Model training method and device, and method for predicting wind power

Abstract

This application proposes a training method and device for a wind power prediction model, and a method for predicting wind power, relating to the field of artificial intelligence technology. The method includes: obtaining multiple sets of historical monitoring data and corresponding actual output power of sample wind turbines during operation; performing feature extraction on any set of historical monitoring data to obtain corresponding key features; and training a prediction model based on each key feature and corresponding actual output power to obtain a trained prediction model; wherein the prediction model is a support vector regression (SVR) model. Thus, the wind power prediction model can be trained, and the prediction effect of the model can be improved, that is, the accuracy and reliability of the model prediction results can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a training method and device for a wind power prediction model, and a method for predicting wind power. Background Art

[0002] With the increasing depletion of fossil energy and environmental pollution, the development of renewable energy is receiving increasing attention. Wind power, as a green, sustainable, and renewable energy source, is gaining increasing attention worldwide. However, wind power is intermittent and random, and its integration into the grid poses significant challenges to the security and stability of power systems. Accurate short-term wind power forecasts can provide a reliable scientific basis for grid scheduling, ensuring grid security and improving wind power utilization. Summary of the Invention

[0003] The present application aims to solve one of the technical problems in the related art at least to a certain extent.

[0004] This application proposes a training method and device for a wind power prediction model, and a method for predicting wind power, so as to realize the training of the wind power prediction model and improve the prediction effect of the model, that is, to improve the accuracy and reliability of the model prediction results.

[0005] The first embodiment of the present application provides a method for training a wind power prediction model, including:

[0006] Obtain multiple sets of historical monitoring data and corresponding actual output power of sample wind turbines during operation;

[0007] Perform feature extraction on any set of historical monitoring data to obtain the corresponding key features;

[0008] Based on each of the key features and the corresponding actual output power, a prediction model is trained to obtain a trained prediction model; wherein the prediction model is a support vector regression SVR model.

[0009] A second embodiment of the present application provides a method for predicting wind power, including:

[0010] Acquire a trained prediction model using the wind power prediction model training method provided in the embodiment of the first aspect;

[0011] Obtain target monitoring data of the target wind turbine during operation;

[0012] Performing feature extraction on the target monitoring data to obtain corresponding target features;

[0013] Based on the target characteristics, the trained prediction model is used to predict the output power of the target wind turbine generator set to obtain the target output power.

[0014] The third embodiment of the present application provides a training device for a wind power prediction model, comprising:

[0015] An acquisition module is used to obtain multiple sets of historical monitoring data and corresponding actual output power of sample wind turbines during operation;

[0016] The extraction module is used to extract features from any set of historical monitoring data to obtain the corresponding key features;

[0017] A training module is used to perform model training on the prediction model based on each of the key features and the corresponding actual output power to obtain a trained prediction model; wherein the prediction model is a support vector regression SVR model.

[0018] A fourth embodiment of the present application provides a device for predicting wind power, including:

[0019] A first acquisition module, configured to acquire a trained prediction model using the wind power prediction model training method provided in the embodiment of the first aspect;

[0020] The second acquisition module is used to obtain target monitoring data of the target wind turbine when it is in operation;

[0021] An extraction module, configured to extract features from the target monitoring data to obtain corresponding target features;

[0022] A prediction module is used to predict the output power of the target wind turbine generator system based on the target characteristics and using the trained prediction model to obtain the target output power.

[0023] The fifth aspect embodiment of the present application proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the program, it implements the training method of the wind power prediction model proposed in the first aspect embodiment of the present application, or the method for predicting wind power proposed in the second aspect embodiment of the present application.

[0024] The sixth aspect embodiment of the present application proposes a non-temporary computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, it implements the training method of the wind power prediction model proposed in the first aspect embodiment of the present application, or the method for predicting wind power proposed in the second aspect embodiment of the present application.

[0025] The seventh aspect embodiment of the present application proposes a computer program product. When the instructions in the computer program product are executed by a processor, the training method of the wind power prediction model proposed in the first aspect embodiment of the present application, or the method for predicting wind power proposed in the second aspect embodiment of the present application is executed.

[0026] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0028] Figure 1 A flow chart of a method for training a wind power prediction model provided in Example 1 of the present application;

[0029] Figure 2 A flow chart of a method for training a wind power prediction model provided in Example 2 of the present application;

[0030] Figure 3 A flow chart of a method for predicting wind power provided in the third embodiment of the present application;

[0031] Figure 4 A schematic structural diagram of a wind power prediction model training device provided in the fourth embodiment of the present application;

[0032] Figure 5 A schematic structural diagram of a device for predicting wind power provided in Example 5 of the present application;

[0033] Figure 6 A schematic block diagram of an example electronic device that can be used to implement embodiments of the present application is shown. DETAILED DESCRIPTION

[0034] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0035] In related technologies, the prediction of wind power mainly adopts physical models, statistical models, machine learning and hybrid models.

[0036] Among them, the physical model combines numerical weather forecasts and basic parameters of wind farms, and calculates the predicted values ​​through a set of high-dimensional equations. However, the calculations of the physical model are complex and the cost of collecting information is high; the statistical model mainly works by establishing the relationship between input and output. However, the statistical model usually establishes a linear model and cannot handle the nonlinear nature of wind power series; with the development of artificial intelligence, machine learning with stronger nonlinear learning capabilities has been widely used in wind power prediction, mainly including support vector machines, extreme learning machines and artificial neural networks, etc.

[0037] Wind power prediction models can be categorized into single models and hybrid models based on their structure. Hybrid models typically achieve higher wind power prediction accuracy than single models. Currently, researchers have proposed a variety of hybrid models that combine the strengths of different models. While relevant literature has demonstrated the high stability and accuracy of machine learning hybrid models in predicting wind power, these models often fail to fully account for factors influencing wind power and under-optimize machine learning algorithms.

[0038] In response to at least one of the above problems, the present application proposes a training method for a wind power prediction model and a method for predicting wind power.

[0039] The following describes a method for training a wind power prediction model and a method for predicting wind power according to an embodiment of the present application with reference to the accompanying drawings.

[0040] Figure 1 This is a flow chart of the training method of the wind power prediction model provided in Example 1 of the present application.

[0041] In the embodiment of the present application, the training method of the wind power prediction model is configured in a training device for a wind power prediction model. The training device for a wind power prediction model can be applied to any electronic device so that the electronic device can perform the training function of the wind power prediction model.

[0042] Among them, the electronic device can be any device with computing capabilities, such as a personal computer (PC), a mobile terminal, a server, etc. The mobile terminal can be, for example, a mobile phone, a tablet computer, a personal digital assistant, a wearable device, and other hardware devices with various operating systems, touch screens and / or display screens.

[0043] like Figure 1 As shown, the training method of the wind power prediction model may include the following steps:

[0044] Step 101: Acquire multiple sets of historical monitoring data and corresponding actual output power of sample wind turbines during operation.

[0045] In a possible implementation of the embodiment of the present application, the historical monitoring data may include at least one of weather forecast data, unit operation data, and wind tower data.

[0046] Among them, weather forecast data can be but is not limited to temperature, humidity, precipitation, cloud cover, etc., and this application does not impose any restrictions on this.

[0047] Among them, the unit operation data may include generator speed, voltage, current, etc., which is not limited in this application.

[0048] The wind tower data may include wind speed, air pressure, wind direction, etc., which is not limited in this application.

[0049] In the embodiment of the present application, any set of historical monitoring data of the sample wind turbines may have corresponding actual output power.

[0050] In an embodiment of the present application, multiple sets of historical monitoring data and corresponding actual output power of the sample wind turbines during operation can be obtained.

[0051] Step 102: extract features from any set of historical monitoring data to obtain corresponding key features.

[0052] In an embodiment of the present application, for any set of historical monitoring data, feature extraction is performed on the historical monitoring data to obtain corresponding key features.

[0053] As a possible implementation method, based on multiple sets of historical monitoring data, a feature selection algorithm can be used to determine the target features; for any historical monitoring data, feature extraction is performed on the historical monitoring data according to the target features to obtain the corresponding key features.

[0054] In the embodiments of the present application, the feature selection algorithm may be, for example, the Pearson correlation coefficient method, the random forest algorithm, etc., and the present application does not impose any restrictions on this.

[0055] As an example, based on multiple sets of historical monitoring data, the Pearson correlation coefficient method can be used to select relevant features related to the actual output power, and the four features with the highest correlation with the actual output power can be selected as the target features of the model. The Pearson correlation coefficient method measures the linear correlation between the relevant features and the actual output power by taking the quotient of the covariance and standard deviation between the two features. In this way, the target features of multiple sets of historical monitoring data can be effectively determined; thus, for any historical monitoring data, feature extraction can be performed on the historical monitoring data based on the target features to obtain the corresponding key features.

[0056] Step 103 : Based on each key feature and the corresponding actual output power, the prediction model is trained to obtain a trained prediction model.

[0057] The prediction model may be a SVR (Support Vector Regression) model.

[0058] In an embodiment of the present application, the prediction model can be trained based on each key feature and the corresponding actual output power to obtain a trained prediction model, which can enable the prediction model to learn the correspondence between the key features of the wind turbine and the actual output power, thereby obtaining a trained prediction model.

[0059] The wind power prediction model training method of the present embodiment obtains multiple sets of historical monitoring data and the corresponding actual output power of sample wind turbines during operation; performs feature extraction on any set of historical monitoring data to obtain corresponding key features; and trains the prediction model based on each key feature and the corresponding actual output power to obtain a trained prediction model; wherein the prediction model is a support vector regression (SVR) model. This method can train the wind power prediction model and improve the model's prediction performance, that is, improve the accuracy and reliability of the model's prediction results.

[0060] In order to clearly illustrate how the prediction model is trained based on each key feature and the corresponding actual output power in the above embodiments of the present application to obtain a trained prediction model, the present application also proposes a training method for a wind power prediction model.

[0061] Figure 2 This is a flow chart of the training method of the wind power prediction model provided in the second embodiment of the present application.

[0062] like Figure 2 As shown, the training method of the wind power prediction model may include the following steps:

[0063] Step 201 : obtaining multiple sets of historical monitoring data and corresponding actual output power of sample wind turbines during operation.

[0064] Step 202: extract features from any set of historical monitoring data to obtain corresponding key features.

[0065] The execution process of step 201 to step 202 can refer to the execution process of any embodiment of the present application and will not be repeated here.

[0066] Step 203 : Based on the key features and the actual output power of the sample wind turbines, the prediction model is trained using the Seagull optimization algorithm to obtain a trained prediction model.

[0067] In an embodiment of the present application, the prediction model may be trained based on the Seagull optimization algorithm based on the key features and the actual output power of the sample wind turbines to obtain a trained prediction model.

[0068] As a possible implementation method, the seagull population of the seagull optimization algorithm is initialized to obtain an initial population; wherein the position value of each seagull in the seagull population represents the value of the hyperparameter of the prediction model; based on the key features and the actual output power of the sample wind turbine, and the initial population, the prediction model is iterated based on the seagull optimization algorithm for no more than a set number of iterations, so that the optimal seagull in the seagull population is updated based on the fitness value of each seagull in any round of iteration.

[0069] Among them, the hyperparameters of the prediction model can be, for example, the penalty factor in support vector regression SVR, the parameters of the RBF (Radial Basis function) kernel function, etc., and this application does not impose any restrictions on this.

[0070] The number of iterations may be pre-set, such as 5000, 10000, etc., and this application does not impose any restrictions on this.

[0071] Among them, the optimal seagull can be determined based on the fitness value of each seagull in the seagull population; among them, the fitness value of any seagull can be determined by inputting key features into a prediction model using the position value of the corresponding seagull, and using an objective function according to the output of the prediction model and the actual output power of the sample wind turbine.

[0072] As an example, assume that the total number of seagulls in the seagull population is 5, and there is seagull A among them. The total number of groups of historical monitoring data is N, and the key feature of the jth group of historical monitoring data is x j , j∈[1,N] and j is a positive integer. When the key feature input to the prediction model (i.e., SVR model) using the position value of Seagull A is x j When the prediction model corresponds to the output y j , the objective function can be expressed as follows:

[0073]

[0074] in, is the actual output power corresponding to the jth group of historical monitoring data.

[0075] According to formula (1), the fitness value of seagull A can be determined as A for:

[0076] Fitness A =MSE; (2)

[0077] Thus, the fitness value of Seagull A can be determined.

[0078] Similar to determining the fitness value of seagull A, the fitness values ​​of other seagulls in the above seagull population can also be determined, which will not be described in detail here.

[0079] It should be noted that the above examples of objective functions are only exemplary. In practical applications, the objective function may be other, for example, the objective function may be a root mean square error (RMSE) function, etc. The corresponding objective function may be set as needed.

[0080] It should also be noted that the optimal seagull can be determined based on the fitness values ​​of each seagull in the seagull population, and this application does not limit the method for determining the optimal seagull. For example, the seagull with the largest fitness value in the seagull population can be determined as the optimal seagull, or the seagull with the smallest fitness value in the seagull population can be determined as the optimal seagull.

[0081] In a possible implementation of the embodiment of the present application, after obtaining the initial population, the initial population can be optimized using Kent chaotic mapping to obtain an initial population with a more uniform distribution.

[0082] To clearly illustrate how the best seagull in the seagull population is updated based on the fitness value of each seagull during any round of iteration, in a possible implementation of the embodiment of the present application, updating the best seagull in the seagull population may include the following steps:

[0083] 1. For any round of iteration, if the order of the iteration process is not greater than the set number of iterations, the position value and fitness value of the optimal seagull in the previous round can be obtained.

[0084] As an example, for the t+1th round of iteration, when the order of the iteration process (t+1) is not greater than the set number of iterations, the position value and fitness value of the optimal seagull in the previous round can be obtained, that is, the position value and fitness value of the optimal seagull in the tth round, where t is a positive integer.

[0085] 2. Based on the position value and fitness value of the best seagull in the previous round, the position of any seagull in the seagull population is updated, and the updated position value of the seagull is used as the position value of the seagull in this round.

[0086] As a possible implementation method, based on the fitness values ​​of each seagull in the previous round of seagull population, the average fitness value of the previous round of seagull population can be determined, and the fitness value of the worst seagull in the previous round can be determined; for any seagull in the seagull population, the following formula can be used to determine the position value of the seagull in this round based on the position value and fitness value of the best seagull in the previous round, the average fitness value of the seagull population in the previous round, the fitness value of the worst seagull in the previous round, and the position value of the seagull:

[0087]

[0088] Among them, L s (t) = AP s (t); (4)

[0089] M s (t) = B(P bs (t)-P s (t)); (5)

[0090] D s (t+1)=|L s (t)+M s (t)|; (6)

[0091]

[0092] B=2r1A 2 ; (8)

[0093]

[0094]

[0095] t is the order of the previous iteration process, t+1 is the order of the current iteration process; the number of seagull population is n; P s (·) is the position value of the seagull during the corresponding iteration; L s (t) represents the new position of the seagull without colliding with other seagulls; A is the anti-collision variable used to represent the seagull's search movement behavior in the given search space; B is the random number responsible for balancing global and local searches; P bs (t) is the position value of the optimal seagull in the last round of iteration; M s (t) is the distance the seagull moves toward the optimal seagull position; D s (t) represents the new position of the seagull; R is a random number in the interval (-2, 2) that follows a normal distribution; f c is a number that decreases linearly from 2 to 0, used to control the frequency of variable A; r1 is a random number in [0,1]; F s i(t) is the fitness value of the i-th seagull in the last iteration; F avg (t) is the average fitness value of the seagull population in the previous iteration; F bs (t) is the fitness value of the optimal seagull in the last round of iteration, F ls (t) is the fitness value of the worst seagull in the previous iteration; x, y, and z are the three-dimensional motion trajectories of the seagull in a spiral shape in the air when attacking prey; r is the radius of each spiral; θ∈[0,2π] is a random angle value; u and v are constants used to control the shape of the spiral.

[0096] It is understandable that the best seagull in any round can be determined based on the fitness value of each seagull in any round of seagull population. Conversely, the worst seagull in any round can also be determined.

[0097] 3. For any seagull in this round of seagull population, determine the fitness value of this round of seagull based on the position value of the seagull.

[0098] That is to say, for any seagull in this round of seagull population, the key features of each historical monitoring data are input into the prediction model using the position value of the seagull. According to the output of the prediction model and the actual output power corresponding to each historical monitoring data, the objective function is used to determine the fitness value of the seagull in this round.

[0099] 4. Determine the optimal seagull in this round based on the fitness values ​​of each seagull in this round and the fitness value of the optimal seagull in the previous round.

[0100] As an example, the fitness value of each seagull in this round can be compared with the fitness value of the best seagull in the previous round. When there is a seagull B among the seagulls in this round whose fitness value is not greater than the fitness value of the best seagull in the previous round, and the fitness value of the seagull B in this round is not greater than the fitness value of any other seagull except this seagull in this round, it can be determined that the seagull B is the best seagull in this round.

[0101] In this way, the optimal seagull in the seagull population can be updated in any round of iteration.

[0102] In another possible implementation of the embodiment of the present application, when the order of the iterative process is greater than the set number of iterations, the position value of the optimal seagull in the previous round can be used as the optimal solution for the hyperparameters of the prediction model, and the current iterative process can be stopped.

[0103] In this way, the optimal solution of the hyperparameters of the prediction model can be effectively obtained, thereby realizing the training of the prediction model.

[0104] The training method of the wind power prediction model in the embodiment of the present application performs model training based on the Seagull optimization algorithm on the prediction model based on key features and the actual output power of the sample wind turbines to obtain a trained prediction model. Therefore, based on the Seagull optimization algorithm, the convergence speed of the model can be accelerated and the global optimization ability of the model can be improved, thereby realizing model training of the prediction model and improving the model recognition ability.

[0105] The above is an embodiment corresponding to the training method of the wind power prediction model. The following is an embodiment of applying the trained prediction model.

[0106] Figure 3 This is a flow chart of the method for predicting wind power provided in the third embodiment of the present application.

[0107] like Figure 3 As shown, the method for predicting wind power may include the following steps:

[0108] Step 301: A trained prediction model is obtained by using a training method for a wind power prediction model.

[0109] In an embodiment of the present application, the training method of the wind power prediction model in any of the above embodiments may be used to obtain a trained prediction model.

[0110] Step 302: Acquire target monitoring data of the target wind turbine generator system during operation.

[0111] In the embodiment of the present application, the target wind turbine generator set may be a wind turbine generator set whose power is to be predicted.

[0112] In the embodiment of the present application, the target monitoring data may include weather forecast data, unit operation data, wind tower data, etc., and the present application does not impose any restrictions on this.

[0113] In an embodiment of the present application, target monitoring data of a target wind turbine set during operation may be obtained.

[0114] Step 303: extract features from the target monitoring data to obtain corresponding target features.

[0115] In an embodiment of the present application, feature extraction may be performed on target monitoring data to obtain corresponding target features.

[0116] It should be noted that the method for obtaining target features of target monitoring data in this embodiment is similar to the method for obtaining key features of each historical monitoring data of sample wind turbines in step 102, and will not be described in detail here.

[0117] Step 304 : Based on the target features, the trained prediction model is used to predict the output power of the target wind turbine to obtain the target output power.

[0118] It is understandable that the trained prediction model has learned the correspondence between key features and output power. Therefore, in this application, the target features can be input into the trained prediction model to predict the output power of the target wind turbine and obtain the target output power.

[0119] The method for predicting wind power in an embodiment of the present application comprises: obtaining a trained prediction model by using a wind power prediction model training method; obtaining target monitoring data of a target wind turbine during operation; performing feature extraction on the target detection data to obtain corresponding target features; and predicting the output power of the target wind turbine using the trained prediction model based on the target features to obtain a target output power. Thus, by using the trained prediction model to identify the target features of the target monitoring data of the target wind turbine, the target output power of the target wind turbine can be effectively determined.

[0120] With the above Figures 1 to 2 Corresponding to the training method of the wind power prediction model provided in the embodiment, the present application also provides a training device for the wind power prediction model. Since the training device for the wind power prediction model provided in the embodiment of the present application is consistent with the above-mentioned Figures 1 to 2 The training method of the wind power prediction model provided in the embodiment corresponds to the embodiment, so the implementation method of the wind power prediction model training method is also applicable to the training device of the wind power prediction model provided in the embodiment of the present application, and will not be described in detail in the embodiment of the present application.

[0121] Figure 4 This is a structural diagram of a training device for a wind power prediction model provided in Example 4 of the present application.

[0122] like Figure 4 As shown, the wind power prediction model training device 400 may include: an acquisition module 401 , an extraction module 402 and a training module 403 .

[0123] The acquisition module 401 is used to acquire multiple sets of historical monitoring data and corresponding actual output power of the sample wind turbines during operation.

[0124] The extraction module 402 is used to perform feature extraction on any set of historical monitoring data to obtain corresponding key features.

[0125] The training module 403 is used to perform model training on the prediction model based on each key feature and the corresponding actual output power to obtain a trained prediction model; wherein the prediction model is a support vector regression SVR model.

[0126] In a possible implementation of the embodiment of the present application, the training module 403 is used to: perform model training based on the Seagull optimization algorithm on the prediction model based on key features and actual output power of sample wind turbines to obtain a trained prediction model.

[0127] In a possible implementation of an embodiment of the present application, the training module 403 is used to: initialize the seagull population of the seagull optimization algorithm to obtain an initial population; wherein the position value of each seagull in the seagull population represents the value of the hyperparameter of the prediction model; based on the key features and the actual output power of the sample wind turbine, and the initial population, the prediction model is iterated for no more than a set number of iterations based on the seagull optimization algorithm, so as to update the optimal seagull in the seagull population based on the fitness value of each seagull in any round of iteration; wherein the optimal seagull is determined based on the fitness value of each seagull in the seagull population; the fitness value of any seagull is determined by inputting the key features into the prediction model using the position value of the corresponding seagull, and using the objective function according to the output of the prediction model and the actual output power of the sample wind turbine.

[0128] In a possible implementation of the embodiment of the present application, the training module 403 is used to: for any round of iteration process, when the order of the iteration process is not greater than the set number of iterations, obtain the position value and fitness value of the best seagull in the previous round; based on the position value and fitness value of the best seagull in the previous round, update the position of any seagull in the seagull population, and use the updated position value of the seagull as the position value of the seagull in this round; for any seagull in the seagull population in this round, determine the fitness value of the seagull in this round based on the position value of the seagull; determine the best seagull in this round according to the fitness value of each seagull in this round and the fitness value of the best seagull in the previous round.

[0129] In a possible implementation of the embodiment of the present application, the training module 403 is configured to: determine, based on the fitness values ​​of each seagull in the previous round of seagull population, an average of the fitness values ​​of the previous round of seagull population, and determine the fitness value of the worst seagull in the previous round; and for any seagull in the seagull population, use the following formula to determine the position value of the seagull in the current round based on the position value and fitness value of the best seagull in the previous round, the average of the fitness values ​​of the previous round of seagull population, the fitness value of the worst seagull in the previous round, and the position value of the seagull:

[0130]

[0131] Among them, L s(t) = AP s (t);

[0132] M s (t) = B(P bs (t)-P s (t));

[0133] D s (t+1)=|L s (t)+M s (t)|;

[0134]

[0135] B=2r1A 2 ;

[0136]

[0137]

[0138] t is the order of the previous iteration process, t+1 is the order of the current iteration process; the number of seagull population is n; P s (·) is the position value of the seagull during the corresponding iteration; L s (t) represents the new position of the seagull without colliding with other seagulls; A is the anti-collision variable used to represent the seagull's search movement behavior in the given search space; B is the random number responsible for balancing global and local searches; P bs (t) is the position value of the optimal seagull in the last round of iteration; M s (t) is the distance the seagull moves toward the optimal seagull position; D s (t) represents the new position of the seagull; R is a random number in the interval (-2, 2) that follows a normal distribution; f c is a number that decreases linearly from 2 to 0, used to control the frequency of variable A; r1 is a random number in [0,1]; F s i (t) is the fitness value of the i-th seagull in the last iteration; F avg (t) is the average fitness value of the seagull population in the previous iteration; F bs (t) is the fitness value of the optimal seagull in the last round of iteration, F ls (t) is the fitness value of the worst seagull in the previous iteration; x, y, and z are the three-dimensional motion trajectories of the seagull in a spiral shape in the air when attacking prey; r is the radius of each spiral; θ∈[0,2π] is a random angle value; u and v are constants used to control the shape of the spiral.

[0139] In a possible implementation of the embodiment of the present application, the wind power prediction model training device 400 may further include:

[0140] The optimization module is used to optimize the initial population using Kent chaotic mapping to obtain an initial population with a more uniform distribution.

[0141] In a possible implementation of the embodiment of the present application, the historical monitoring data includes at least one of weather forecast data, unit operation data, and wind tower data.

[0142] In a possible implementation of an embodiment of the present application, the extraction module 402 is used to: determine target features based on multiple groups of historical monitoring data using a feature selection algorithm; and for any historical monitoring data, perform feature extraction on the historical monitoring data according to the target features to obtain corresponding key features.

[0143] The training device for a wind power prediction model in an embodiment of the present application obtains multiple sets of historical monitoring data and the corresponding actual output power of sample wind turbines during operation; performs feature extraction on any set of historical monitoring data to obtain corresponding key features; and trains a prediction model based on each key feature and the corresponding actual output power to obtain a trained prediction model, wherein the prediction model is a support vector regression (SVR) model. This allows for training of the wind power prediction model, improving the model's prediction performance, namely, improving the accuracy and reliability of the model's prediction results.

[0144] With the above Figure 3 The method for predicting wind power provided in the embodiment corresponds to the method for predicting wind power. The present application also provides a device for predicting wind power. Figure 3 The method for predicting wind power provided in the embodiment corresponds to the method for predicting wind power, so the implementation method of the method for predicting wind power is also applicable to the device for predicting wind power provided in the embodiment of the present application, and will not be described in detail in the embodiment of the present application.

[0145] Figure 5 This is a structural diagram of a device for predicting wind power provided in Example 5 of the present application.

[0146] like Figure 5 As shown, the device 500 for predicting wind power may include: a first acquisition module 501 , a second acquisition module 502 , an extraction module 503 and a prediction module 504 .

[0147] The first acquisition module 501 is configured to acquire a trained prediction model using the wind power prediction model training method of any of the aforementioned embodiments.

[0148] The second acquisition module 502 is used to acquire target monitoring data of the target wind turbine during operation.

[0149] The extraction module 503 is used to extract features from the target monitoring data to obtain corresponding target features.

[0150] The prediction module 504 is configured to predict the output power of the target wind turbine generator system based on the target characteristics using a trained prediction model to obtain the target output power.

[0151] The device for predicting wind power according to an embodiment of the present application uses a wind power prediction model training method to obtain a trained prediction model; obtains target monitoring data of a target wind turbine during operation; performs feature extraction on the target detection data to obtain corresponding target features; and based on the target features, uses the trained prediction model to predict the output power of the target wind turbine to obtain a target output power. Thus, by using the trained prediction model to identify the target features of the target monitoring data of the target wind turbine, the target output power of the target wind turbine can be effectively determined.

[0152] In order to implement the above embodiments, the present invention also proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the program, it implements the training method of the wind power prediction model proposed in any of the aforementioned embodiments of the present invention, or the method for predicting wind power.

[0153] In order to implement the above embodiments, the present invention also proposes a non-temporary computer-readable storage medium on which a computer program is stored, characterized in that when the program is executed by a processor, it implements the training method of the wind power prediction model proposed in any of the aforementioned embodiments of the present invention, or the method for predicting wind power.

[0154] In order to implement the above embodiments, the present invention also proposes a computer program product. When the instructions in the computer program product are executed by a processor, the training method of the wind power prediction model or the method for predicting wind power proposed in any of the aforementioned embodiments of the present invention is executed.

[0155] According to an embodiment of the present invention, the present invention further provides an electronic device, a non-transitory computer-readable storage medium, and a computer program product.

[0156] like Figure 6As shown, electronic device 12 is implemented as a general-purpose computing device. Components of electronic device 12 may include, but are not limited to, one or more processors or processing units 16, system memory 28, and a bus 18 that connects various system components (including system memory 28 and processing unit 16).

[0157] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures. Examples of such architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnection (PCI) bus.

[0158] The electronic device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 12, including volatile and non-volatile media, removable and non-removable media.

[0159] The memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The electronic device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be configured to read and write non-removable, non-volatile magnetic media ( Figure 6 Not shown, often called a "hard drive"). Although Figure 6Although not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a floppy disk) and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a Compact Disc Read Only Memory (CD-ROM), a Digital Video Disc Read Only Memory (DVD-ROM), or other optical media) may be provided. In these cases, each drive may be connected to bus 18 via one or more data media interfaces. Memory 28 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of various embodiments of the present invention.

[0160] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in memory 28. Such program modules 42 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may include an implementation of a network environment. Program modules 42 generally implement the functions and / or methods of the embodiments described herein.

[0161] The electronic device 12 can also communicate with one or more external devices 14 (e.g., a keyboard, pointing device, display 24, etc.), one or more devices that enable a user to interact with the electronic device 12, and / or any device that enables the electronic device 12 to communicate with one or more other computing devices (e.g., a network card, a modem, etc.). This communication can occur via an input / output (I / O) interface 22. Furthermore, the electronic device 12 can communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network such as the Internet) via a network adapter 20. As shown, the network adapter 20 communicates with other modules of the electronic device 12 via the bus 18. It should be understood that, although not shown, other hardware and / or software modules can be used in conjunction with the electronic device 12, including but not limited to microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0162] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, such as implementing the methods mentioned in the above embodiments.

[0163] In the description of this specification, the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0164] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0165] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention pertain.

[0166] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and a portable compact disc read-only memory (CDROM). Furthermore, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing it in another suitable manner if necessary, and then storing it in a computer memory.

[0167] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0168] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0169] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing module, or each unit may exist physically separately, or two or more units may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or in the form of software functional modules. If the integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium.

[0170] The storage medium mentioned above may be a read-only memory, a magnetic disk, or an optical disk, etc. Although the embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and are not to be construed as limiting the present invention. Persons skilled in the art may make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A training method for a wind power prediction model, characterized in that: The method comprises: Obtain multiple sets of historical monitoring data and corresponding actual output power of sample wind turbines during operation; Perform feature extraction on any set of historical monitoring data to obtain the corresponding key features; Based on each of the key features and the corresponding actual output power, a prediction model is trained to obtain a trained prediction model; wherein the prediction model is a support vector regression (SVR) model; The method of training the prediction model based on each of the key features and the corresponding actual output power to obtain a trained prediction model includes: Based on the key features and the actual output power of the sample wind turbines, performing model training based on the Seagull optimization algorithm on the prediction model to obtain a trained prediction model; The method of performing model training based on the Seagull optimization algorithm on the prediction model based on the key features and the actual output power of the sample wind turbine to obtain a trained prediction model includes: Initializing a seagull population of a seagull optimization algorithm to obtain an initial population; wherein the position value of each seagull in the seagull population represents the value of a hyperparameter of the prediction model; Based on the key features and the actual output power of the sample wind turbines, and the initial population, the prediction model is subjected to an iterative process based on the seagull optimization algorithm for no more than a set number of iterations, so as to update the optimal seagull in the seagull population based on the fitness value of each seagull in any round of the iteration process; wherein, the optimal seagull is determined based on the fitness value of each seagull in the seagull population; the fitness value of any seagull is determined by inputting the key features into the prediction model using the position value of the corresponding seagull, and using the objective function according to the output of the prediction model and the actual output power of the sample wind turbines.

2. The method according to claim 1, characterized in that The updating of the optimal seagull in the seagull population based on the fitness value of each seagull in any round of the iteration process includes: For any round of the iterative process, when the order of the iterative process is not greater than the set number of iterations, obtaining the position value and fitness value of the optimal seagull in the previous round; Based on the position value and fitness value of the optimal seagull in the previous round, the position of any seagull in the seagull population is updated, so that the updated position value of the seagull is used as the position value of the seagull in this round; For any seagull in the seagull population in this round, determine the fitness value of the seagull in this round based on the position value of the seagull; The optimal seagull of this round is determined according to the fitness value of each seagull in this round and the fitness value of the optimal seagull in the previous round.

3. The method according to claim 2, characterized in that The updating of the position of any seagull in the seagull population based on the position value and fitness value of the best seagull in the previous round, so as to use the updated position value of the seagull as the position value of the seagull in this round, includes: According to the fitness values ​​of each seagull in the seagull population in the previous round, determine the average fitness value of the seagull population in the previous round, and determine the fitness value of the worst seagull in the previous round; For any seagull in the seagull population, the position value of the seagull in this round is determined using the following formula based on the position value and fitness value of the best seagull in the previous round, the average fitness value of the seagull population in the previous round, the fitness value of the worst seagull in the previous round, and the position value of the seagull: ； in, ; ； ； ； ； ； ； t is the order of the previous iteration process, t+1 is the order of the current iteration process; the number of seagull population is n; P s (·) is the position value of the seagull in the corresponding iteration process; L s (t) represents the new position of the seagull without colliding with other seagulls; A is an anti-collision variable used to represent the seagull's search movement behavior in a given search space; B is a random number responsible for balancing global and local searches; P bs (t) is the position value of the optimal seagull in the last round of iteration; M s (t) is the distance the seagull moves toward the optimal seagull position; D s (t) represents the new position of the seagull; R is a random number in the interval (-2, 2) that follows a normal distribution; f c is a number that decreases linearly from 2 to 0, used to control the frequency of variable A; r1 is a random number in [0,1]; F s i (t) is the fitness value of the i-th seagull in the last iteration; F avg (t) is the average fitness value of the seagull population in the previous iteration; F bs (t) is the fitness value of the optimal seagull in the last round of iteration, F ls (t) is the fitness value of the worst seagull in the previous iteration; x, y, and z are the three-dimensional motion trajectories of the seagull in a spiral shape in the air when attacking prey; r is the radius of each spiral; θ∈[0,2π] is a random angle value; u and v are constants used to control the shape of the spiral.

4. The method according to claim 1, wherein After initializing the seagull population of the seagull optimization algorithm to obtain an initial population, the method further includes: The initial population is optimized by using Kent chaotic mapping to obtain an initial population with a more uniform distribution.

5. The method according to claim 1, wherein The historical monitoring data includes at least one of weather forecast data, unit operation data and wind tower data.

6. The method according to any one of claims 1 to 5, characterized in that: The feature extraction of any set of historical monitoring data to obtain corresponding key features includes: Determining target features using a feature selection algorithm based on the multiple sets of historical monitoring data; For any of the historical monitoring data, feature extraction is performed on the historical monitoring data according to the target feature to obtain corresponding key features.

7. A method for predicting wind power, characterized in that: The method comprises: Obtaining a trained prediction model using the method according to any one of claims 1 to 6; Obtain target monitoring data of the target wind turbine during operation; Performing feature extraction on the target monitoring data to obtain corresponding target features; Based on the target characteristics, the trained prediction model is used to predict the output power of the target wind turbine generator set to obtain the target output power.

8. A training device for a wind power prediction model, characterized in that: The device comprises: An acquisition module is used to obtain multiple sets of historical monitoring data and corresponding actual output power of sample wind turbines during operation; The extraction module is used to extract features from any set of historical monitoring data to obtain the corresponding key features; A training module, configured to perform model training on a prediction model based on each of the key features and the corresponding actual output power to obtain a trained prediction model; wherein the prediction model is a support vector regression (SVR) model; The training module is specifically used to: Based on the key features and the actual output power of the sample wind turbines, performing model training based on the Seagull optimization algorithm on the prediction model to obtain a trained prediction model; The training module is specifically used to: Initializing a seagull population of a seagull optimization algorithm to obtain an initial population; wherein the position value of each seagull in the seagull population represents the value of a hyperparameter of the prediction model; Based on the key features and the actual output power of the sample wind turbines, and the initial population, the prediction model is subjected to an iterative process based on the seagull optimization algorithm for no more than a set number of iterations, so as to update the optimal seagull in the seagull population based on the fitness value of each seagull in any round of the iteration process; wherein, the optimal seagull is determined based on the fitness value of each seagull in the seagull population; the fitness value of any seagull is determined by inputting the key features into the prediction model using the position value of the corresponding seagull, and using the objective function according to the output of the prediction model and the actual output power of the sample wind turbines.

Images