Medium and long term wind speed prediction method and system

Abstract

The invention is suitable for the technical field of wind speed prediction, and provides a medium and long term wind speed prediction method and system, and the method comprises the following steps: collecting wind speed data, and constructing an input-output data set; searching an optimal intrinsic mode component number K and a penalty coefficient alpha in the VMD by adopting an artemisinin optimization algorithm, and obtaining an IMF component; an LSTM network model is constructed, the hidden unit number, the maximum training period and the initial learning rate of the LSTM network model are used as targets of a mirage optimization algorithm for optimization, optimal parameters of the LSTM network model are reserved, and a FATA-LSTM network model is obtained; training, testing and verifying the FATA-LSTM network model by using the IMF component, and generating a wind speed prediction model; and outputting a prediction result. According to the method, multi-scale features are effectively extracted, noise interference is reduced, generalization ability is improved, and prediction accuracy in a complex wind speed environment is higher.

Description

Technical Field

[0001] The present invention belongs to the technical field of wind speed prediction, and particularly relates to a medium and long-term wind speed prediction method and system. Background Technique

[0002] Wind energy plays an important role in solving carbon emission problems and energy shortages. However, its significant non-stationarity and intermittency pose challenges to the stable operation of the power grid, power generation efficiency, and power quality. Precise wind speed prediction technology is crucial for ensuring the stable operation of the power system and optimizing economic dispatching, and can help wind farms formulate scientific and reasonable control strategies, thereby improving the utilization efficiency of wind energy.

[0003] Current wind speed prediction methods include physical models (such as numerical weather prediction), statistical models (such as ARIMA), and intelligent models based on machine learning. In recent years, hybrid models combining signal decomposition and deep learning have received extensive attention due to their high accuracy.

[0004] However, there are bottlenecks in existing medium and long-term wind power prediction technologies: the variational mode decomposition (VMD) method lacks adaptability. In signal sources with low signal-to-noise ratio and frequency domain aliasing, it is prone to over-filtering or incomplete filtering phenomena. At the same time, the speculation of the number of signal sources is often inaccurate. When decomposing signal sources, problems such as incorrect filtering of signal sources, inaccurate speculation of the number of signal sources, and excessive time-frequency domain distribution often occur; the fitting ability of the LSTM model highly depends on the setting of hyperparameters, and traditional hyperparameter tuning methods are inefficient and inaccurate; the real-time monitoring and prediction method based on sensor data has high costs, is limited by the accuracy and reliability of equipment, has low economy, and has limited promotion prospects.

[0005] In summary, the existing technology mainly shows problems such as low prediction accuracy and poor economy in medium and long-term wind power prediction, and it is difficult to effectively support the further development of wind power. Summary of the Invention

[0006] The purpose of the embodiments of the present invention is to provide a medium and long-term wind speed prediction method, aiming to solve the problems proposed in the above background technique.

[0007] The embodiments of the present invention are implemented as follows. A medium and long-term wind speed prediction method includes the following steps:

[0008] Collect wind speed data and construct an input-output data set;

[0009] Use the artemisinin optimization algorithm to find the optimal number of intrinsic mode functions K and the penalty coefficient α in VMD, and obtain IMF components;

[0010] Build an LSTM network model, optimize the number of hidden units, maximum training epochs, and initial learning rate of the LSTM network model as the objectives of the mirage optimization algorithm, retain the optimal parameters of the LSTM network model, and obtain the FATA-LSTM network model;

[0011] Use the IMF components to train, test, and validate the FATA-LSTM network model to generate a wind speed prediction model;

[0012] Output the prediction result.

[0013] Another object of the embodiments of the present invention is to provide a medium- and long-term wind speed prediction system for implementing the above medium- and long-term wind speed prediction method, including:

[0014] An acquisition module for acquiring wind speed data and constructing an input-output data set;

[0015] An IMF component acquisition module for using the artemisinin optimization algorithm to find the optimal number of intrinsic mode functions K and penalty coefficient α in VMD and obtain the IMF components;

[0016] An FATA-LSTM network model construction module for building an LSTM network model, optimizing the number of hidden units, maximum training epochs, and initial learning rate of the LSTM network model as the objectives of the mirage optimization algorithm, retaining the optimal parameters of the LSTM network model, and obtaining the FATA-LSTM network model;

[0017] A wind speed prediction model generation module for using the IMF components to train, test, and validate the FATA-LSTM network model to generate a wind speed prediction model;

[0018] An output module for outputting the prediction result.

[0019] A medium- and long-term wind speed prediction method provided by the embodiments of the present invention optimizes the number of hidden units, maximum training epochs, and initial learning rate of the LSTM network model by using the mirage optimization algorithm (FATA), solves the problem of difficult accurate selection of hyperparameters, and can accurately output the wind speed change trend in real time through the prediction of IMF components;

[0020] By combining the AO-VMD decomposition technology to preprocess the original wind speed signal, effectively extract multi-scale features and reduce noise interference, improve the generalization ability of the model, and make the prediction accuracy of the model higher in a complex wind speed environment. Description of the Drawings

[0021] Figure 1 It is a flowchart of a medium- and long-term wind speed prediction method provided by the embodiments of the present invention;

[0022] Figure 2 Structural diagram of a medium - and long - term wind speed prediction system provided by an embodiment of the present invention;

[0023] Figure 3 Summer wind speed data graph provided by an embodiment of the present invention;

[0024] Figure 4 Comparison graph of prediction results of each model provided by an embodiment of the present invention. Detailed implementation manners

[0025] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0026] The following describes the specific implementation of the present invention in detail with reference to specific embodiments.

[0027] As Figure 1 shown, it is a flowchart of a medium - and long - term wind speed prediction method provided by an embodiment of the present invention, including the following steps:

[0028] S1. Collect historical wind speed time - series data of the target wind farm, construct an input - output data set, and prepare data for subsequent decomposition and prediction:

[0029] Obtain hourly wind speed records, denoted as the original signal data, with a length of T;

[0030] Use the sliding window method to construct a data set. Take the wind speed data of 24 consecutive hours as the input feature x, predict the wind speed value y of the next hour, and generate a feature matrix x and a target vector y with the number of samples being T - 24.

[0031] S2. Optimize the variational mode decomposition (VMD) for wind speed data and influencing factors by using the artemisinin optimization algorithm (AO) to obtain a set of IMF components with different frequency - domain characteristics. The artemisinin optimization algorithm realizes finding the optimal number of intrinsic mode components K and the penalty coefficient α by dynamically adjusting the penalty factor α and the number of modes K:

[0032] S201. Define the range of the joint optimization parameter set of the core parameters of the VMD decomposition: where the range of K is from 2 to 30, and the range of α is from 500 to 8000, respectively representing the preset lower limit to the upper limit of the number of decomposition modes and the preset lower limit to the upper limit of the bandwidth constraint;

[0033] S202. Initialize the artemisinin population using the good point set strategy, which can overcome the problem that the original AO's initialization process based on uniform random numbers cannot fully cover the search space, thus avoiding key problems such as poor convergence performance and difficulty in maintaining population diversity. Each artemisinin agent represents a set of VMD parameter combinations. , and the specific process is as follows:

[0034] Determine the population size N = 2, the problem dimension dim = 2 (solving K and α in the VMD parameters of two problems), and the maximum number of iterations MaxFEs = 20;

[0035] Within the parameter range (the lower bound lb = [2, 500] and the upper bound ub = [30, 8000] of the parameter ), find the smallest prime number greater than 2 * dim + 3 as the benchmark for generating the good point set, and randomly generate the initial population pop, expressed as:

[0036] ;

[0037] ;

[0038] ;

[0039] where ; ; ;

[0040] S203. Execute the comprehensive elimination stage, simulate the comprehensive elimination effect in the initial stage of artemisinin treatment of malaria, and search the VMD parameter space through the global perturbation strategy:

[0041] S2031. Initialize the artemisinin population using the good point set strategy and calculate the fitness of each agent: , and the objective function is the reconstruction error of VMD decomposition:

[0042] ; <U+

[0043] In the formula, is the original wind speed signal; is the k-th IMF component;

[0044] Obtain the current optimal agent best and the optimal fitness Bestfitness, corresponding to the current optimal VMD parameter combination ;

[0045] S2032. Enter the iterative loop, calculate the dynamic control parameters, and update the dynamic control parameters according to the current function evaluation times FEs and the maximum evaluation times MaxFEs and the disturbance amplitude control parameter :

[0046] ;

[0047] ;

[0048] S2033. Traverse each artemisinin agent i and dimension j (j = 1 corresponds to K, j = 2 corresponds to α), perform VMD decomposition on each agent position pop(i, j), calculate the fitness Fitness(i), and evaluate whether the position needs to be updated;

[0049] S2034. Judge and update the agent position. If the random number , perform global disturbance update:

[0050] ;

[0051] ;

[0052] S204. Execute the local clearance stage, simulate the local clearance effect of artemisinin in a specific infected area, and optimize the VMD parameters through the local adjustment strategy:

[0053] S2041. Calculate the normalized fitness:

[0054] ;

[0055] If , perform local disturbance update:

[0056] ;

[0057] In the formula, A is a random permutation of the population index; , rand is a random number from 0 to 1;

[0058] S205. Execute the late consolidation stage, simulate the residual cleaning and rechecking effects of artemisinin in the final stage, and optimize the VMD parameters through the mutation and boundary processing mechanism (whether it is the previous comprehensive elimination, local clearance, or mutation, it may cause the particle position to exceed the predefined search space boundary. Therefore, in the embodiments of the present invention, the boundary processing mechanism is used to adjust all out-of-bounds dimensions back to the legal range):

[0059] S2051. Complete the mutation position update:

[0060] ;

[0061] Among them, is the original position of the population; is the current global optimal solution;

[0062] S2052. Check whether the population is out of bounds:

[0063] ;

[0064] If not satisfied, perform the following operations:

[0065] Set a random number to randomly determine whether to execute the reflection strategy or the adaptive reset strategy. Set the probability of executing the reflection strategy to 30% and the adaptive reset strategy to 70%:

[0066] Reflection strategy:

[0067] If it exceeds the upper bound:

[0068] ;

[0069] If it exceeds the lower bound:

[0070] ;

[0071] Perform the clamping operation:

[0072] If the initial is far beyond the boundary, a single reflection may cause to fall outside the other side of the boundary. The clamping operation is used to ensure that the value after reflection is still within the effective search space The formula is:

[0073] ;

[0074] Adaptive reset strategy:

[0075] ;

[0076] In the meta - heuristic optimization algorithm, the position update operations of the population (such as moving, mutation, etc.) may cause it to go out of bounds, that is, the coordinates of the population in a certain dimension exceed the predefined upper and lower limits of the search space , so how to efficiently and reasonably handle these out - of - bounds populations is a key problem that must be faced;

[0077] The artemisinin algorithm adopts a relatively simple and direct strategy: when the population goes out of bounds in a certain dimension, it will be forced to reset to the position of the current global optimum (best) in the corresponding dimension, which brings obvious limitations: loss of diversity, premature convergence, decline in global search ability, etc. To overcome the defects of the original AO, the embodiments of the present invention introduce a boundary handling method that combines reflection and adaptive reset mechanisms:

[0078] Reflection mechanism: When a population crosses the boundary, a reflection operation is performed with a low probability. The core idea is to "bounce" the out-of-bounds population back into the search space from the boundary, just like light hitting a mirror.

[0079] Adaptive reset: When a population crosses a boundary, it is reset to the position of the current global optimal solution in the corresponding dimension. This ensures that the vast majority of out-of-bounds populations can be quickly guided to the most promising area currently known, accelerating the algorithm's convergence and preventing the population from getting "lost" in the search space.

[0080] S205. Determine whether the current evaluation number exceeds the maximum evaluation number. If so, take the current global optimal position as the final output to obtain the optimized VMD parameter combination. Otherwise, return to step S202 to continue optimizing.

[0081] S3. Build an LSTM network model. Use the number of hidden units, maximum training cycle, and initial learning rate of the LSTM network model as the targets of the mirage optimization algorithm for optimization. Retain the optimal parameters of the LSTM network model and obtain the FATA-LSTM network model:

[0082] S301, the number of hidden units of the LSTM network model , maximum training cycle and the initial learning rate As the light individual in the mirage optimization algorithm, the population is initialized as follows:

[0083] Determine the population size N=10, the maximum number of evaluations MaxFEs=20, within the parameter range (i.e. parameter , upper boundary ) randomly generates the initial population:

[0084] ;

[0085] Where, Indicates the position of the i-th ray individual in the j-th dimension; and The upper and lower bounds of the two points in the j-th dimension respectively; r is a random number between (0,1); i=1, 2, ..., N;

[0086] The mirage optimization algorithm is divided into a refraction phase and a reflection phase, which are determined by a random number r1 (between (0, 1)) and the group quality factor IP, as well as a random number r2 (between (0, 1)) and the body quality factor p. If r1>IP, it is the refraction phase and enters the direct refraction mode; if r1≤IP and r2<p, it enters the indirect refraction mode; if r1≤IP and r2≥p, it enters the total internal reflection mode.

[0087] S302. Simulation of the refraction stage, simulating the optical refraction process of a mirage. In the initial stage of optimization, individual light rays explore the hyperparameter space through refraction behavior, with a relatively wide coverage range. As the iteration progresses, the refraction range gradually converges. The refraction stage is divided into a direct refraction mode and an indirect refraction mode. The conversion between the direct refraction mode and the indirect refraction mode is determined by the population quality factor IP and the individual quality factor p. The specific process includes:

[0088] S3021. Simulation of the direct refraction stage, initialize , calculate the fitness of each individual light ray:

[0089] , the objective function is the mean square error of LSTM on the test set:

[0090] ;

[0091] In the formula, is the data of the original data test set; is the model prediction data;

[0092] Obtain the current global optimal position gBest and the optimal fitness gBestScore, sort the fitness , calculate the population quality factor:

[0093] ;

[0094] In the formula, represents the cumulative integral sum of the current population; is the worst historical integral value; is the best historical integral value; is the worst integral value, is a very small positive number;

[0095] Compare the random number r1 with IP. If r1 > IP, enter the direct refraction mode, and update the position as:

[0096] ;

[0097] In the formula, rand is a random number between (0, 1). If the fitness of the new position is better than the current one, update ;

[0098] S3022. Simulation of the indirect refraction mode. If r1 ≤ IP, calculate the individual quality factor p:

[0099] ;

[0100] The random number r2 is compared with p. If r2 < p, enter the indirect refraction mode, and update the position to:

[0101] ;

[0102] In the formula, is the value of the global optimal position in the j-th dimension, , ;

[0103] If the fitness of the new position is better than the current one, update .

[0104] S303. Simulation reflection stage, simulate the total internal reflection process of mirage, and adjust the LSTM hyperparameters through light reflection:

[0105] Initialize the population using the result of step S302. If r1 ≤ IP and r2 ≥ p, enter the reflection stage, and update the position to:

[0106] ;

[0107] ;

[0108] In the formula, is the individual index of the random reference ray; ; ; is the reflection coefficient;

[0109] If the fitness of the new position is better than the current one, update [[ID=4,7]] ;

[0110] S304. Determine whether the current evaluation times FEs exceed the maximum evaluation times MaxFEs. If exceeded, determine the current global optimal position as the best hyperparameter combination of the LSTM network model, otherwise return to step S302 to continue optimization.

[0111] S4. Use the optimized , and to configure the LSTM network. Divide the IMF components decomposed in step S2 into training set, validation set and test set according to a certain proportion. Use the training set to train the FATA-LSTM network model, use the test set to test the FATA-LSTM network model, and use the validation set to validate the FATA-LSTM network model to obtain the wind speed prediction model;

[0112] S5. Input the real-time collected wind speed data into the optimized prediction model and output the prediction result.

[0113] Such asFigure 2 As shown in the figure, it is a structural block diagram of a medium- and long-term wind speed prediction system provided by an embodiment of the present invention, including:

[0114] An acquisition module 100, configured to acquire wind speed data and construct an input-output data set;

[0115] An IMF component acquisition module 200, configured to optimize VMD using the artemisinin algorithm to find the optimal number of intrinsic mode components K and the penalty coefficient α, and acquire IMF components;

[0116] A FATA-LSTM network model construction module 300, configured to construct an LSTM network model, optimize the number of hidden units, the maximum training cycle, and the initial learning rate of the LSTM network model as the objectives of the mirage optimization algorithm, retain the optimal parameters of the LSTM network model, and obtain a FATA-LSTM network model;

[0117] A wind speed prediction model generation module 400, configured to train, test, and validate the FATA-LSTM network model using IMF components to generate a wind speed prediction model;

[0118] An output module 500, configured to output a prediction result.

[0119] Performance verification:

[0120] Taking the wind speed data of Changchun City from April 1 to June 30, 2023 as verification data, the original data is as Figure 3 shown. Using the wind speed prediction model provided by the embodiment of the present invention and various network models such as LSTM for prediction, the results are as Figure 4 shown. The prediction result data of each model is shown in Table 1:

[0121] Table 1

[0122]

[0123] It can be seen that using the model provided by the embodiment of the present invention for prediction, the root mean square error is 0.23145, the mean absolute error is 0.18501, and the mean absolute percentage error is 3.4809%, which are all better than the single LSTM, VMD-LSTM, and AO-VMD-LSTM models.

[0124] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A medium- and long-term wind speed prediction method, characterized in that, It includes the following steps: Collect wind speed data and construct an input-output data set; Use the artemisinin optimization algorithm to find the optimal number of intrinsic mode functions K and the penalty coefficient α in VMD, and obtain the IMF components; Construct an LSTM network model, optimize the number of hidden units, the maximum number of training epochs, and the initial learning rate of the LSTM network model as the objectives of the mirage optimization algorithm, retain the optimal parameters of the LSTM network model, and obtain the FATA-LSTM network model; Use the IMF components to train, test, and validate the FATA-LSTM network model to generate a wind speed prediction model; Output the prediction results.

2. The medium- and long-term wind speed prediction method according to claim 1, characterized in that The step of collecting wind speed data and constructing an input-output data set specifically includes: Obtain hourly wind speed records, denoted as the original signal data, with a length of T; Use the sliding window method to construct a data set. Use the wind speed data for 24 consecutive hours as the input feature x to predict the wind speed value y for the next hour, and generate a feature matrix x and a target vector y with a sample number of T - 24.

3. The medium- and long-term wind speed prediction method according to claim 1, characterized in that The step of using the artemisinin optimization algorithm to find the optimal number of intrinsic mode functions K and the penalty coefficient α in VMD and obtain the IMF components specifically includes: Define the range of the joint optimization parameter set of the core parameters of VMD decomposition: where the range of K is from 2 to 30, and the range of α is from 500 to 8000, which respectively represent the preset lower limit to upper limit of the number of decomposition modes and the preset lower limit to upper limit of the bandwidth constraint; Initialize the artemisinin population using the good point set strategy, where each artemisinin agent represents a set of VMD parameter combinations ; Execute the comprehensive elimination stage, simulate the comprehensive elimination effect in the initial stage of artemisinin treatment of malaria, and search the VMD parameter space through the global perturbation strategy; Execute the local clearance stage, simulate the local clearance effect of artemisinin in a specific infected area, and optimize the VMD parameters through the local adjustment strategy; Execute the late consolidation stage, simulate the residual cleaning and review effect of artemisinin in the final stage, and optimize the VMD parameters through the mutation and boundary processing mechanism; Judge whether the current evaluation quantity exceeds the maximum evaluation quantity. If it exceeds, use the current global optimal position as the final output to obtain the optimized VMD parameter combination. Otherwise, continue the optimization.

4. The medium- and long-term wind speed prediction method according to claim 3, wherein The step of initializing the artemisinin population using the good point set strategy, where each artemisinin agent represents a set of VMD parameter combinations is specifically as follows: Determine the population size N = 2, the problem dimension dim = 2, and the maximum number of iterations MaxFEs = 20; Within the parameter range Find the smallest prime number greater than 2*dim + 3 As the benchmark for generating the good point set, randomly generate the initial population pop, expressed as: ； ； ； In the formula, ; ; .

5. The medium- and long-term wind speed prediction method according to claim 4, characterized in that The step of executing the comprehensive elimination stage, simulating the comprehensive elimination effect in the initial stage of artemisinin treatment of malaria, and searching the VMD parameter space through the global perturbation strategy specifically includes: Initialize the artemisinin population using the good point set strategy and calculate the fitness of each agent: , and the objective function is the reconstruction error of VMD decomposition: ； In the formula, is the original wind speed signal; is the k-th IMF component; Obtain the current optimal agent best and the optimal fitness Bestfitness, corresponding to the current optimal VMD parameter combination ; Enter the iterative loop, calculate the dynamic control parameters, and update the dynamic control parameters according to the current function evaluation times FEs and the maximum evaluation times MaxFEs and the perturbation amplitude control parameter : ； ； Traverse each artemisinin agent i and dimension j, judge and update the agent position. If the random number , perform global perturbation update: ； 。 6. The medium- and long-term wind speed prediction method according to claim 5, characterized in that, The step of executing the local clearance stage, simulating the local clearance effect of artemisinin in a specific infected area, and optimizing the VMD parameters through the local adjustment strategy specifically includes: Calculate the normalized fitness: ； If , perform local perturbation update: ； Where A is a random permutation of the population indices; , and rand is a random number between 0 and 1.

7. The medium- and long-term wind speed prediction method according to claim 6, wherein The step of executing the late consolidation stage, simulating the effect of artemisinin in the final stage, and optimizing the VMD parameters through the mutation and fusion reflection and adaptive boundary processing mechanism specifically includes: Complete the update of the mutation position: ； In the formula, is the original position of the population; is the current global optimal solution; Check whether the population exceeds the boundary: ； If not satisfied, perform the following operations: Set a random number to randomly determine whether to execute the reflection strategy or the adaptive reset strategy: Reflection strategy: If it exceeds the upper bound: ； If it exceeds the lower bound: ； Perform the clamping operation, and the formula is: ； Adaptive reset strategy: 。 8. The medium- and long-term wind speed prediction method according to claim 1, characterized in that, The steps of constructing the LSTM network model, taking the number of hidden units, the maximum number of training epochs, and the initial learning rate of the LSTM network model as the objectives of the mirage optimization algorithm for optimization, and retaining the optimal parameters of the LSTM network model to obtain the FATA-LSTM network model are specifically as follows: Taking the number of hidden units, the maximum number of training epochs, and the initial learning rate of the LSTM network model as the light individuals in the mirage optimization algorithm for population initialization, specifically: Determine the population size N = 10 and the maximum number of evaluations MaxFEs = 20. Randomly generate the initial population within the parameter range , , : ； In the formula, represents the position of the i-th light individual in the j-th dimension; and are the upper and lower bounds of two points in the j-th dimension respectively; r is a random number between (0, 1); i = 1, 2,...., N; Simulating the refraction stage, simulating the optical refraction process of the mirage, which is divided into the direct refraction mode and the indirect refraction mode, and optimizing the LSTM hyperparameters through light propagation; Simulating the reflection stage, simulating the total internal reflection process of the mirage, and adjusting the LSTM hyperparameters through light reflection; Judging whether the current number of evaluation times FEs exceeds the maximum number of evaluation times MaxFEs. If it exceeds, determining the current global optimal position as the best hyperparameter combination of the LSTM network model, otherwise continuing to optimize.

9. The medium- and long-term wind speed prediction method according to claim 8, wherein The steps of the above-mentioned simulating the refraction stage, simulating the optical refraction process of the mirage, which is divided into the direct refraction mode and the indirect refraction mode, and optimizing the LSTM hyperparameters through light propagation are specifically as follows: Simulate the direct refraction stage and initialize , calculate the fitness of each ray individual: , and the objective function is the mean square error of LSTM on the test set: ； In the formula, is the data of the original data test set; is the data predicted by the model; Obtain the current global optimal position gBest and the optimal fitness gBestScore, and sort the fitness values. , and calculate the population quality factor: ； Where, Expressed as the sum of the cumulative integrals of the current population; The worst points value in history; The best points value in history; is the worst integral value; is a very small positive number; Comparing the random number r1 with IP. If r1 > IP, entering the direct refraction mode, and updating the position as: ； where rand is a random number between (0, 1), and if the fitness of the new position is better than the current one, update ; Simulating the indirect refraction mode. If r1 ≤ IP, calculating the individual quality factor p: ； Comparing the random number r2 with p. If r2 < p, entering the indirect refraction mode, and updating the position as: ； Wherein, , ; If the fitness of the new position is better than the current one, update .

10. The medium- and long-term wind speed prediction method according to claim 9, wherein The steps of the above-mentioned simulating the reflection stage, simulating the total internal reflection process of the mirage, and adjusting the LSTM hyperparameters through light reflection are specifically as follows: If r1 ≤ IP and r2 ≥ p, entering the reflection stage, and updating the position as: ； ； In the formula, is the individual index of the random reference ray; ; ; is the reflection coefficient; If the fitness of the new position is better than the current one, update .

11. A medium- and long-term wind speed prediction system for implementing the medium- and long-term wind speed prediction method according to any one of claims 1-10, characterized in that, Including: An acquisition module, used for acquiring wind speed data and constructing an input-output data set; An IMF component acquisition module, used for using the artemisinin optimization algorithm to find the optimal number of intrinsic mode functions K and the penalty coefficient α in VMD, and obtaining the IMF components; An FATA-LSTM network model construction module, used for constructing an LSTM network model, taking the number of hidden units, the maximum number of training epochs, and the initial learning rate of the LSTM network model as the objectives of the mirage optimization algorithm for optimization, and retaining the optimal parameters of the LSTM network model to obtain the FATA-LSTM network model; A wind speed prediction model generation module, used for training, testing, and validating the FATA-LSTM network model using the IMF components, and generating a wind speed prediction model; An output module, used for outputting the prediction result.

Images