Wind speed prediction method based on time sequence extension and related equipment

Abstract

The invention relates to the technical field of wind speed prediction, and provides a wind speed prediction method based on time sequence extension and related equipment, and the method comprises the steps: determining a plurality of target meteorological features from all meteorological features based on all feature values; based on all the feature values of all the target meteorological features, performing time sequence extension on all the wind speeds by using the proxy model to obtain an initial predicted wind speed; optimizing the proxy model by using an improved white whale optimization algorithm to obtain an optimized proxy model; integrating the wind speeds at the T moments and the initial predicted wind speed into a wind speed sequence, and decomposing and reconstructing the wind speed sequence to obtain a reconstructed sequence comprising a plurality of reconstructed wind speeds; and removing the reconstructed wind speed at the next moment corresponding to the current moment in the reconstructed sequence to obtain a final wind speed sequence, and performing wind speed prediction on the final wind speed sequence by using the optimized proxy model to obtain a final predicted wind speed. According to the method, the reliability of wind speed prediction can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of wind speed prediction, and in particular to a wind speed prediction method based on time series extension and related equipment. Background Art

[0002] Bridges in mountainous areas need to take into account changes in wind speed, especially special wind conditions such as valley winds and cliff winds. This not only affects the design of the bridge, but also poses new challenges to safety during construction. Accurate wind speed prediction can not only ensure the safety of bridge design and construction, but also provide reliable evaluation indicators for the long-term operation of the bridge.

[0003] At present, the commonly used wind speed prediction methods can be divided into three categories: physical methods, statistical methods and machine learning methods. Machine learning is a science of artificial intelligence, and its main research content is how to improve the performance of specific algorithms in experience learning. A large number of studies have shown that the wind speed prediction method based on machine learning is higher in accuracy and efficiency than the physical method and statistical method, and is an ideal solution to the wind speed prediction problem. At the same time, the wind speed sequence often contains a lot of noise, which is very unfavorable for wind speed prediction. To solve this problem, the wind speed sequence is denoised based on various decomposition algorithms, and the architecture of nesting the decomposition algorithm and machine learning is widely used. However, in recent years, studies have shown that decomposing the test set together during the data decomposition process may lead to prediction information leakage, which greatly improves the prediction accuracy. In fact, when a shorter data is inserted at the end of the sequence, the sub-modes of its decomposition will be significantly different. When the test set, which originally belongs to unknown data, is inserted into the training set and the validation set and then decomposed together, the accuracy of the model will be overestimated, which means that the model will be difficult to use for other data sets. It can be seen that the existing wind speed prediction methods have the problem of poor reliability of wind speed prediction. Summary of the invention

[0004] The present application provides a wind speed prediction method based on time series extension and related equipment, which can solve the problem of poor reliability of wind speed prediction.

[0005] In a first aspect, the present application provides a wind speed prediction method based on time series extension, the wind speed prediction method comprising:

[0006] Obtain the characteristic value and wind speed of each meteorological feature affecting the wind speed in the target area at T moments, and determine multiple target meteorological features from all meteorological features based on all characteristic values; the Tth moment is the current moment;

[0007] Based on all characteristic values ​​of all target meteorological characteristics, the proxy model is used to extend the time series of all wind speeds to obtain the initial predicted wind speed at the next moment of the current moment;

[0008] The proxy model is optimized using the improved White Whale optimization algorithm to obtain the optimized proxy model;

[0009] The wind speeds at T moments and the initial predicted wind speed are integrated into a wind speed sequence, and the wind speed sequence is decomposed and reconstructed to obtain a reconstructed sequence including multiple reconstructed wind speeds;

[0010] The reconstructed wind speed at the next moment corresponding to the current moment in the reconstructed sequence is removed to obtain the final wind speed sequence, and the optimized proxy model is used to predict the wind speed of the final wind speed sequence to obtain the final predicted wind speed of the target area at the next moment of the current moment.

[0011] Optionally, multiple target meteorological features are determined from all meteorological features based on all feature values, including:

[0012] For each meteorological feature, the variance of the meteorological feature is calculated according to all the characteristic values ​​corresponding to the meteorological feature, and it is determined whether the variance is greater than the variance threshold. If so, the meteorological feature is used as a candidate meteorological feature;

[0013] For each candidate meteorological feature, the mutual information value between the candidate meteorological feature and the wind speed is calculated, and it is determined whether the mutual information value is greater than the mutual information threshold. If so, the candidate meteorological feature is used as a target meteorological feature.

[0014] Optionally, the proxy model is optimized using an improved White Whale optimization algorithm to obtain an optimized proxy model, including:

[0015] An initial beluga whale population is created, and the initial beluga whale population is mapped using chaotic mapping to obtain a beluga whale population in multiple dimensions; the initial beluga whale population includes multiple initial beluga whale individuals, and each initial beluga whale individual is a set of hyperparameters of the proxy model;

[0016] The number of iterations is increased by 1, and the adaptive step size and balance factor are calculated based on the number of iterations;

[0017] The beluga population in each dimension is updated according to the balance factor and the adaptive step size, so as to obtain a new beluga population in each dimension including a plurality of new beluga individuals;

[0018] Calculate activation parameters based on the adaptive step size, and calculate activation gradients based on the activation parameters;

[0019] Determine whether the activation gradient is less than 0;

[0020] If so, all new beluga populations in each dimension are updated using a hybrid strategy to obtain the final beluga population in each dimension; the final beluga population includes multiple final beluga individuals;

[0021] Determine whether the number of iterations is greater than or equal to the preset number of iterations;

[0022] If yes, a final beluga individual is selected from all final beluga individuals as the target beluga individual, and the target beluga individual is used to optimize the hyperparameters in the proxy model to obtain the optimized proxy model;

[0023] Otherwise, take the final white whale population in each dimension as the white whale population in each dimension and return the number of iterations plus 1, and calculate the adaptive step size and balance factor according to the number of iterations.

[0024] Optionally, the initial beluga population is mapped using chaotic mapping to obtain beluga populations in multiple dimensions, including:

[0025] By formula:

[0026] X n+1,j =RX n,j (1-X n,j )

[0027] Calculate the jth beluga individual X in the n+1th dimension n+1,j ;

[0028] Among them, R represents the random chaos parameter, X n,j represents the jth beluga individual in the nth dimension, n = 1, 2, ..., N, N represents the number of dimensions, j = 1, 2, ..., J, J represents the number of beluga individuals in the beluga population, when n = 1, X 1,j Represents the jth initial beluga whale individual in the initial beluga whale population.

[0029] Optionally, calculate the adaptive step size and balance factor based on the number of iterations, including:

[0030] By formula:

[0031]

[0032] Calculate the adaptive step size a(t) for the tth iteration;

[0033] Where a(t-1) represents the adaptive step size of the t-1th iteration, t = 1, 2, ..., T max , T max represents the preset number of iterations, ε represents a constant, represents the first-order moment of the t-th iteration, Represents the second-order moment of the t-th iteration:

[0034]

[0035]

[0036] m t =β1m t-1+(1-β1)g t

[0037] V t =β2V t-1 +(1-β2)g t 2

[0038] Among them, m t represents the initial first-order moment of the βth iteration, V t represents the initial second-order moment of the tth iteration, β1 and β2 are constant parameters, g t Represents the gradient of fitness decrease, m t-1 represents the initial first-order moment of the t-1th iteration, V t-1 represents the initial second-order moment of the t-1th iteration;

[0039] By formula:

[0040]

[0041] Calculate the balance factor Bf;

[0042] Among them, B0 represents a random number.

[0043] Optionally, the beluga population in each dimension is updated according to the balance factor and the adaptive step size to obtain a new beluga population in each dimension including multiple new beluga individuals, including:

[0044] Determine whether the balance factor is greater than the preset balance factor;

[0045] If yes, the position update formula in the exploration phase is used to update the beluga whale individuals in each beluga whale population, and a new beluga whale population including multiple new beluga whale individuals in each dimension is obtained;

[0046] Otherwise, the position update formula in the development phase is used to update the beluga whale individuals in each beluga whale population to obtain a new beluga whale population including multiple new beluga whale individuals in each dimension.

[0047] Optionally, calculate activation parameters based on adaptive step size, including:

[0048] By formula:

[0049]

[0050] Calculate the activation parameter K;

[0051] Among them, A t represents the weight coefficient of the tth iteration, and Δf represents the fitness change rate:

[0052]

[0053]

[0054] Among them, A min represents the final control parameter, A max represents the initial control parameters, represents the highest fitness of the tth iteration, represents the highest fitness at the t-1th iteration.

[0055] Optionally, all new beluga populations in each dimension are updated using a hybrid strategy to obtain the final beluga population in each dimension, including:

[0056] By formula:

[0057]

[0058]

[0059] Calculate the jth final beluga individual in the nth dimension

[0060] in, represents the jth new beluga individual in the nth dimension, represents the speed of the j-th new beluga individual in the n-th dimension at the t+1-th iteration, represents the speed of the jth new beluga individual in the nth dimension at the tth iteration, w represents the inertia weight factor, c1 represents the individual learning factor, c2 represents the social learning factor, r5 and r6 are both random numbers, represents the optimal position that the j-th new beluga individual in the n-th dimension has searched in the t-th iteration, represents the optimal beluga individual in the tth iteration;

[0061] For each dimension, all the final beluga whale individuals corresponding to the dimension are integrated into one population to obtain the final beluga whale population of the dimension.

[0062] In a second aspect, the present application provides a wind speed prediction device based on time series extension, comprising:

[0063] A determination module obtains the characteristic value and wind speed of each meteorological feature affecting the wind speed in the target area at T moments, and determines multiple target meteorological features from all meteorological features based on all characteristic values; the Tth moment is the current moment;

[0064] The time extension module uses the proxy model to extend the time series of all wind speeds based on all characteristic values ​​of all target meteorological characteristics to obtain the initial predicted wind speed at the next moment of the current moment;

[0065] The optimization module optimizes the proxy model using the improved White Whale optimization algorithm to obtain the optimized proxy model;

[0066] An integration module integrates the wind speeds at T moments and the initial predicted wind speed into a wind speed sequence, and decomposes and reconstructs the wind speed sequence to obtain a reconstructed sequence including multiple reconstructed wind speeds;

[0067] The wind speed prediction module removes the reconstructed wind speed corresponding to the next moment of the current moment in the reconstructed sequence to obtain the final wind speed sequence, and uses the optimized proxy model to predict the wind speed of the final wind speed sequence to obtain the final predicted wind speed of the target area at the next moment of the current moment.

[0068] In a third aspect, an embodiment of the present application provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned wind speed prediction method based on time series extension when executing the above-mentioned computer program.

[0069] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, it implements the above-mentioned wind speed prediction method based on time series extension.

[0070] The above solution of the present application has the following beneficial effects:

[0071] In an embodiment of the present application, the characteristic value and wind speed of each meteorological characteristic affecting the wind speed in the target area at T moments are obtained, and multiple target meteorological characteristics are determined from all meteorological characteristics based on all characteristic values, and then all wind speeds are time-series extended using a proxy model based on all characteristic values ​​of all target meteorological characteristics to obtain an initial predicted wind speed at the next moment of the current moment, and then the proxy model is optimized using an improved Beluga optimization algorithm to obtain an optimized proxy model, and then the wind speeds at T moments and the initial predicted wind speeds are integrated into a wind speed sequence, and the wind speed sequence is decomposed and reconstructed to obtain a reconstructed sequence including multiple reconstructed wind speeds, and finally the reconstructed wind speed at the next moment corresponding to the current moment in the reconstructed sequence is removed to obtain a final wind speed sequence, and the optimized proxy model is used to predict the wind speed of the final wind speed sequence to obtain the final predicted wind speed of the target area at the next moment of the current moment. Among them, determining the target meteorological characteristics from all meteorological characteristics can reduce the amount of data input into the proxy model and improve the efficiency of wind speed prediction. Optimizing the proxy model with the improved Beluga optimization algorithm can effectively improve the performance of the proxy model so that the performance of the proxy model meets expectations, thereby improving the accuracy of wind speed prediction. Decomposing and reconstructing the wind speed sequence can reduce the noise in the wind speed data. Removing the reconstructed wind speed at the next moment corresponding to the current moment in the reconstructed sequence can suppress the boundary effect of the data sequence. Predicting the wind speed based on the data sequence after data removal can effectively improve the reliability of the wind speed prediction.

[0072] Other beneficial effects of the present application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0074] Figure 1 A flow chart of a wind speed prediction method based on time series extension provided in one embodiment of the present application;

[0075] Figure 2 A schematic diagram of a modal decomposition result provided by an embodiment of the present application;

[0076] Figure 3 A schematic diagram of a reconstruction sequence provided in an embodiment of the present application;

[0077] Figure 4 A schematic diagram showing a comparison between a predicted wind speed value and an actual wind speed value provided in an embodiment of the present application;

[0078] Figure 5 A schematic diagram of the structure of a wind speed prediction device based on time series extension provided in one embodiment of the present application;

[0079] Figure 6 A schematic diagram of the structure of a terminal device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0080] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0081] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0082] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0083] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0084] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0085] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0086] In response to the problem of poor reliability of existing wind speed predictions, an embodiment of the present application provides a wind speed prediction method based on time series extension. The wind speed prediction method obtains the characteristic value and wind speed of each meteorological feature that affects the wind speed in the target area at T moments, and determines multiple target meteorological features from all meteorological features based on all characteristic values, and then uses a proxy model to perform time series extension on all wind speeds based on all characteristic values ​​of all target meteorological features to obtain an initial predicted wind speed at the next moment of the current moment, and then uses an improved Beluga optimization algorithm to optimize the proxy model to obtain an optimized proxy model, and then integrates the wind speeds at T moments and the initial predicted wind speed into a wind speed sequence, and decomposes and reconstructs the wind speed sequence to obtain a reconstructed sequence including multiple reconstructed wind speeds, and finally removes the reconstructed wind speed at the next moment corresponding to the current moment in the reconstructed sequence to obtain a final wind speed sequence, and uses the optimized proxy model to perform wind speed prediction on the final wind speed sequence to obtain the final predicted wind speed of the target area at the next moment of the current moment. Among them, determining the target meteorological characteristics from all meteorological characteristics can reduce the amount of data input into the proxy model and improve the efficiency of wind speed prediction. Optimizing the proxy model with the improved Beluga optimization algorithm can effectively improve the performance of the proxy model so that the performance of the proxy model meets expectations, thereby improving the accuracy of wind speed prediction. Decomposing and reconstructing the wind speed sequence can reduce the noise in the wind speed data. Removing the reconstructed wind speed at the next moment corresponding to the current moment in the reconstructed sequence can suppress the boundary effect of the data sequence. Predicting the wind speed based on the data sequence after data removal can effectively improve the reliability of the wind speed prediction.

[0087] Next, an exemplary description is given of the wind speed prediction method based on time series extension provided in the present application.

[0088] like Figure 1 As shown, the wind speed prediction method based on time series extension provided by the present application includes the following steps:

[0089] Step 11, obtaining the characteristic value and wind speed of each meteorological feature affecting the wind speed in the target area at T moments, and determining a plurality of target meteorological features from all meteorological features based on all characteristic values.

[0090] The Tth moment is the current moment. For example, the current moment is 9 o'clock, T=5, then the 5 moments can be 7 o'clock, 7:30, 8 o'clock, 8:30, 9 o'clock, and the next moment of the current moment is 9:30. The above target area is the area where wind speed prediction is required, such as the bridge site. The meteorological characteristics can be temperature, rainfall, pressure difference, etc. After obtaining the original data of meteorological characteristics, the original data is normalized and standardized. The existing normalization formula can be used to normalize the original data. The formula Standardize the original data, x ‘ represents the normalized raw data, x represents the raw data, μ represents the mean of the raw data, and σ represents the standard deviation of the raw data. Then, the characteristic values ​​of meteorological characteristics are extracted from the normalized and standardized raw data. The characteristic values ​​of meteorological characteristics can be extracted using algorithms such as principal component analysis.

[0091] In some embodiments of the present application, the original data and wind speed of each meteorological feature affecting the wind speed in the target area at T times can be obtained by using a sensor or other equipment. The above step of determining multiple target meteorological features from all meteorological features based on all feature values ​​is specifically as follows:

[0092] In the first step, for each meteorological feature, the variance of the meteorological feature is calculated according to all the eigenvalues ​​corresponding to the meteorological feature, and it is determined whether the variance is greater than the variance threshold. If so, the meteorological feature is taken as a candidate meteorological feature.

[0093] Specifically, through the formula:

[0094]

[0095] Calculate the variance Var(X) of the meteorological characteristics.

[0096] Among them, x i represents the i-th eigenvalue corresponding to the meteorological feature, μ represents the mean of the eigenvalues ​​of the meteorological feature, and n represents the total number of eigenvalues ​​of the meteorological feature.

[0097] It should be noted that if the variance is less than or equal to the variance threshold, the meteorological feature corresponding to the variance is removed.

[0098] Exemplarily, the variance threshold is 0.8, and the variance of the first meteorological feature is 0.9, which is greater than the variance threshold. In this case, the meteorological feature is taken as a candidate meteorological feature.

[0099] In the second step, for each candidate meteorological feature, the mutual information value between the candidate meteorological feature and the wind speed is calculated, and it is determined whether the mutual information value is greater than the mutual information threshold. If so, the candidate meteorological feature is used as a target meteorological feature.

[0100] Specifically, through the formula:

[0101]

[0102] Calculate the mutual information value I(X;Y) between the candidate meteorological features and the wind speed.

[0103] Wherein, x represents a specific observation value of the candidate meteorological feature, X represents a set of specific observation values ​​of the candidate meteorological feature, y represents a specific observation value of wind speed, Y represents a set of specific observation values ​​of wind speed, p(x,y) represents the joint probability distribution between x and y, p(x) represents the marginal probability distribution of x, and p(y) represents the marginal probability distribution of y.

[0104] It should be noted that if the mutual information value is less than or equal to the mutual information threshold, the candidate meteorological feature corresponding to the mutual information value is removed.

[0105] Exemplarily, the mutual information value between the candidate meteorological feature and the wind speed is 1.6, which is less than the mutual information threshold of 1.9, and the candidate meteorological feature is removed.

[0106] In some embodiments of the present application, the correlation between the meteorological characteristics and the wind speed may also be analyzed by calculating the Pearson correlation coefficient between the meteorological characteristics and the wind speed. The calculation formula of the Pearson correlation coefficient is:

[0107]

[0108] Among them, r represents the value of Pearson correlation coefficient, X i represents the i-th specific observation value of the meteorological characteristic, represents the average value of specific observations of meteorological characteristics, Y i represents the i-th specific observation value of wind speed, represents the average value of the specific observations of wind speed, and n represents the total sample size.

[0109] It is worth mentioning that by calculating the variance and mutual information value of meteorological characteristics, the correlation between meteorological characteristics and wind speed can be analyzed, and only multiple meteorological characteristics with high correlation with wind speed can be retained, thereby reducing the amount of data and improving the efficiency of wind speed prediction.

[0110] Step 12, based on all characteristic values ​​of all target meteorological characteristics, all wind speeds are time-series extended using a proxy model to obtain an initial predicted wind speed at the next moment of the current moment.

[0111] In some embodiments of the present application, the proxy model may be a combination of a gated recurrent unit (GRU) and an attention mechanism, where the characteristic values ​​of all target meteorological features and all wind speeds are integrated into a matrix, input into the GRU, and then the attention mechanism is used to calculate the output data of the GRU to obtain the initial predicted wind speed at the next moment of the current moment. Specifically, the calculation formula of the GRU is:

[0112]

[0113]

[0114]

[0115]

[0116] Among them, z (t) represents the tth update gate, x (t) represents the input matrix, h (t-1) represents the t-1th hidden vector, r (t) represents the tth reset gate, g (t) represents the potential hidden vector, h (t) represents the tth hidden vector, W xz , W hz , W xr , W hr , W xg , W hg Both represent weights, b z 、b r 、b g Both represent bias terms.

[0117] The calculation formula of the attention mechanism is:

[0118] Vatq=atten(Vecq)

[0119]

[0120] χ=ξ T Vec

[0121] Where Vec represents the input data of the attention mechanism, Vat represents the importance of each element in Vec, q is the index value of the element in Vec, ξ represents the attention weight vector, χ represents the output of the attention mechanism, ( ) T Represents a transpose operation.

[0122] Exemplarily, before performing this step, the proxy model may be pre-trained using a loss function, where the loss function is:

[0123]

[0124] Among them, Loss represents the value of the loss function, n represents the number of wind speed samples, and y i represents the i-th wind speed sample, represents the i-th predicted wind speed, β j represents the regression coefficient, λ represents the regularization parameter, and p represents the number of features.

[0125] Step 13, optimizing the proxy model using the improved White Whale optimization algorithm to obtain an optimized proxy model.

[0126] In some embodiments of the present application, the above-mentioned step of optimizing the proxy model using the improved White Whale optimization algorithm to obtain the optimized proxy model is specifically as follows:

[0127] The first step is to create an initial beluga whale population and use chaotic mapping to map the initial beluga whale population to obtain beluga whale populations in multiple dimensions.

[0128] The above-mentioned initial beluga population includes multiple initial beluga individuals, each of which is a set of hyperparameters of the proxy model (such as learning rate, hidden layer size, batch size and other hyperparameters).

[0129] Exemplarily, a set of hyperparameters of the proxy model is randomly generated to obtain an initial beluga whale individual, and all the initial beluga whale individuals are integrated into a population to obtain an initial beluga whale population.

[0130] In some embodiments of the present application, the above-mentioned step of mapping the initial beluga whale population using chaotic mapping to obtain beluga whale populations in multiple dimensions is specifically as follows:

[0131] By formula:

[0132] X n+1,j =RX n,j (1-X n,j )

[0133] Calculate the jth beluga individual X in the n+1th dimension n+1,j .

[0134] Among them, R represents the random chaos parameter, X n,j represents the jth beluga individual in the nth dimension, n = 1, 2, ..., N, N represents the number of dimensions, j = 1, 2, ..., J, J represents the number of beluga individuals in the beluga population, when n = 1, X 1,j Represents the jth initial beluga whale individual in the initial beluga whale population.

[0135] In the second step, the number of iterations is increased by 1, and the adaptive step size and balance factor are calculated based on the number of iterations.

[0136] It should be noted that the initial number of iterations is 0.

[0137] Specifically, through the formula:

[0138]

[0139] Compute the adaptive step size a(t) for the tth iteration.

[0140] Where a(t-1) represents the adaptive step size of the t-1th iteration, t = 1, 2, ..., T max , T max represents the preset number of iterations, ε represents a constant, represents the first-order moment of the t-th iteration, Represents the second-order moment of the t-th iteration:

[0141]

[0142]

[0143] m t =β1m t-1 +(1-β1)g t

[0144] V t =β2V t-1 +(1-β2)g t 2

[0145] Among them, m t represents the initial first-order moment of the tth iteration, V t represents the initial second-order moment of the tth iteration, β1 and β2 are constant parameters, g t Represents the gradient of fitness decrease, m t-1 represents the initial first-order moment of the t-1th iteration, V t-1 Represents the initial second-order moment of the t-1th iteration. When t=1, m0 and V0 are preset values.

[0146] By formula:

[0147]

[0148] Calculate the balance factor Bf.

[0149] Among them, B0 represents a random number.

[0150] The third step is to update the beluga whale population in each dimension according to the balance factor and the adaptive step size to obtain a new beluga whale population in each dimension including multiple new beluga whale individuals.

[0151] Specifically, it is determined whether the balance factor is greater than a preset balance factor.

[0152] If so, the position update formula in the exploration phase is used to update the beluga whale individuals in each beluga whale population to obtain a new beluga whale population including multiple new beluga whale individuals in each dimension.

[0153] Otherwise, the position update formula in the development phase is used to update the beluga whale individuals in each beluga whale population to obtain a new beluga whale population including multiple new beluga whale individuals in each dimension.

[0154] It should be noted that the position update formula in the above exploration phase is:

[0155]

[0156] in, represents the jth beluga individual in the nth dimension after update, represents the jth beluga individual in the pnth dimension, represents the rth beluga whale individual in the p1th dimension, r1 and r2 are both random numbers in the range of (0,1), j≠r, j,r∈J, J represents the number of beluga whale individuals in the beluga whale population, pn≠n≠p1, pn,n,p1∈N, N represents the number of dimensions.

[0157] The position update formula in the above development phase is:

[0158]

[0159]

[0160] in, represents the jth beluga individual in the nth dimension after the update. r3 and r4 are random numbers used to control the amplitude and direction of the update. represents the optimal beluga individual of the tth iteration, represents the jth beluga individual in the nth dimension, represents the rth beluga individual in the nth dimension, LF represents the Levy flight function, C1 represents the random jump intensity, v and μ are random numbers from the normal distribution, β represents a constant, and σ represents the standard deviation.

[0161] In the fourth step, the activation parameters are calculated based on the adaptive step size, and the activation gradient is calculated according to the activation parameters.

[0162] Specifically, through the formula:

[0163]

[0164] Calculate the activation parameter K.

[0165] Among them, At represents the weight coefficient of the tth iteration, and Δf represents the fitness change rate:

[0166]

[0167]

[0168] Among them, A min represents the final control parameter, A max represents the initial control parameters, represents the highest fitness of the tth iteration, represents the highest fitness at the t-1th iteration.

[0169] The activation gradient is obtained by taking the derivative of the activation parameter with respect to the number of iterations t.

[0170] It should be noted that the fitness is the fitness of the new Beluga individual, which can be calculated by the fitness function in the Beluga optimization algorithm.

[0171] The fifth step is to determine whether the activation gradient is less than 0.

[0172] If yes, then all new beluga populations in each dimension are updated using the hybrid strategy to obtain the final beluga population in each dimension. Otherwise, all new beluga populations are not updated using the hybrid strategy.

[0173] The final beluga population includes multiple final beluga individuals.

[0174] Specifically, through the formula:

[0175]

[0176]

[0177] Calculate the jth final beluga individual in the nth dimension

[0178] in, represents the jth new beluga individual in the nth dimension, represents the speed of the j-th new beluga individual in the n-th dimension at the t+1-th iteration, represents the speed of the jth new beluga individual in the nth dimension at the tth iteration, w represents the inertia weight factor, c1 represents the individual learning factor, c2 represents the social learning factor, r5 and r6 are both random numbers, represents the optimal position that the j-th new beluga individual in the n-th dimension has searched in the t-th iteration, represents the optimal beluga individual in the tth iteration.

[0179] For each dimension, all the final beluga whale individuals corresponding to the dimension are integrated into one population to obtain the final beluga whale population of the dimension.

[0180] Step 6: Determine whether the number of iterations is greater than or equal to the preset number of iterations.

[0181] If so, a final beluga individual is selected from all final beluga individuals as the target beluga individual, and the target beluga individual is used to optimize the hyperparameters in the proxy model to obtain the optimized proxy model.

[0182] Otherwise, take the final white whale population in each dimension as the white whale population in each dimension and return the number of iterations plus 1, and calculate the adaptive step size and balance factor according to the number of iterations.

[0183] It should be noted that when selecting the target beluga whale individuals from all the final beluga whale individuals, random selection is performed. In order to ensure that the number of beluga whale individuals in the beluga whale population remains unchanged in each iteration, before returning the number of iterations plus 1 and calculating the adaptive step size and balance factor according to the number of iterations, it is necessary to simulate the whale fall behavior and lose some individuals. The expression is:

[0184]

[0185] in, represents the jth final beluga individual in the nth dimension after updating the simulated whale fall behavior. represents the jth final beluga individual in the nth dimension, X step Indicates the step length of the whale's descent. r7, r8, and r9 are all random numbers.

[0186] Exemplarily, the number of iterations is 9, and the preset number of iterations is 10. If the number of iterations does not reach the preset number of iterations, the number of iterations plus 1 is returned, and the steps of calculating the adaptive step size and the balance factor are calculated according to the number of iterations. At this time, the number of iterations is 10, which reaches the preset number of iterations. A final beluga individual is randomly selected from all the final beluga individuals at this time as the target beluga individual, and the hyperparameters in the target beluga individual are substituted into the proxy model to obtain the optimized proxy model.

[0187] It is worth mentioning that optimizing the proxy model using the improved Beluga optimization algorithm can effectively improve the performance of the proxy model, so that the performance of the proxy model meets expectations, thereby improving the accuracy of wind speed prediction.

[0188] Step 14: Integrate the wind speeds at T moments and the initial predicted wind speed into a wind speed sequence, and decompose and reconstruct the wind speed sequence to obtain a reconstructed sequence including a plurality of reconstructed wind speeds.

[0189] The data in the above wind speed sequence and the reconstructed sequence are the same as the total number of moments. If there are T+1 moments including the next moment of the current moment, the reconstructed sequence includes T+1 data.

[0190] In some embodiments of the present application, the wind speed sequence can be decomposed and reconstructed using a singular spectrum analysis method. Specifically, the wind speed sequence is first written as a trajectory equation:

[0191]

[0192] Among them, t1 represents the first wind speed, t2 represents the second wind speed, and t l represents the lth wind speed, t l+1 represents the l+1th wind speed, t m-l+1 represents the m-l+1th wind speed, t m-l+2 Indicates the m-l+2th wind speed, t m Indicates the mth wind speed.

[0193] Perform singular value decomposition of F into the following form:

[0194]

[0195] Among them, l×(m-l+1) represents the dimension of F, K l×l represents the left singular vector matrix, V represents the right singular vector matrix, i represents the singular value number, and (m-l+1)×(m-l+1) represents the dimension of V.

[0196] By the formula P = F × F T Calculate the covariance matrix P and calculate multiple eigenvalues ​​of P [λ1,λ2,...,λ l ], i∈l, then is the i-th singular value, and the corresponding eigenvector is Therefore, there are:

[0197]

[0198]

[0199] After the data decomposition is completed, the first M modes are selected according to the energy percentage to reconstruct the input time series. The specific calculation formula is as follows:

[0200]

[0201] Where E(k) represents the energy ratio of the first k modes, W k (i) represents the i-th wind speed in the reconstructed time series of the first k-order modes, and W(i) represents the i-th wind speed in the wind speed sequence.

[0202] Then, the first k-order modes when the energy ratio is greater than a preset energy ratio (such as 95%) are selected as the first M-order modes, and the elements at corresponding positions in the reconstruction time series corresponding to the first M-order modes are added to obtain a reconstructed sequence.

[0203] It is worth mentioning that wind speed data usually contains a lot of noise. Decomposing and reconstructing the wind speed sequence can reduce the noise in the wind speed data.

[0204] The above singular value decomposition is illustrated below with a specific example. The modal decomposition result is as follows: Figure 2 As shown in the figure, the horizontal axis represents time and the vertical axis represents wind speed value. Figure 2 Part a is the wind speed curve of submode 1 after decomposition, part b is the wind speed curve of submode 2, part c is the wind speed curve of submode 3, part d is the wind speed curve of submode 4, part e is the wind speed curve of submode 5, part f is the wind speed curve of submode 6, part g is the wind speed curve of submode 7, part h is the wind speed curve of submode 8, part i is the wind speed curve of submode 9, and part j is the wind speed curve of submode 10.

[0205] Reconstruct the sequence as Figure 3 As shown in the figure, the horizontal axis represents time, the vertical axis represents the wind speed value, the unit is meter per second (m / s), and the two curves are the wind speed curve of the wind speed sequence (original sequence) and the wind speed curve of the reconstructed sequence (SSA decomposition and reconstruction).

[0206] Step 15, remove the reconstructed wind speed corresponding to the next moment of the current moment in the reconstructed sequence to obtain a final wind speed sequence, and use the optimized proxy model to predict the wind speed of the final wind speed sequence to obtain the final predicted wind speed of the target area at the next moment of the current moment.

[0207] It should be noted that, as can be seen from step 14, the multiple reconstructed wind speeds in the reconstruction sequence correspond one-to-one to multiple moments. In this step, the reconstructed wind speed corresponding to the next moment of the current moment in the reconstruction sequence is removed, and the reconstructed wind speeds corresponding to other moments are retained to suppress the boundary effect.

[0208] In some embodiments of the present application, the final wind speed sequence and all eigenvalues ​​of all target meteorological characteristics in step 11 are integrated into a matrix, and the matrix is ​​input into the optimized proxy model for wind speed prediction to obtain the final predicted wind speed of the target area at the next moment after the current moment.

[0209] For example, in order to further improve the accuracy of the final predicted wind speed, the optimized proxy model can be used to perform two wind speed predictions and calculate the difference between the two final predicted wind speeds. If the difference is less than the preset difference, it means that the accuracy of the final predicted wind speed meets expectations. If the difference is greater than or equal to the preset difference, it means that the accuracy of the final predicted maple tree does not meet expectations and the proxy model needs to be further adjusted.

[0210] It is worth mentioning that determining the target meteorological characteristics from all meteorological characteristics can reduce the amount of data input into the proxy model and improve the efficiency of wind speed prediction. Optimizing the proxy model using the improved Beluga optimization algorithm can effectively improve the performance of the proxy model so that the performance of the proxy model meets expectations, thereby improving the accuracy of wind speed prediction. Decomposing and reconstructing the wind speed sequence can reduce the noise in the wind speed data. Removing the reconstructed wind speed at the next moment corresponding to the current moment in the reconstructed sequence can suppress the boundary effect of the data sequence. Predicting the wind speed based on the data sequence after data removal can effectively improve the reliability of the wind speed prediction.

[0211] The wind speed prediction method of the present application is exemplified below with reference to a specific example.

[0212] like Figure 4 As shown in the figure, the horizontal axis represents time, and the vertical axis represents the wind speed value, the unit is meter per second (m / s), and the two curves are the predicted values ​​obtained by using the method of the present application to predict the wind speed, and the corresponding true values ​​of the wind speed.

[0213] It can be seen that the wind speed prediction method of the present application can effectively improve the accuracy of wind speed prediction.

[0214] The following is an exemplary description of the wind speed prediction device based on time series extension provided in the present application.

[0215] like Figure 5 As shown, the embodiment of the present application provides a wind speed prediction device based on time series extension, and the wind speed prediction device based on time series extension 500 includes:

[0216] Determination module 501, obtains the characteristic value and wind speed of each meteorological feature affecting the wind speed in the target area at T moments, and determines multiple target meteorological features from all meteorological features based on all characteristic values; the Tth moment is the current moment;

[0217] The time series extension module 502 uses the proxy model to extend the time series of all wind speeds based on all characteristic values ​​of all target meteorological characteristics to obtain the initial predicted wind speed at the next moment of the current moment;

[0218] The optimization module 503 optimizes the proxy model using the improved White Whale optimization algorithm to obtain an optimized proxy model;

[0219] An integration module 504 integrates the wind speeds at T moments and the initial predicted wind speed into a wind speed sequence, and decomposes and reconstructs the wind speed sequence to obtain a reconstructed sequence including a plurality of reconstructed wind speeds;

[0220] The wind speed prediction module 505 removes the reconstructed wind speed corresponding to the next moment of the current moment in the reconstructed sequence to obtain a final wind speed sequence, and uses the optimized proxy model to perform wind speed prediction on the final wind speed sequence to obtain the final predicted wind speed of the target area at the next moment of the current moment.

[0221] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the present application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0222] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0223] like Figure 6 As shown, an embodiment of the present application provides a terminal device. The terminal device D10 of this embodiment includes: at least one processor D100 ( Figure 6 Only one processor is shown in the figure), a memory D101, and a computer program D102 stored in the memory D101 and executable on the at least one processor D100, wherein the processor D100 implements the steps of any of the above-mentioned method embodiments when executing the computer program D102.

[0224] Specifically, when the processor D100 executes the computer program D102, it obtains the characteristic value and wind speed of each meteorological feature that affects the wind speed in the target area at T moments, and determines multiple target meteorological features from all meteorological features based on all characteristic values, and then uses a proxy model to extend the time series of all wind speeds based on all characteristic values ​​of all target meteorological features to obtain the initial predicted wind speed at the next moment of the current moment, and then uses the improved Beluga optimization algorithm to optimize the proxy model to obtain an optimized proxy model, and then integrates the wind speeds at T moments and the initial predicted wind speed into a wind speed sequence, and decomposes and reconstructs the wind speed sequence to obtain a reconstructed sequence including multiple reconstructed wind speeds, and finally removes the reconstructed wind speed at the next moment corresponding to the current moment in the reconstructed sequence to obtain a final wind speed sequence, and uses the optimized proxy model to predict the wind speed of the final wind speed sequence to obtain the final predicted wind speed of the target area at the next moment of the current moment. Among them, determining the target meteorological characteristics from all meteorological characteristics can reduce the amount of data input into the proxy model and improve the efficiency of wind speed prediction. Optimizing the proxy model with the improved Beluga optimization algorithm can effectively improve the performance of the proxy model so that the performance of the proxy model meets expectations, thereby improving the accuracy of wind speed prediction. Decomposing and reconstructing the wind speed sequence can reduce the noise in the wind speed data. Removing the reconstructed wind speed at the next moment corresponding to the current moment in the reconstructed sequence can suppress the boundary effect of the data sequence. Predicting the wind speed based on the data sequence after data removal can effectively improve the reliability of the wind speed prediction.

[0225] The processor D100 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0226] In some embodiments, the memory D101 may be an internal storage unit of the terminal device D10, such as a hard disk or memory of the terminal device D10. In other embodiments, the memory D101 may also be an external storage device of the terminal device D10, such as a plug-in hard disk, a smart memory card (SMC, SmartMedia Card), a secure digital (SD, Secure Digital) card, a flash card (Flash Card), etc. equipped on the terminal device D10. Further, the memory D101 may also include both an internal storage unit of the terminal device D10 and an external storage device. The memory D101 is used to store an operating system, an application program, a boot loader (BootLoader), data and other programs, such as the program code of the computer program, etc. The memory D101 may also be used to temporarily store data that has been output or is to be output.

[0227] An embodiment of the present application further 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 in the above-mentioned method embodiments can be implemented.

[0228] An embodiment of the present application provides a computer program product. When the computer program product runs on a terminal device, the terminal device can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0229] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium that can carry the computer program code to the wind speed prediction method device / terminal device based on time extension. For example, a USB flash drive, a mobile hard disk, a disk or an optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0230] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0231] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0232] The above is a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles described in the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A wind speed prediction method based on time series extension, characterized in that: include: Obtaining the characteristic value and wind speed of each meteorological feature affecting the wind speed in the target area at T times, and determining a plurality of target meteorological features from all meteorological features based on all characteristic values; The Tth moment is the current moment; Based on all characteristic values ​​of all target meteorological characteristics, the proxy model is used to extend the time series of all wind speeds to obtain the initial predicted wind speed at the next moment of the current moment; The proxy model is optimized using an improved White Whale optimization algorithm to obtain an optimized proxy model; Integrating the wind speeds at the T moments and the initial predicted wind speed into a wind speed sequence, and decomposing and reconstructing the wind speed sequence to obtain a reconstructed sequence including a plurality of reconstructed wind speeds; The reconstructed wind speed at the next moment corresponding to the current moment in the reconstructed sequence is removed to obtain a final wind speed sequence, and the wind speed of the final wind speed sequence is predicted using the optimized proxy model to obtain a final predicted wind speed of the target area at the next moment of the current moment.

2. The wind speed prediction method according to claim 1, characterized in that: The step of determining a plurality of target meteorological features from all meteorological features based on all feature values ​​includes: For each of the meteorological features, respectively, calculate the variance of the meteorological feature according to all the feature values ​​corresponding to the meteorological feature, and determine whether the variance is greater than a variance threshold, and if so, use the meteorological feature as a candidate meteorological feature; For each of the candidate meteorological features, the mutual information value between the candidate meteorological feature and the wind speed is calculated, and it is determined whether the mutual information value is greater than a mutual information threshold. If so, the candidate meteorological feature is used as a target meteorological feature.

3. The wind speed prediction method according to claim 1, characterized in that: The improved White Whale optimization algorithm is used to optimize the proxy model to obtain an optimized proxy model, including: Creating an initial beluga whale population, and mapping the initial beluga whale population using chaotic mapping to obtain a beluga whale population in multiple dimensions; the initial beluga whale population includes multiple initial beluga whale individuals, and each initial beluga whale individual is a set of hyperparameters of the proxy model; The number of iterations is increased by 1, and the adaptive step size and balance factor are calculated based on the number of iterations; The beluga whale population in each dimension is updated according to the balance factor and the adaptive step size to obtain a new beluga whale population including a plurality of new beluga whale individuals in each dimension; Calculating an activation parameter based on the adaptive step size, and calculating an activation gradient according to the activation parameter; Determine whether the activation gradient is less than 0; If so, all new beluga whale populations in each dimension are updated using a hybrid strategy to obtain a final beluga whale population in each dimension; the final beluga whale population includes multiple final beluga whale individuals; Determine whether the number of iterations is greater than or equal to a preset number of iterations; If yes, a final beluga individual is selected from all final beluga individuals as a target beluga individual, and the target beluga individual is used to optimize the hyperparameters in the proxy model to obtain an optimized proxy model; Otherwise, the final beluga population of each dimension is used as the beluga population of each dimension, and the number of iterations plus 1 is returned, and the steps of calculating the adaptive step size and the balance factor according to the number of iterations.

4. The wind speed prediction method according to claim 3, characterized in that: The initial beluga whale population is mapped by chaotic mapping to obtain beluga whale populations in multiple dimensions, including: By formula: X n+1,j =RX n,j (1-X n,j ) Calculate the jth beluga individual X in the n+1th dimension n+1,j ; Among them, R represents the random chaos parameter, X n,j represents the jth beluga individual in the nth dimension, n = 1, 2, ..., N, N represents the number of dimensions, j = 1, 2, ..., J, J represents the number of beluga individuals in the beluga population, when n = 1, X 1,j Represents the jth initial beluga whale individual in the initial beluga whale population.

5. The wind speed prediction method according to claim 4, characterized in that: The step of calculating the adaptive step size and the balance factor according to the number of iterations includes: By formula: Calculate the adaptive step size a(t) for the tth iteration; Where a(t-1) represents the adaptive step size of the t-1th iteration, t = 1, 2, ..., T max , T max represents the preset number of iterations, ε represents a constant, represents the first-order moment of the t-th iteration, Represents the second-order moment of the t-th iteration: m t =β1m t-1 +(1-β1)g t V t =β2V t-1 +(1-β2)g t 2 Among them, m t represents the initial first-order moment of the tth iteration, V t represents the initial second-order moment of the tth iteration, β1 and β2 are constant parameters, g t Represents the gradient of fitness decrease, m t-1 represents the initial first-order moment of the t-1th iteration, V t-1 represents the initial second-order moment of the t-1th iteration; By formula: Calculate the balance factor Bf; Among them, B0 represents a random number.

6. The wind speed prediction method according to claim 5, characterized in that: The beluga population in each dimension is updated according to the balance factor and the adaptive step size to obtain a new beluga population including a plurality of new beluga individuals in each dimension, including: Determining whether the balance factor is greater than a preset balance factor; If yes, the position update formula of the exploration phase is used to update the beluga whale individuals in each beluga whale population to obtain a new beluga whale population including multiple new beluga whale individuals in each dimension; Otherwise, the position update formula in the development phase is used to update the beluga whale individuals in each beluga whale population to obtain a new beluga whale population including a plurality of new beluga whale individuals in each dimension.

7. The wind speed prediction method according to claim 6, characterized in that: The calculating of activation parameters based on the adaptive step size comprises: By formula: Calculate the activation parameter K; Among them, A t represents the weight coefficient of the tth iteration, and Δf represents the fitness change rate: Among them, A min represents the final control parameter, A max represents the initial control parameters, represents the highest fitness of the tth iteration, represents the highest fitness at the t-1th iteration.

8. The wind speed prediction method according to claim 1, characterized in that: The hybrid strategy is used to update all new beluga populations in each dimension to obtain the final beluga population in each dimension, including: By formula: Calculate the jth final beluga individual in the nth dimension in, represents the jth new beluga individual in the nth dimension, represents the speed of the j-th new beluga individual in the n-th dimension at the t+1-th iteration, represents the speed of the jth new beluga individual in the nth dimension at the tth iteration, w represents the inertia weight factor, c1 represents the individual learning factor, c2 represents the social learning factor, r5 and r6 are both random numbers, represents the optimal position that the j-th new beluga individual in the n-th dimension has searched in the t-th iteration, represents the optimal beluga individual in the tth iteration; For each dimension, all final beluga whale individuals corresponding to the dimension are integrated into one population to obtain the final beluga whale population of the dimension.

9. A terminal 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 wind speed prediction method based on time series extension as described in any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the wind speed prediction method based on time series extension according to any one of claims 1 to 8 is implemented.

Images