data sets;

[0020] The prediction module is used to input N*N groups of model input data sets into the optimized improved TimeMixer model to obtain N*N prediction results, and calculate the final short-term wind power prediction result based on the N*N prediction results.

[0021] The short-term wind power prediction system based on signal decomposition and model optimization of the present invention is further improved in that:

[0022] Furthermore, the improved TimeMixer model is optimized by a hybrid solar wind optimization algorithm.

[0023] Furthermore, the process of calculating the final short-term wind power forecast result according to the N*N forecast results is as follows:

[0024] A stack denoising autoencoder is used to configure nonlinear weights for the N*N prediction results, and then the N*N prediction results are weighted summed to obtain a final short-term wind power prediction result.

[0025] Furthermore, N different signal decomposition algorithms include but are not limited to: wavelet packet decomposition (WPD), empirical mode decomposition (EMD), empirical wavelet transform (EWT), variational mode decomposition (VMD), characteristic mode decomposition (FMD), local mean decomposition (LMD), symplectic geometric mode decomposition (SGMD), extreme symmetric mode decomposition (ESMD), singular spectrum analysis (SSA) and group decomposition (SWD).

[0026] In a third aspect of the present invention, the present invention provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the short-term wind power prediction method based on signal decomposition and model optimization when executing the computer program.

[0027] In a fourth aspect of the present invention, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the short-term wind power prediction method based on signal decomposition and model optimization are implemented.

[0028] The present invention has the following beneficial effects:

[0029] The short-term wind power prediction method, system, equipment and medium based on signal decomposition and model optimization described in the present invention, during specific operation, decomposes the original wind power sequence into stable signals through primary and secondary signal decomposition technology and then predicts, which effectively copes with the non-stationarity of wind power, and the prediction effect is better than that of directly predicting wind power by a single model. In addition, the present invention adopts the improved TimeMixer model optimized by the hybrid solar wind algorithm for prediction, which improves the prediction accuracy of each wind power component, can solve the disadvantage of relatively large prediction error of a single model, and improves the final prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0031] Figure 1 is a flow chart of the method of the present invention;

[0032] Figure 2 It is a structural diagram of the improved TimeMixer model in the present invention;

[0033] Figure 3 Flowchart of the Hybrid Solar Wind Optimization algorithm (HSWO). DETAILED DESCRIPTION

[0034] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only an embodiment of a part of the present invention, not all embodiments, and is not intended to limit the scope of the present invention. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concepts disclosed in the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present invention.

[0035] The accompanying drawings show schematic diagrams of structures according to embodiments disclosed in the present invention. These figures are not drawn to scale, and some details are magnified and some details may be omitted for the purpose of clear expression. The shapes of various regions and layers shown in the figures and the relative sizes and positional relationships therebetween are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may additionally design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0036] Embodiment 1

[0037] refer to Figure 1 The short-term wind power prediction method based on signal decomposition and model optimization of the present invention comprises the following steps:

[0038] 1) Collect numerical weather forecast (NWP) data and wind power data of wind farms, remove outliers and normalize the collected wind power data;

[0039] 2) Decomposing the normalized wind power data using N different signal decomposition algorithms to obtain N sets of wind power sequence sets after primary decomposition; then permuting and combining the N different signal decomposition methods to obtain N*(N-1) combinations, performing secondary decomposition on the wind power sequence to obtain N*(N-1) sets of data sets, and configuring weather forecast data for the N*(N-1) sets of data sets and the N sets of wind power sequence sets after primary decomposition to form N*N sets of model input data sets;

[0040] 3) respectively inputting the N*N groups of model input data sets into the improved TimeMixer model optimized by the Hybrid Solar Wind Optimization (HSWO) algorithm to obtain N*N prediction results;

[0041] 4) A stacked denoising autoencoder (SDAE) is used to configure nonlinear weights for the N*N prediction results, and then weighted summation is performed on the N*N prediction results to obtain a final short-term wind power prediction result.

[0042] The N different signal decomposition algorithms are wavelet packet decomposition (WPD), empirical mode decomposition (EMD), empirical wavelet transform (EWT), variational mode decomposition (VMD), eigenmode decomposition (FMD), local mean decomposition (LMD), symplectic geometric mode decomposition (SGMD), extreme symmetric mode decomposition (ESMD), singular spectrum analysis (SSA) and group decomposition (SWD).

[0043] The secondary decomposition is as follows: first, decompose the wind power sequence using decomposition algorithm 1 to obtain components {x 1 ,x 2 ,…,x p}, then the component x with the highest frequency in the component set decomposed by decomposition algorithm 1 p , decompose using decomposition algorithm 2 to obtain {y 1 ,y 2 ,…,y q}, then the wind power subsequence set that needs to be predicted is {x 1 ,x 2 ,…,x p-1 ,y 1 ,…,y q}.

[0044] It should be noted that the TimeMixer model comes from an article published by Tsinghua University and Ant Group at ICLR 2024, "TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting". The model architecture diagram is as follows Figure 2 The present invention replaces the Linear predictor used in TimeMixer in the original paper with iTransformer to obtain I TimeMixer, as shown in Figure 2 As shown in Figure 2. iTransformer (Inverted Transformer) is an innovative time series prediction model that aims to solve the problems of performance degradation and computational explosion of traditional Transformer models when predicting time series with a large lookback window.

[0045] The hybrid solar wind optimization algorithm (HSWO) is a novel meta-heuristic optimization algorithm proposed in this paper. Figure 3 As shown in Figure 1, the specific steps of optimizing the hyperparameters of the improved TimeMixer model using HSWO are as follows:

[0046] a1) Initialize the value range of each hyperparameter of the improved TimeMixer model, with the upper bound as UB and the lower bound as LB, and set the number of iterations of the HSWO overall optimization algorithm to T*;

[0047] a2) Randomly generate N populations within the range of hyperparameter values, and the individuals in each population are equivalent to a set of model hyperparameters (parameter 1, parameter 2, ..., parameter D);

[0048] a3) using the prediction accuracy as the individual fitness value, calculating the fitness values ​​of all individuals in the N populations in step a2);

[0049] a4) performing a Harris Hawk optimization on each of the N populations, and finally obtaining N optimal individuals. The specific method of the Harris Hawk optimization is as follows: each population contains m individuals, the number of iterations of each population is set to T1, and the m individuals are globally explored, converted from global exploration to local development, and locally developed to update the population;

[0050] a5) Calculate the fitness value of each updated individual, and update the optimal individual representing the current number of iterations according to the size of the fitness value;

[0051] a6) Return to step a4) and continue to complete the Harris Hawk optimization until the upper limit of the number of iterations T1 is reached. Each population will have T1 historical optimal individuals. The fitness of these individuals is compared, and the individual with the highest fitness value is selected as the optimal individual of the population. Finally, N optimal individuals from N populations are obtained;

[0052] a7) The N optimal individuals obtained in step a6) are used as the initial population of the solar wind algorithm (SWO). Inspired by the solar wind, a stream of charged particles (solar wind) moves in the air under the influence of the magnetic field environment near the north and south poles of the earth, and is affected by the atmosphere, magnetic field and high-energy charged particles. The present invention proposes spiral motion through the movement process of high-energy charged particles spiraling around inductive magnetic lines; then, by comprehensively considering the energy, speed and trajectory of charged particles, as well as the composition and conditions in the atmosphere, the concept of circular wandering is proposed. Finally, through the phenomenon of continuous friction between high-energy charged particles during flight, a particle friction strategy is proposed. Set the current evaluation number s=0 of the SWO algorithm, the maximum evaluation number is S, the time step t2=0, and the maximum number of iterations is set to T2.

[0053] a8) The process of performing the solar wind optimization operation is:

[0054] a81) Calculate the fitness value of each individual, sort the particle swarm according to the fitness value, and update the current optimal solution x best ;

[0055] a82) Optimize by following the steps below:

[0056] a821) Calculate the speed of the individual according to formula (1).

[0057]

[0058] Among them, the integral constant Z, the charge q carried by the charged particles and the strength B of the earth's magnetic field are all constant. For simplicity, in the present invention, the values ​​of Z, q and B are set to 1; to increase randomness, the mass m of the solar wind charged particles is set to i The value of is a random number between [100,120], and the damping factor α is a random number between [0.01,0.012].

[0059] a822) Calculate the circular path of each high-energy particle individual, as shown in the following equations (2) to (5):

[0060] L(u)~|u -β ,1<β≤2.5(2)

[0061] A 0 =L(u)×(x avg (j)-x(i,j))+LB+r1 (UB-LB) / 2 (3)

[0062]

[0063] B 0 =L(u)×(x best (j)-x(i,j))+LB+r 1 (UB-LB) / 2(5)

[0064] The random walk strategy in formula (2) is used to enhance the global exploration ability of the metaheuristic algorithm because it is essentially a random non-Gaussian walk. Where β represents an important parameter for adjusting stability, u∈(1,e] represents the random step size, M is the total mass of the high-energy particle swarm, and x avg is the center of mass position of the high-energy particle group, which is calculated by equation (4). x(i,j) is the current position of the high-energy particle, and x avg (j)-x(i,j) represents the trend of particle movement. 0 and B 0 A circular complex simulated by a random walk drives energetic particles between the poles and the equator.

[0065] a823) Calculate three adaptive weights w that change with each iteration of the algorithm 1 、w 2 and 3 Specifically, according to equations (6) to (8), we can get:

[0066]

[0067]

[0068]

[0069] Among them, w 1 To control the weight of the spiral motion, w 2 and 3 is the weight of the circular step, as w 1 As the algorithm iterates, A 0 and B 0 The weight, that is, w 2 and w 3 , will decrease. During the algorithm iteration process, global search and local development rely on constantly changing weights to achieve balance and explore the optimal solution.

[0070] a824) Perform the following operations on each individual high-energy particle:

[0071] a8241) Update the particle position according to Equation (9).

[0072]

[0073] where x new (i, j) is the position of the energy particle after the update, r 2 ∈[0, 1], r 3 ∈[0, 1] represents the interference caused by uncontrollable environmental factors and other factors where the particle is located, and the constant c = 0.001.

[0074] a8242) Calculate the fitness value of the new particle, update x best and increase the evaluation times s.

[0075] a8243) When s < S, return to step a8241); otherwise, go to step a825) to update x new (i, j).

[0076] a825) When r 6 > 0.9, execute the particle friction strategy according to Equation (10) for new position update:

[0077] x new (i, j) = x(i, j) + r 4 w 2 B 0 + r 5 w 3 (x(k, j) - x(i, j))(10)

[0078] where r 4 ∈[0, 1], r 5 ∈[0, 1], r 6 ∈[0, 1] are random numbers.

[0079] a826) When the fitness value of x is greater than the fitness value of x new , update x through the greedy search algorithm; otherwise, sort x according to the fitness value, update x best , and increase the time step t2;

[0080] a9) Judge whether the current iteration number t2 is less than or equal to T2. If it is satisfied, go to step a8); otherwise, go to step a10);

[0081] a10) Judge whether the current total iteration number is ≤ T*. If it is satisfied, randomly select K (K < N) from the historical optimal individuals in SWO to replace K individuals in each Harris hawk population in step a2), and return to step a2) to continue optimization; if not, output the optimal solution.

[0082] Embodiment 2

[0083] To prove the effectiveness of the present invention, the data from a wind farm in Jiangsu from July 1, 2010 to July 12, 2012 are used to train and test the model, with a time resolution of 1 hour. The data from July 1, 2010 to May 21, 2012 are used for training, while the data after February 21, 2012 are used for testing. The standard mean absolute error and standard root mean square error of the three prediction models are shown in Table 1.

[0084] Table 1

[0085] method Standard mean absolute error Standardized root mean square error Method of the present invention 0.0261 0.0537 HSWO+ITimeMixer 0.0313 0.0557 ITimeMixer 0.0366 0.0566 TimeMixer 0.0378 0.0632

[0086] Embodiment 3

[0087] The short-term wind power prediction system based on signal decomposition and model optimization of the present invention comprises:

[0088] The acquisition module is used to collect the weather forecast data and wind power data of the wind farm and normalize the collected wind power data;

[0089] A primary decomposition module is used to decompose the normalized wind power data using N different signal decomposition algorithms to obtain N sets of wind power sequence sets after primary decomposition;

[0090] A secondary decomposition module, which arranges and combines N different signal decomposition methods to obtain N*(N-1) combinations, and then performs secondary decomposition on the wind power sequence to obtain N*(N-1) sets of data sets, and configures weather forecast data for the N*(N-1) sets of data sets and the N sets of wind power sequence sets after the primary decomposition, respectively, to form N*N sets of model input data sets;

[0091] The prediction module is used to input N*N groups of model input data sets into the optimized improved TimeMixer model to obtain N*N prediction results, and calculate the final short-term wind power prediction result based on the N*N prediction results.

[0092] In one embodiment of the present invention, the improved TimeMixer model is optimized by a hybrid solar wind optimization algorithm.

[0093] In one implementation mode of the present invention, the process of calculating the final short-term wind power forecast result according to the N*N forecast results is as follows:

[0094] A stack denoising autoencoder is used to configure nonlinear weights for the N*N prediction results, and then the N*N prediction results are weighted summed to obtain a final short-term wind power prediction result.

[0095] In one embodiment of the present invention, N different signal decomposition algorithms include but are not limited to: wavelet packet decomposition (WPD), empirical mode decomposition (EMD), empirical wavelet transform (EWT), variational mode decomposition (VMD), characteristic mode decomposition (FMD), local mean decomposition (LMD), symplectic geometric mode decomposition (SGMD), extreme symmetric mode decomposition (ESMD), singular spectrum analysis (SSA) and group decomposition (SWD).

[0096] The division of modules in the embodiments of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present application may be integrated into a processor, or may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0097] Embodiment 4

[0098] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the short-term wind power prediction method based on signal decomposition and model optimization are implemented. The memory may include a memory, such as a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk memory, etc. The processor, the network interface, and the memory are interconnected through an internal bus. The internal bus may be an industrial standard architecture bus, a peripheral component interconnection standard bus, an extended industrial standard structure bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The memory is used to store programs. Specifically, the program may include a program code, and the program code includes computer operation instructions. The memory may include a memory and a non-volatile memory, and provide instructions and data to the processor.

[0099] Embodiment 5

[0100] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the short-term wind power prediction method based on signal decomposition and model optimization are implemented. Specifically, the computer-readable storage medium includes, but is not limited to, for example, a volatile memory and / or a non-volatile memory. The volatile memory may include a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may include a read-only memory (ROM), a hard disk, a flash memory, an optical disk, a magnetic disk, etc.

[0101] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0102] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0103] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0104] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0105] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A short-term wind power prediction method based on signal decomposition and model optimization, characterized in that: include: Collect weather forecast data and wind power data of wind farms, remove outliers and normalize the collected wind power data; The normalized wind power data are decomposed using N different signal decomposition algorithms to obtain N sets of wind power sequence sets after one decomposition; Arrange and combine N different signal decomposition methods to obtain N*(N-1) combinations, then perform secondary decomposition on the wind power sequence to obtain N*(N-1) sets of data sets, and configure weather forecast data for the N*(N-1) sets of data sets and the N sets of wind power sequence sets after the primary decomposition, respectively, to form N*N sets of model input data sets; Input N*N groups of model input data sets into the optimized improved TimeMixer model to obtain N*N prediction results, and calculate the final short-term wind power prediction result according to the N*N prediction results.

2. The short-term wind power prediction method based on signal decomposition and model optimization according to claim 1 is characterized in that: The improved TimeMixer model is optimized by a hybrid solar wind optimization algorithm.

3. The short-term wind power prediction method based on signal decomposition and model optimization according to claim 1 is characterized in that: The process of calculating the final short-term wind power forecast result according to the N*N forecast results is as follows: A stack denoising autoencoder is used to configure nonlinear weights for the N*N prediction results, and then the N*N prediction results are weighted summed to obtain a final short-term wind power prediction result.

4. The short-term wind power prediction method based on signal decomposition and model optimization according to claim 1 is characterized in that: N different signal decomposition algorithms include but are not limited to: wavelet packet decomposition (WPD), empirical mode decomposition (EMD), empirical wavelet transform (EWT), variational mode decomposition (VMD), eigenmode decomposition (FMD), local mean decomposition (LMD), symplectic geometric mode decomposition (SGMD), extreme symmetric mode decomposition (ESMD), singular spectrum analysis (SSA) and group decomposition (SWD).

5. A short-term wind power prediction system based on signal decomposition and model optimization, characterized in that: include: The acquisition module is used to collect the weather forecast data and wind power data of the wind farm, and to remove abnormal values ​​and normalize the collected wind power data; A primary decomposition module is used to decompose the normalized wind power data using N different signal decomposition algorithms to obtain N sets of wind power sequence sets after primary decomposition; A secondary decomposition module, which arranges and combines N different signal decomposition methods to obtain N*(N-1) combinations, and then performs secondary decomposition on the wind power sequence to obtain N*(N-1) sets of data sets, and configures weather forecast data for the N*(N-1) sets of data sets and the N sets of wind power sequence sets after the primary decomposition, respectively, to form N*N sets of model input data sets; The prediction module is used to input N*N groups of model input data sets into the optimized improved TimeMixer model to obtain N*N prediction results, and calculate the final short-term wind power prediction result based on the N*N prediction results.

6. The short-term wind power prediction system based on signal decomposition and model optimization according to claim 5 is characterized in that: The improved TimeMixer model is optimized by a hybrid solar wind optimization algorithm.

7. The short-term wind power prediction system based on signal decomposition and model optimization according to claim 5 is characterized in that: The process of calculating the final short-term wind power forecast result according to the N*N forecast results is as follows: A stack denoising autoencoder is used to configure nonlinear weights for the N*N prediction results, and then the N*N prediction results are weighted summed to obtain a final short-term wind power prediction result.

8. The short-term wind power prediction system based on signal decomposition and model optimization according to claim 5 is characterized in that: N different signal decomposition algorithms include but are not limited to: wavelet packet decomposition (WPD), empirical mode decomposition (EMD), empirical wavelet transform (EWT), variational mode decomposition (VMD), eigenmode decomposition (FMD), local mean decomposition (LMD), symplectic geometric mode decomposition (SGMD), extreme symmetric mode decomposition (ESMD), singular spectrum analysis (SSA) and group decomposition (SWD).

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the short-term wind power prediction method based on signal decomposition and model optimization as described in any one of claims 1 to 4 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the short-term wind power prediction method based on signal decomposition and model optimization as described in any one of claims 1 to 4 are implemented.

Images