Training Method, Device and Computer Equipment for Wind Power Prediction Model

Abstract

The present application relates to a training method, device and computer equipment for a wind power prediction model. The method includes: obtaining a sample meteorological data set, where the sample meteorological data set includes sample wind speeds at each moment within a preset period and sample wind power corresponding to the sample wind speeds; dividing each of the sample wind speeds by a model according to wind speed fluctuations to obtain a sample wind speed group corresponding to each wind speed fluctuation process; for each wind speed fluctuation process, training an initial wind power prediction model according to the sample wind speed group corresponding to the wind speed fluctuation process and the sample wind power group corresponding to the sample wind speed group, to obtain a wind power prediction model corresponding to the wind speed fluctuation process. Using this method can improve the accuracy of the wind power prediction model.

Description

Technical Field

[0001] The present application relates to the field of power system operation, and particularly to a method, device, and computer equipment for training a wind power prediction model. Background Art

[0002] With the continuous development of wind power, the proportion of wind power generation in the power system has gradually increased. However, due to the strong volatility and uncertainty of wind power, it will affect the real-time energy supply-demand balance of the power system. Accurate wind power prediction technology can effectively reduce the uncertainty of wind power and improve the ability of the power system to maintain the real-time energy supply-demand balance.

[0003] In traditional wind power prediction models, only weather conditions with long-term regular characteristics can be predicted, while the meteorological characteristic data of short-term numerical weather forecasts cannot guarantee strict regular characteristics, resulting in low accuracy of short-term wind power prediction data. Summary of the Invention

[0004] Based on this, it is necessary to provide a method, device, and computer equipment for training a wind power prediction model for the above technical problems.

[0005] In a first aspect, the present application provides a method for training a wind power prediction model. The method includes:

[0006] Obtain a sample meteorological data set, where the sample meteorological data set includes sample wind speeds at each moment within a preset time period and sample wind power corresponding to the sample wind speeds;

[0007] Divide each of the sample wind speeds according to a wind speed fluctuation division model to obtain a sample wind speed group corresponding to each wind speed fluctuation process;

[0008] For each wind speed fluctuation process, train an initial wind power prediction model according to the sample wind speed group corresponding to the wind speed fluctuation process and the sample wind power group corresponding to the sample wind speed group to obtain a wind power prediction model corresponding to the wind speed fluctuation process.

[0009] Optionally, the step of dividing each of the sample wind speeds according to a wind speed fluctuation division model to obtain a sample wind speed group corresponding to each wind speed fluctuation process includes:

[0010] Select each valid sample wind speed that meets a preset screening condition from each of the sample wind speeds;

[0011] Determine a sample wind speed group corresponding to each wind speed fluctuation process according to each of the valid sample wind speeds and the wind speed fluctuation division model.

[0012] Optionally, the wind speed fluctuation division model includes the sample ratio corresponding to each wind speed fluctuation process; the determining of the sample wind speed group corresponding to each wind speed fluctuation process according to each of the effective sample wind speeds and the wind speed fluctuation division model includes:

[0013] Determining each peak value in the wind speed sequence formed by each of the effective sample wind speeds, and the sample interval corresponding to each peak value in the wind speed sequence;

[0014] Dividing the peak value distribution sequence according to the sample ratio corresponding to each wind speed fluctuation process to obtain the wind speed threshold corresponding to each wind speed fluctuation process; the peak value distribution sequence is obtained by sorting the number of peak values in each preset wind speed interval in ascending order of wind speed;

[0015] Determining the wind speed range corresponding to each wind speed fluctuation process according to the wind speed threshold corresponding to each wind speed fluctuation process;

[0016] For each wind speed fluctuation process, determining the peak values within the wind speed range corresponding to the wind speed fluctuation process, and taking the effective sample wind speeds within the sample interval corresponding to the peak values as the sample wind speed group of the wind speed fluctuation process.

[0017] Optionally, the training of the initial wind power prediction model respectively according to the sample wind speed group of each wind speed fluctuation process and the sample wind power group corresponding to the sample wind speed group of each wind speed fluctuation process to determine the wind power prediction model corresponding to each wind speed fluctuation process includes:

[0018] For each wind speed fluctuation process, determining the predicted wind power group corresponding to the sample wind speed group and the group of model parameter values corresponding to the sample wind speed group according to the sample wind speed group of the wind speed fluctuation process and the initial wind power prediction model;

[0019] Determining the optimal model parameter value according to the group of model parameter values, the predicted wind power group and the slime mold algorithm;

[0020] Determining the function value of the evaluation function according to the predicted wind power group corresponding to the sample wind speed group, the sample wind power group corresponding to the sample wind speed group and the evaluation function;

[0021] When the preset iteration stop condition is not satisfied, recording the function value of the evaluation function and the optimal model parameter value, and returning to execute the step of determining the predicted wind power group corresponding to the sample wind speed group and the group of model parameter values corresponding to the sample wind speed group according to the sample wind speed group of the wind speed fluctuation process and the initial wind power prediction model until the preset iteration stop condition is satisfied, and determining the function value of each loss function and the optimal model parameter value corresponding to the function value of each loss function;

[0022] Determine the function value of the target loss function and the corresponding target optimal model parameter value of the target loss function according to the function values of the respective loss functions and the optimization algorithm.

[0023] Based on the target optimal model parameter value, obtain the wind power prediction model for the wind speed fluctuation process.

[0024] In a second aspect, the present application provides a method for predicting wind power. The method includes:

[0025] Obtain a target meteorological data set, where the target meteorological data set is the target wind speed at each moment within a preset time period.

[0026] Divide each of the target wind speeds according to the wind speed fluctuation division model to obtain a target wind speed group corresponding to each wind speed fluctuation process.

[0027] For each wind speed fluctuation process, input the target wind speed group of the wind speed fluctuation process into the wind power prediction model corresponding to the wind speed fluctuation process to determine the target predicted wind power group corresponding to the fluctuation process.

[0028] Arrange each of the target predicted wind powers in the time series of the target wind speed corresponding to the target predicted wind power to obtain the predicted wind power.

[0029] Wherein, the wind power prediction model is trained by the training method of the wind power prediction model according to any one of claims 1 to 4.

[0030] In a third aspect, the present application provides a training device for a wind power prediction model. The device includes:

[0031] An acquisition module, configured to acquire a sample meteorological data set, where the sample meteorological data set includes the sample wind speed at each moment within a preset time period and the sample wind power corresponding to the sample wind speed.

[0032] A division module, configured to divide each of the sample wind speeds according to the wind speed fluctuation division model to obtain a sample wind speed group corresponding to each wind speed fluctuation process.

[0033] A training module, configured to, for each wind speed fluctuation process, train an initial wind power prediction model according to the sample wind speed group corresponding to the wind speed fluctuation process and the sample wind power group corresponding to the sample wind speed group to obtain the wind power prediction model corresponding to the wind speed fluctuation process.

[0034] Optionally, the division module is specifically configured to:

[0035] Among the respective sample wind speeds, select each valid sample wind speed that meets a preset screening condition.

[0036] Determine the sample wind speed group corresponding to each of the wind speed fluctuation processes according to each of the effective sample wind speeds and the wind speed fluctuation division model.

[0037] Optionally, the wind speed fluctuation division model includes the sample ratio corresponding to each wind speed fluctuation process; the division module is specifically configured to:

[0038] Determine each peak value in the wind speed sequence formed by each of the effective sample wind speeds, and the sample interval corresponding to each peak value in the wind speed sequence;

[0039] Divide the peak value distribution sequence according to the sample ratio corresponding to each wind speed fluctuation process to obtain the wind speed threshold corresponding to each wind speed fluctuation process; the peak value distribution sequence is obtained by sorting the number of peak values in each preset wind speed interval in ascending order of wind speed;

[0040] Determine the wind speed range corresponding to each wind speed fluctuation process according to the wind speed threshold corresponding to each wind speed fluctuation process;

[0041] For each wind speed fluctuation process, determine the peak value within the wind speed range corresponding to the wind speed fluctuation process, and use the effective sample wind speed within the sample interval corresponding to the peak value as the sample wind speed group of the wind speed fluctuation process.

[0042] Optionally, the training module is specifically configured to:

[0043] For each wind speed fluctuation process, determine the predicted wind power group corresponding to the sample wind speed group and the model parameter value group corresponding to the sample wind speed group according to the sample wind speed group of the wind speed fluctuation process and the initial wind power prediction model;

[0044] Determine the optimal model parameter value according to the model parameter value group, the predicted wind power group and the slime mold algorithm;

[0045] Determine the function value of the evaluation function according to the predicted wind power group corresponding to the sample wind speed group, the sample wind power group corresponding to the sample wind speed group and the evaluation function;

[0046] In the case of not meeting the preset iteration stop condition, record the function value of the evaluation function and the optimal model parameter value, and return to execute the step of determining the predicted wind power group corresponding to the sample wind speed group and the model parameter value group corresponding to the sample wind speed group according to the sample wind speed group of the wind speed fluctuation process and the initial wind power prediction model, until the preset iteration stop condition is met, and determine the function value of each loss function and the optimal model parameter value corresponding to the function value of each loss function;

[0047] Determine the function value of the target loss function and the corresponding target optimal model parameter value of the target loss function according to the function values of the respective loss functions and the optimization algorithm;

[0048] Based on the target optimal model parameter value, obtain the wind power prediction model for the wind speed fluctuation process.

[0049] In a fourth aspect, the present application provides a device for predicting wind power, the device comprising:

[0050] An acquisition module, configured to acquire a target meteorological data set, where the target meteorological data set is the target wind speed at each moment within a preset time period;

[0051] A partitioning module, configured to partition each of the target wind speeds according to a wind speed fluctuation partitioning model to obtain each target wind speed group corresponding to each wind speed fluctuation process;

[0052] A determination module, configured to, for each wind speed fluctuation process, input the target wind speed group of the wind speed fluctuation process into the wind power prediction model corresponding to the wind speed fluctuation process, and determine the target predicted wind power group corresponding to the fluctuation process;

[0053] A sorting module, configured to sort the respective target predicted wind powers according to the time series of the target wind speeds corresponding to the target predicted wind powers to obtain the predicted wind power;

[0054] Wherein, the wind power prediction model is trained by the training method of the wind power prediction model according to any one of claims 1 to 4.

[0055] In a fifth aspect, the present application provides a computer device. The computer device includes: a memory and a processor, the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the method according to any one of the first aspect or the second aspect are implemented.

[0056] In a sixth aspect, the present application provides a computer-readable storage medium. The storage medium includes: a computer program stored thereon, and is characterized in that when the computer program is executed by a processor, the steps of the method according to any one of the first aspect or the second aspect are implemented.

[0057] In a seventh aspect, the present application provides a computer program product. The computer program product includes: a computer program, and is characterized in that when the computer program is executed by a processor, the steps of the method according to any one of the first aspect or the second aspect are implemented.

[0058] The training method, device, and computer equipment for the above wind power prediction model include obtaining a sample meteorological data set, where the sample meteorological data set contains sample wind speeds at each moment within a preset time period and the sample wind power corresponding to the sample wind speeds; dividing each of the sample wind speeds according to wind speed fluctuations by a model to obtain a sample wind speed group corresponding to each wind speed fluctuation process; for each wind speed fluctuation process, training an initial wind power prediction model based on the sample wind speed group corresponding to the wind speed fluctuation process and the sample wind power group corresponding to the sample wind speed group to obtain a wind power prediction model corresponding to the wind speed fluctuation process. By carefully dividing each sample wind speed with different magnitudes of wind speed, obtaining a sample wind speed group corresponding to each wind speed fluctuation process, and separately training the wind power prediction model with the sample wind speed groups corresponding to each wind speed fluctuation process, the accuracy of the wind power prediction model is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a schematic flowchart of the training method for the wind power prediction model in an embodiment;

[0060] Figure 2 It is a schematic flowchart of the sample wind speed division method in an embodiment;

[0061] Figure 3 It is a schematic flowchart of the method for determining the sample wind speed group in an embodiment;

[0062] Figure 4 It is an example diagram of the distribution of sample wind speeds in a time series in an embodiment;

[0063] Figure 5 It is an example diagram of the sample peak distribution in an embodiment;

[0064] Figure 6 It is an example diagram of the wind speed threshold in an embodiment;

[0065] Figure 7 It is a schematic flowchart of the training steps of the wind power prediction model in an embodiment;

[0066] Figure 8 It is a schematic flowchart of the method for predicting wind power in an embodiment;

[0067] Figure 9 It is a schematic flowchart of the training method for the wind power prediction model in another embodiment;

[0068] Figure 10 It is a schematic structural diagram of the training device for the wind power prediction model in an embodiment;

[0069] Figure 11Schematic structural diagram of a wind power prediction device in an embodiment;

[0070] Figure 12 Internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0071] To make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0072] The training method for the wind power prediction model provided by the embodiments of the present application can be applied to a terminal, a server, or a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the terminal may include, but is not limited to, various personal computers, laptop computers, tablet computers, Internet of Things devices, etc. The terminal is used to obtain a sample meteorological data set, divide each sample wind speed in the sample meteorological data set through a fluctuation process division model, and obtain a wind power prediction model corresponding to each wind speed fluctuation process through training.

[0073] In one embodiment, as Figure 1 shown, a training method for a wind power prediction model is provided. Taking the application of this method to a terminal as an example, the method includes the following steps:

[0074] Step S101, obtain a sample meteorological data set.

[0075] Among them, the sample meteorological data set includes sample wind speeds at each moment within a preset time period and sample wind power corresponding to the sample wind speeds.

[0076] In this embodiment, the terminal obtains historical meteorological data (i.e., sample meteorological data) at each moment within a preset time period through a cloud network such as a meteorological bureau, a meteorological mapping laboratory, or a meteorological observation station. The historical meteorological data includes historical wind speed data (i.e., sample wind speeds) and wind power corresponding to the historical wind speeds (i.e., sample wind power). The preset time period is the time period involved in the historical wind speed data set. According to the sample wind speed data set obtained by the terminal and the set of sample wind power corresponding to each sample wind speed data, the sample meteorological data set is determined.

[0077] Step S102, divide each sample wind speed according to the wind speed fluctuation division model to obtain a sample wind speed group corresponding to each wind speed fluctuation process.

[0078] In this embodiment, the terminal divides each obtained sample wind speed through the wind speed fluctuation division model to determine the fluctuation process to which each sample wind speed belongs, so as to determine the sample wind speed group in this fluctuation process.

[0079] Step S103: For each wind speed fluctuation process, train the initial wind power prediction model based on the sample wind speed group corresponding to the wind speed fluctuation process and the sample wind power group corresponding to the sample wind speed group, to obtain the wind power prediction model corresponding to the wind speed fluctuation process.

[0080] In this embodiment, for each fluctuation process, the terminal inputs the sample wind speed group of this fluctuation process and each sample wind speed in the sample wind speed group of this fluctuation process into the initial wind power prediction model, and trains the initial wind power prediction model to obtain the wind power prediction model corresponding to this wind speed fluctuation process.

[0081] Based on the above solution, by carefully dividing each sample wind speed with different magnitudes of wind speed, the sample wind speed groups corresponding to each wind speed fluctuation process are obtained, and the sample wind speed groups corresponding to each wind speed fluctuation process are respectively used to train the wind power prediction model, so as to obtain the wind power prediction models corresponding to each wind speed fluctuation process, improving the accuracy of the wind power prediction model.

[0082] Optionally, as Figure 2 shown, dividing each sample wind speed according to the wind speed fluctuation division model to obtain the sample wind speed groups corresponding to each wind speed fluctuation process, including:

[0083] Step S201: In each sample wind speed, select each effective sample wind speed that meets the preset screening conditions.

[0084] In this embodiment, the preset screening conditions may include a zero output wind speed threshold, a maximum wind speed threshold, and a preset slope threshold. Among them, the zero output wind speed threshold can be determined according to the lowest power generation wind speed of the wind power generation equipment, the maximum wind speed threshold can be determined according to the maximum wind speed that the wind power generation equipment can withstand, and this maximum wind speed threshold can be 25 m / s. The slope threshold can be determined according to the wind speed change slope with the smallest number of occurrences in the historical meteorological data. First, exclude the sample wind speeds less than or equal to the zero output wind speed threshold, secondly, exclude the sample wind speeds greater than the maximum wind speed threshold, and finally, exclude the sample wind speeds whose change slope between two adjacent sample wind speeds is greater than the preset slope threshold. The terminal screens each sample wind speed through the above preset screening conditions, and selects the sample wind speeds that pass the screening conditions as effective sample wind speeds. The specific preset screening formula is:

[0085] K = |(V i -V j ) / (T i -T j )| ≤ a

[0086] V i ≥ b

[0087] Vi ≤c

[0088] In the above formula, i is the serial number obtained by arranging the moments of the wind speed of the sample to be screened in the collected time period, j is the previous or next serial number of the serial number obtained by arranging the moments of the wind speed of the sample to be screened in the collected time period, K is the wind speed change slope between the sample wind speeds corresponding to two adjacent moments, V i is the wind speed of the sample to be screened, V i+1 is the sample wind speed corresponding to the previous or next serial number of the serial number obtained by arranging the moments of the wind speed of the sample to be screened in the collected time period, T i is the moment at which the wind speed of the sample to be screened is located, T j is the moment at which the sample wind speed corresponding to the previous or next serial number of the serial number obtained by arranging the moments of the wind speed of the sample to be screened in the collected time period is located, a is the slope threshold, b is the maximum wind speed threshold, and C is the zero output wind speed threshold.

[0089] Step S202: Determine the sample wind speed groups corresponding to each wind speed fluctuation process according to each effective sample wind speed and the wind speed fluctuation division model.

[0090] In this embodiment, the terminal divides different sample wind speeds into each wind speed fluctuation process through the wind speed fluctuation division model for each effective sample wind speed, and forms the sample wind speed groups corresponding to each wind speed fluctuation process.

[0091] Based on the above solution, the terminal obtains the effective sample wind speed by screening the sample wind speed, and divides the effective sample wind speed into each wind speed fluctuation process, which improves the reliability and universality of the training sample, thereby indirectly improving the accuracy and practicability of the wind power model.

[0092] Optionally, as Figure 3 shown, the wind speed fluctuation division model includes the sample ratios corresponding to each wind speed fluctuation process; determining the sample wind speed groups corresponding to each wind speed fluctuation process according to each effective sample wind speed and the wind speed fluctuation division model includes:

[0093] Step S301: Determine each peak value in the wind speed sequence formed by each effective sample wind speed, and the sample interval corresponding to each peak value in the wind speed sequence.

[0094] In this embodiment, the terminal sorts the effective sample wind speeds in the wind speed sequence to obtain a set of effective sample wind speeds. The wind speed sequence is a wind speed sequence formed by the effective sample wind speeds in the order of the time when the sample wind speeds are collected. The effective sample wind speeds in the wind speed sequence will have different wind speed change situations due to the different magnitudes of the effective sample wind speeds, thus forming peaks and valleys. The terminal determines the effective sample wind speed corresponding to the minimum wind speed point as the valley, determines the effective sample wind speed corresponding to the maximum wind speed point as the peak, and determines the set of effective sample wind speeds between two adjacent valleys as the sample interval. The peak within this sample interval is determined as the peak corresponding to this sample interval. Therefore, one peak corresponds to one sample interval. For example Figure 4 As shown, the minimum sample wind speed point (left) and the minimum sample wind speed point (right) are valleys, the maximum sample wind speed point (peak value) is the peak, and the interval between the minimum sample wind speed point (left) and the minimum sample wind speed point (right) is the sample interval, and this maximum sample wind speed point (peak value) is the peak corresponding to this sample interval.

[0095] Step S302: Divide the peak distribution sequence according to the sample ratio corresponding to each wind speed fluctuation process to obtain the wind speed threshold corresponding to each wind speed fluctuation process.

[0096] Among them, the peak distribution sequence is obtained by sorting the number of peaks in each preset wind speed interval in ascending order of wind speed.

[0097] In this embodiment, the terminal compares the ranges where all peak wind speeds are located, divides the wind speed range formed by the minimum peak and the maximum peak in this range into different wind speed intervals (i.e., preset wind speed intervals) in an equal-proportion manner; classifies each peak into the corresponding wind speed interval, and determines the number of peaks corresponding to different wind speed intervals; finally, arranges each wind speed interval in ascending order of peaks to obtain the peak distribution sequence. For example Figure 5 As shown, it is a peak distribution sequence diagram in an embodiment, where the abscissa is the peak value and the ordinate is the number of peaks. The equal proportion can be a proportion with an interval of 5 m / s. The wind speed intervals are 0 m / s - 5 m / s, 5 m / s - 10 m / s,..., 25 m / s - 30 m / s. When the peak value is 6 m / s, the interval to which this peak value belongs is 5 m / s - 10 m / s.

[0098] The terminal determines the number of samples corresponding to each wind speed fluctuation process according to the sample ratio corresponding to each wind speed fluctuation process and the total number of peaks; and divides the peak distribution sequence by each number of samples, so that the number of peaks corresponding to each divided wind speed fluctuation process meets the number of samples corresponding to each wind speed fluctuation process. The abscissa corresponding to the boundary point of the wind speed fluctuation process is determined as the wind speed threshold corresponding to each wind speed fluctuation process, and the wind speed threshold can be a small fluctuation threshold and a large fluctuation threshold. In another embodiment, as Figure 6 shown, it is a wind speed threshold diagram. Among them, the peak distribution sequence can also be represented by the frequency of the number of peaks corresponding to the wind speed under different wind speed conditions in the total number of peaks (i.e., peak frequency). Among them, the abscissa is the wind speed, the ordinate is the peak frequency, and the area formed between the abscissa and the ordinate is the peak probability (i.e., the proportion of the number of peaks in the wind speed interval in the total number of peaks). When the sample ratio corresponding to each wind speed fluctuation process is 1:1:1, it is determined that the sample probability corresponding to each wind speed fluctuation process is 1 / 3. The peak probability is divided by the sample probability, and it is determined that the peak probability corresponding to each wind speed fluctuation process is 1 / 3. Then, the boundary point (i.e., the abscissa wind speed) of the peak probability corresponding to each wind speed fluctuation process is determined as the wind speed threshold corresponding to each wind speed fluctuation process.

[0099] Step S303: Determine the wind speed range corresponding to each wind speed fluctuation process according to the wind speed threshold corresponding to each wind speed fluctuation process.

[0100] In this embodiment, the terminal determines the wind speed range corresponding to the small fluctuation process as the wind speed range between the zero output wind speed threshold and the small fluctuation threshold according to the wind speed threshold corresponding to each wind speed fluctuation process; determines the wind speed range corresponding to the medium fluctuation process as the wind speed range between the small fluctuation threshold and the large fluctuation threshold; and determines the wind speed range corresponding to the large fluctuation process as the wind speed range between the large fluctuation threshold and the maximum wind speed threshold.

[0101] Step S304: For each wind speed fluctuation process, determine the peaks within the wind speed range corresponding to the wind speed fluctuation process, and use the effective sample wind speeds within the sample interval corresponding to the peaks as the sample wind speed group of the wind speed fluctuation process.

[0102] In this embodiment, for each wind speed fluctuation process, the terminal determines the sample interval corresponding to each peak within the wind speed range corresponding to the obtained wind speed fluctuation process as the sample interval corresponding to the wind speed fluctuation process; and determines the effective sample wind speeds within the sample interval as the effective sample wind speeds corresponding to the wind speed fluctuation process; the set of effective sample wind speeds formed by all the effective sample wind speeds in the wind speed fluctuation process is determined as the sample wind speed group in each wind speed fluctuation process. The specific expression of the wind speed fluctuation process is as follows:

[0103]

[0104] In the above formula, W v1 is the small wind speed fluctuation process, W v2 is the medium wind speed fluctuation process, W v3 is the large wind speed fluctuation process, ε v0 is the zero output threshold, ε v1 is the small fluctuation threshold, ε v2 is the large fluctuation threshold, ε vmax is the maximum wind speed threshold, v i is the wind speed corresponding to the effective sample wind speed.

[0105] Based on the above scheme, by dividing the model according to each effective sample wind speed and wind speed fluctuation, the sample wind speed groups corresponding to each wind speed fluctuation process are determined, ensuring that the wind power prediction models corresponding to each fluctuation process can be trained, and improving the accuracy and practicability of the wind power models corresponding to each fluctuation process.

[0106] Optionally, as Figure 7 shown, according to the sample wind speed groups of each wind speed fluctuation process and the sample wind power groups corresponding to the sample wind speed groups of each wind speed fluctuation process, the initial wind power prediction model is trained respectively to determine the wind power prediction models corresponding to each wind speed fluctuation process, including:

[0107] Step S701, for each wind speed fluctuation process, according to the sample wind speed group of the wind speed fluctuation process and the initial wind power prediction model, determine the predicted wind power group corresponding to the sample wind speed group, and the model parameter value group corresponding to the sample wind speed group.

[0108] In this embodiment, the initial wind power prediction model can be an ELM network model. The terminal inputs each sample wind speed in the sample wind speed group of the wind speed fluctuation process into the initial wind power prediction model respectively. After the first sample wind speed is input into the initial wind power prediction model, the initial wind power model will automatically change the parameters, and the parameters are random parameters for the next operation. After each sample wind speed is input, the predicted wind power corresponding to the sample wind speed and the parameter value of the wind power prediction model corresponding to the sample wind speed will be obtained. After the terminal inputs the sample wind speeds in the sample wind speed group into the wind power prediction model one by one, the predicted wind power group corresponding to the sample wind speed group and the wind power prediction model parameter value group corresponding to the sample wind speed group will be obtained. Similarly, for each wind speed fluctuation process, the above operations are performed to obtain the predicted wind power group corresponding to the sample wind speed group of each wind speed fluctuation process, and the wind power prediction model parameter value group corresponding to the sample wind speed group of each wind speed fluctuation process.

[0109] Step S702, according to the model parameter value group, the predicted wind power group and the slime mold algorithm, determine the optimal model parameter value.

[0110] In this embodiment, the terminal inputs the model parameter value group and the predicted wind power group into the slime mold algorithm. The slime mold algorithm iteratively optimizes in the predicted wind power group to determine the optimal predicted wind power. According to the optimal predicted wind power, the sample wind speed corresponding to the optimal predicted wind power and the model parameter value corresponding to the sample wind speed are determined. Then, the parameter value corresponding to the sample wind speed is the optimal model parameter value. The slime mold algorithm is an optimized slime mold algorithm, and the slime mold iteration times of the slime mold algorithm are pre-stored in the terminal. As the iteration times increase, the iteration parameters in the slime mold algorithm are adaptively adjusted, so as to ensure that each iterative optimization is not limited to the local optimal solution. The iteration parameter in the traditional slime mold algorithm is the function value of a linear decreasing function from 1 to 0. However, the predicted wind power value has high-dimensional non-linear characteristics. To meet the high-dimensional non-linear characteristics of the predicted wind power value, the terminal first establishes a normal cloud data set from 1 to 0 through the adaptive normal cloud model, ensuring that the total amount of data in the data set is the same as the number of times that need to be iterated in the slime mold algorithm. At the same time, during the iterative operation of the slime mold algorithm, the iteration parameters can obtain the parameter values each time through the normal cloud data set, thereby optimizing the slime mold algorithm and ensuring that the slime mold algorithm does not fall into the local optimal solution as much as possible. Similarly, for each wind speed fluctuation process, the above operations are performed to determine the optimal model parameter values corresponding to each wind speed fluctuation process.

[0111] Step S703: Determine the function value of the evaluation function according to the predicted wind power group corresponding to the sample wind speed group, the sample wind power group corresponding to the sample wind speed group, and the evaluation function.

[0112] In this embodiment, the terminal substitutes the data of the predicted wind power group corresponding to the sample wind speed group and the sample wind power group corresponding to the sample wind speed group into the evaluation function one by one according to the corresponding relationship (that is, one predicted wind power corresponds to one sample wind power), and finally obtains the function value of the evaluation function corresponding to this iteration. The evaluation functions are the root mean square error for evaluating the accuracy of the wind power prediction model and the standard deviation for evaluating the robustness of the wind power prediction model. The specific evaluation function formulas are as follows:

[0113]

[0114] In the formula: RMSE is the root mean square error, n is the total number of samples in the predicted wind power group or the sample wind power group, P i is the sample wind power in the sample wind power group, and P ipre is the predicted wind power in the predicted wind power group.

[0115]

[0116] In the formula: SD is the standard deviation, is the average error between the sample wind power in the sample wind power group and the predicted wind power in the predicted wind power group, n is the total number of samples in the predicted wind power group or the sample wind power group, and P i is the sample wind power in the sample wind power group, and P ipre is the predicted wind power in the predicted wind power group.

[0117] Similarly, for each wind speed fluctuation process, the above operations are performed to determine the function value of the evaluation function corresponding to each wind speed fluctuation process.

[0118] Step S704, when the preset iteration stop condition is not satisfied, record the function value of the evaluation function and the optimal model parameter value, and return to execute the step of determining the predicted wind power group corresponding to the sample wind speed group and the model parameter value group corresponding to the sample wind speed group according to the sample wind speed group of the wind speed fluctuation process and the initial wind power prediction model, until the preset iteration stop condition is satisfied, and determine the function value of each loss function and the optimal model parameter value corresponding to the function value of each loss function.

[0119] In this embodiment, the preset iteration stop condition includes reaching the iteration number, or the deviation between the function values of the evaluation function obtained by two iterations of steps S501 - S703 is less than the preset value. The terminal pre - stores the training iteration number. When the preset iteration stop condition is not satisfied, the terminal records the function value of the loss function and the optimal model parameter value obtained in this iteration, and returns to execute step S501. Until the iteration number is satisfied, or the deviation value between the function values of the loss function obtained by two iterations is less than the preset value, determine the function value of each loss function obtained after each iteration operation, and the optimal model parameter value corresponding to the function value of each loss function. Similarly, for each wind speed fluctuation process, the above operations are performed to determine the function value of each loss function corresponding to each wind speed fluctuation process and the optimal model parameter value corresponding to the function value of each loss function.

[0120] Step S705, according to the function value of each loss function and the optimization algorithm, determine the function value of the target loss function and the target optimal model parameter value corresponding to the function value of the target loss function.

[0121] In this embodiment, the optimization algorithm can be the Pareto selection algorithm. The terminal inputs the function values of each loss function into the optimization algorithm to obtain the function value of the overall satisfaction function corresponding to the function value of each loss function. The terminal selects the largest function value among the function values of the overall satisfaction function, determines the function value of the loss function corresponding to the function value of the overall satisfaction function as the function value of the target loss function, and determines the optimal model parameter value corresponding to the function value of the target loss function as the target optimal model parameter value, where the satisfaction function formula is:

[0122]

[0123]

[0124]

[0125] In the above formula, f i1 is the root mean square error satisfaction function of this iterative operation, f i2 is the standard deviation satisfaction function of this iterative operation, f i is the overall satisfaction function of this iterative operation, RMSE max is the maximum value of the root mean square error in all iterative operations, RMSE min The minimum value of the root mean square error in all iterative operations, RMSE i The root mean square error of this iterative operation, SD max The maximum value of the standard deviation in all iterative operations, SD min The minimum value of the standard deviation in all iterative operations, SD i is the standard deviation of this iterative operation.

[0126] Similarly, for each wind speed fluctuation process, the above operations are all executed to determine the target optimal model parameter values corresponding to each wind speed fluctuation process.

[0127] Step S706, based on the target optimal model parameter values, obtain the wind power prediction model for the wind speed fluctuation process.

[0128] In this embodiment, the terminal determines the wind power prediction model corresponding to the target optimal model parameter value through iterative operations, and this wind power prediction model corresponding to the target optimal model parameter value is the wind power prediction model for this wind speed fluctuation process. Similarly, for each wind speed fluctuation process, the above operations are all executed to determine the wind power prediction models corresponding to each wind speed fluctuation process.

[0129] Based on the above solution, by training the wind power prediction model corresponding to the wind speed fluctuation process and according to the dual-objective optimization strategy, the wind power prediction models corresponding to each wind speed fluctuation process are obtained, improving the universality and accuracy of the wind power prediction model.

[0130] In one embodiment, as Figure 8 shown, a method for predicting wind power is provided, and this method includes the following steps:

[0131] Step S801, obtain the target meteorological data set, and the target meteorological data set is the target wind speed at each moment within a preset time period.

[0132] In this embodiment, the terminal obtains the target wind speeds at each moment within a preset time period to form a target wind speed data set (i.e., the target meteorological data set). The specific processing procedure of this step can refer to the relevant explanation of step S101 above and will not be elaborated here.

[0133] Step S802: Divide each target wind speed according to the wind speed fluctuation division model to obtain each target wind speed group corresponding to each wind speed fluctuation process.

[0134] In this embodiment, the terminal selects each valid target wind speed that meets the preset screening conditions from the obtained target wind speeds; according to each valid target wind speed and the wind speed fluctuation division model, it determines each target wind speed group corresponding to each wind speed fluctuation process. The specific processing procedure of this step can refer to the relevant explanation of step S102 above and will not be elaborated here.

[0135] Step S803: For each wind speed fluctuation process, input the target wind speed group of the wind speed fluctuation process into the wind power prediction model corresponding to the wind speed fluctuation process to determine the target predicted wind power group corresponding to the fluctuation process.

[0136] In this embodiment, the terminal determines the wind power prediction model corresponding to each wind speed fluctuation process, inputs the target wind speed group of the wind speed fluctuation process into the wind power prediction model, and obtains the target predicted wind power group corresponding to the fluctuation process.

[0137] Step S804: Arrange each target predicted wind power according to the time series of the target wind speed corresponding to the target predicted wind power to obtain the predicted wind power.

[0138] Among them, the wind power prediction model is trained by any of the above wind power prediction model training methods.

[0139] In this embodiment, the terminal sorts the target predicted wind power groups corresponding to each wind speed fluctuation process obtained according to the time series of the target wind speed corresponding to the predicted wind power in the target predicted wind power group to obtain a wind power data set with the same time order as the target wind speed data set where the target wind speed is located, and determines that this wind power data set forms the predicted wind power.

[0140] Based on the above solution, by dividing the target wind speed data set to form each target wind speed group corresponding to each wind speed fluctuation process, inputting each target wind speed group corresponding to each wind speed fluctuation process into the wind power prediction model corresponding to the wind speed fluctuation process, obtaining each target predicted wind power, and sorting according to the time series of each target wind speed corresponding to each target predicted wind power, the final predicted wind power is obtained, improving the accuracy of predicting the abundant electric power.

[0141] To verify the effects of the above-mentioned various solutions, this application verifies the effects of the solutions through experiments, and the specific situation is as follows:

[0142] Taking the measured power of a wind farm in western China in 2020 and the corresponding numerical weather forecast data as the data set. The rated power of this wind farm is 54 MW. Considering the seasonal distribution law of wind power generation, the data of the first 3 months of each season are selected as the training set, and the data of the next month are used as the test set.

[0143] Table 1 Prediction correction values in different power intervals

[0144]

[0145] In the above table, ELM is a traditional wind speed prediction model, MFFP is a model for dividing fluctuations without considering thresholds, OMFFP is a model for dividing fluctuations considering thresholds, MAE is the mean absolute error between the predicted wind power and the sample wind power, and RMSE is the root mean square error between the predicted wind power and the sample wind power.

[0146] On the test set, the prediction results are shown in Table 1. The prediction accuracy of the model for dividing fluctuations considering thresholds is higher than that of the model for dividing fluctuations without considering thresholds. This shows that the method of dividing thresholds considering fluctuations improves the prediction accuracy.

[0147] Table 2 Comparison results of prediction models

[0148]

[0149]

[0150] In the above table, OMFFP is a model for dividing fluctuations considering thresholds, CMOSMA is a slime mold algorithm based on a normal cloud model, MOSMA is a double-objective slime mold algorithm, SMA is a single-objective slime mold algorithm, NSGAII is a multi-objective genetic optimization algorithm, ELM is a traditional wind speed prediction model, MAE is the mean absolute error between the predicted wind power and the sample wind power, RMSE is the root mean square error between the predicted wind power and the sample wind power, and SD is the standard deviation between the predicted wind power and the sample wind power.

[0151] On the test set, on the basis of considering thresholds for data fluctuation division, a double-objective prediction method is used for wind power prediction. The prediction results are shown in Table 2. The solution described in this application improves the prediction accuracy while ensuring the robustness of the prediction system. The robustness comparison of the prediction system is as Figure 2 shown.

[0152] This application also provides a training example of a wind power prediction model, such as Figure 9As shown in the figure, the specific processing process includes the following steps:

[0153] Step S901: Obtain a sample meteorological data set, which includes sample wind speeds and corresponding sample wind power at each moment within a preset time period.

[0154] Step S902: Among the sample wind speeds, select each valid sample wind speed that meets the preset screening conditions.

[0155] Step S903: Determine each peak in the wind speed sequence formed by the valid sample wind speeds, and the corresponding sample intervals of each peak in the wind speed sequence.

[0156] Step S904: Divide the peak distribution sequence according to the sample ratio corresponding to each wind speed fluctuation process to obtain the wind speed threshold corresponding to each wind speed fluctuation process; the peak distribution sequence is obtained by sorting the number of peaks within each preset wind speed interval in ascending order of wind speed.

[0157] Step S905: Determine the wind speed range corresponding to each wind speed fluctuation process according to the wind speed threshold corresponding to each wind speed fluctuation process.

[0158] Step S906: For each wind speed fluctuation process, determine the peaks within the wind speed range corresponding to the wind speed fluctuation process, and use the valid sample wind speeds within the sample intervals corresponding to the peaks as the sample wind speed group of the wind speed fluctuation process.

[0159] Step S907: For each wind speed fluctuation process, determine the predicted wind power group corresponding to the sample wind speed group and the model parameter value group corresponding to the sample wind speed group according to the sample wind speed group of the wind speed fluctuation process and the initial wind power prediction model.

[0160] Step S908: Determine the optimal model parameter value according to the model parameter value group, the predicted wind power group, and the slime mold algorithm.

[0161] Step S909: Determine the function value of the evaluation function according to the predicted wind power group corresponding to the sample wind speed group, the sample wind power group corresponding to the sample wind speed group, and the evaluation function.

[0162] Step S910: Determine whether the preset iteration stop condition is satisfied, and record the function value of the evaluation function and the optimal model parameter value.

[0163] If yes, execute step S911; if no, execute step S907.

[0164] Step S911: Determine the function value of the target loss function and the target optimal model parameter value corresponding to the function value of the target loss function according to the function values of the loss functions and the optimization algorithm.

[0165] Step S912: Based on the target optimal model parameter values, obtain a wind power prediction model for the wind speed fluctuation process.

[0166] It should be understood that although the steps in the flowcharts involved in the above embodiments are sequentially shown according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least some of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least some of the steps or stages in other steps or other steps.

[0167] Based on the same inventive concept, an embodiment of the present application further provides a training device for a wind power prediction model for implementing the training method of the wind power prediction model involved above. The solution provided by this device to solve the problem is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the training device for the wind power prediction model provided below can refer to the limitations on the training method of the wind power prediction model in the above text, and will not be repeated here.

[0168] In one embodiment, as Figure 10 shown, a training device for a wind power prediction model is provided, including: an acquisition module 1010, a division module 1020, and a training module 1030, where:

[0169] The acquisition module 1010 is configured to acquire a sample meteorological data set, which includes sample wind speeds and sample wind power corresponding to the sample wind speeds at each moment within a preset time period.

[0170] The division module 1020 is configured to divide each sample wind speed according to the wind speed fluctuation division model to obtain a sample wind speed group corresponding to each wind speed fluctuation process.

[0171] The training module 1030 is configured to, for each wind speed fluctuation process, train the initial wind power prediction model according to the sample wind speed group corresponding to the wind speed fluctuation process and the sample wind power group corresponding to the sample wind speed group, to obtain a wind power prediction model corresponding to the wind speed fluctuation process.

[0172] Optionally, the division module 1020 is specifically configured to:

[0173] Among the sample wind speeds, select each valid sample wind speed that meets the preset screening conditions.

[0174] According to the effective sample wind speeds and the wind speed fluctuation division model, determine the sample wind speed groups corresponding to each wind speed fluctuation process.

[0175] Optionally, the wind speed fluctuation division model includes the sample ratios corresponding to each wind speed fluctuation process; the division module 1020 is specifically configured to:

[0176] Determine each peak value in the wind speed sequence formed by the effective sample wind speeds, and the sample intervals corresponding to each peak value in the wind speed sequence.

[0177] According to the sample ratios corresponding to each wind speed fluctuation process, divide the peak value distribution sequence to obtain the wind speed thresholds corresponding to each wind speed fluctuation process; the peak value distribution sequence is obtained by sorting the number of peak values within each preset wind speed interval in ascending order of wind speed.

[0178] According to the wind speed thresholds corresponding to each wind speed fluctuation process, determine the wind speed ranges corresponding to each wind speed fluctuation process.

[0179] For each wind speed fluctuation process, determine the peak values within the wind speed range corresponding to the wind speed fluctuation process, and use the effective sample wind speeds within the sample intervals corresponding to the peak values as the sample wind speed groups of the wind speed fluctuation process.

[0180] Optionally, the training module 1030 is specifically configured to:

[0181] For each wind speed fluctuation process, according to the sample wind speed group of the wind speed fluctuation process and the initial wind power prediction model, determine the predicted wind power group corresponding to the sample wind speed group and the model parameter value group corresponding to the sample wind speed group.

[0182] According to the model parameter value group, the predicted wind power group, and the slime mold algorithm, determine the optimal model parameter values.

[0183] According to the predicted wind power group corresponding to the sample wind speed group, the sample wind power group corresponding to the sample wind speed group, and the evaluation function, determine the function value of the evaluation function.

[0184] When the preset iteration stop condition is not satisfied, record the function value of the evaluation function and the optimal model parameter values, and return to execute the step of determining the predicted wind power group corresponding to the sample wind speed group and the model parameter value group corresponding to the sample wind speed group according to the sample wind speed group of the wind speed fluctuation process and the initial wind power prediction model, until the preset iteration stop condition is satisfied, and determine the function values of each loss function and the optimal model parameter values corresponding to the function values of each loss function.

[0185] According to the function values of each loss function and the optimization algorithm, determine the function value of the target loss function and the target optimal model parameter values corresponding to the function value of the target loss function.

[0186] Based on the target optimal model parameter values, a wind power prediction model for the wind speed fluctuation process is obtained.

[0187] Based on the same inventive concept, an embodiment of the present application further provides a wind power prediction device for implementing the above-mentioned wind power prediction method. The solution provided by this device for solving problems is similar to the solution described in the above method. Therefore, the specific limitations in one or more of the following embodiments of the wind power prediction device can refer to the limitations on the wind power prediction method in the above text, and will not be repeated here.

[0188] In one embodiment, as Figure 11 shown, a wind power prediction device is provided, including: an acquisition module 1110, a division module 1120, a determination module 1130, and a sorting module 1140, where:

[0189] The acquisition module 1110 is configured to acquire a target meteorological data set, and the target meteorological data set is the target wind speed at each moment within a preset time period.

[0190] The division module 1120 is configured to divide each target wind speed according to the wind speed fluctuation division model to obtain each target wind speed group corresponding to each wind speed fluctuation process.

[0191] The determination module 1130 is configured to, for each wind speed fluctuation process, input the target wind speed group of the wind speed fluctuation process into the wind power prediction model corresponding to the wind speed fluctuation process, and determine the target predicted wind power group corresponding to the fluctuation process.

[0192] The sorting module 1140 is configured to arrange each target predicted wind power according to the time series of the target wind speed corresponding to the target predicted wind power to obtain the predicted wind power.

[0193] Wherein, the wind power prediction model is trained by the training method of the wind power prediction model of any one of the foregoing.

[0194] Each module in the above-mentioned wind power prediction model training device and wind power prediction device can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0195] In one embodiment, a computer device is provided. The computer device can be a terminal, and its internal structure diagram can be as Figure 12As shown in the figure. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be implemented through WIFI, a mobile cellular network, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for training a wind power prediction model and a method for predicting wind power. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covered on the display screen, or buttons, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0196] Those skilled in the art can understand that Figure 12 the structure shown in the figure is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have a different component layout.

[0197] In an embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, the steps in the above method embodiments are implemented.

[0198] In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor, the steps in the above method embodiments are implemented.

[0199] In an embodiment, a computer program product is provided, including a computer program. When the computer program is executed by the processor, the steps in the above method embodiments are implemented.

[0200] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or fully authorized by all parties.

[0201] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0202] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered to be within the scope described in this specification.

[0203] The above embodiments only represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A training method for a wind power prediction model, characterized in that, The method includes: Obtaining a sample meteorological data set, where the sample meteorological data set includes sample wind speeds at each moment within a preset time period and sample wind power corresponding to the sample wind speeds; Dividing each of the sample wind speeds according to a wind speed fluctuation division model to obtain sample wind speed groups corresponding to respective wind speed fluctuation processes; For each wind speed fluctuation process, determining a predicted wind power group corresponding to the sample wind speed group and a model parameter value group corresponding to the sample wind speed group according to the sample wind speed group of the wind speed fluctuation process and an initial wind power prediction model; Determining an optimal model parameter value according to the model parameter value group, the predicted wind power group, and a slime mold algorithm; Determining a function value of the evaluation function according to the predicted wind power group corresponding to the sample wind speed group, the sample wind power group corresponding to the sample wind speed group, and the evaluation function; When a preset iteration stop condition is not satisfied, recording the function value of the evaluation function and the optimal model parameter value, and returning to execute the step of determining the predicted wind power group corresponding to the sample wind speed group and the model parameter value group corresponding to the sample wind speed group according to the sample wind speed group of the wind speed fluctuation process and the initial wind power prediction model, until the preset iteration stop condition is satisfied, and determining function values of respective loss functions and optimal model parameter values corresponding to the function values of the respective loss functions; Determining a function value of a target loss function and a target optimal model parameter value corresponding to the function value of the target loss function according to the function values of the respective loss functions and a preference algorithm; Based on the target optimal model parameter value, obtaining a wind power prediction model for the wind speed fluctuation process.

2. The method according to claim 1, characterized in that, The step of dividing each of the sample wind speeds according to a wind speed fluctuation division model to obtain sample wind speed groups corresponding to respective wind speed fluctuation processes includes: Selecting respective valid sample wind speeds that meet a preset screening condition from among the sample wind speeds; Determining sample wind speed groups corresponding to respective wind speed fluctuation processes according to the valid sample wind speeds and the wind speed fluctuation division model.

3. The method according to claim 2, wherein The wind speed fluctuation division model includes sample ratios corresponding to respective wind speed fluctuation processes; the step of determining sample wind speed groups corresponding to respective wind speed fluctuation processes according to the valid sample wind speeds and the wind speed fluctuation division model includes: Determining respective peaks in a wind speed sequence formed by the valid sample wind speeds and sample intervals corresponding to the peaks in the wind speed sequence; Dividing a peak distribution sequence according to the sample ratios corresponding to respective wind speed fluctuation processes to obtain wind speed thresholds corresponding to respective wind speed fluctuation processes; the peak distribution sequence is obtained by sorting the numbers of peaks within respective preset wind speed intervals in ascending order of wind speed; Determining wind speed ranges corresponding to respective wind speed fluctuation processes according to the wind speed thresholds corresponding to respective wind speed fluctuation processes; For each wind speed fluctuation process, determining peaks within the wind speed range corresponding to the wind speed fluctuation process, and using the valid sample wind speeds within the sample intervals corresponding to the peaks as the sample wind speed group of the wind speed fluctuation process.

4. A method for predicting wind power, characterized in that, The method includes: Obtain a target meteorological data set, where the target meteorological data set is the target wind speed at each moment within a preset time period; Divide each of the target wind speeds according to a wind speed fluctuation division model to obtain a target wind speed group corresponding to each wind speed fluctuation process; For each wind speed fluctuation process, input the target wind speed group of the wind speed fluctuation process into the wind power prediction model corresponding to the wind speed fluctuation process to determine a target predicted wind power group corresponding to the fluctuation process; Arrange each of the target predicted wind powers according to the time series of the target wind speed corresponding to the target predicted wind power to obtain the predicted wind power; Among them, the wind power prediction model is trained by the training method of the wind power prediction model according to any one of claims 1 to 3.

5. A training device for a wind power prediction model, characterized in that, The device includes: An acquisition module for acquiring a sample meteorological data set, where the sample meteorological data set includes sample wind speeds at each moment within a preset time period and sample wind powers corresponding to the sample wind speeds; A division module for dividing each of the sample wind speeds according to a wind speed fluctuation division model to obtain a sample wind speed group corresponding to each wind speed fluctuation process; A training module for, for each wind speed fluctuation process, determining a predicted wind power group corresponding to the sample wind speed group and a model parameter value group corresponding to the sample wind speed group according to the sample wind speed group of the wind speed fluctuation process and an initial wind power prediction model; determining an optimal model parameter value according to the model parameter value group, the predicted wind power group and the slime mold algorithm; determining a function value of the evaluation function according to the predicted wind power group corresponding to the sample wind speed group and the sample wind power group corresponding to the sample wind speed group; in the case of not meeting the preset iteration stop condition, recording the function value of the evaluation function and the optimal model parameter value, and returning to execute the step of determining a predicted wind power group corresponding to the sample wind speed group and a model parameter value group corresponding to the sample wind speed group according to the sample wind speed group of the wind speed fluctuation process and the initial wind power prediction model until the preset iteration stop condition is met, determining function values of each loss function and optimal model parameter values corresponding to the function values of each loss function; determining a function value of a target loss function and a target optimal model parameter value corresponding to the function value of the target loss function according to the function values of each loss function and a preference algorithm; obtaining the wind power prediction model of the wind speed fluctuation process based on the target optimal model parameter value.

6. The device according to claim 5, wherein, The division module is specifically used for: Select each valid sample wind speed that meets a preset screening condition from each of the sample wind speeds; Determine a sample wind speed group corresponding to each wind speed fluctuation process according to each of the valid sample wind speeds and the wind speed fluctuation division model.

7. A wind power prediction device, characterized in that, The device includes: An acquisition module for acquiring a target meteorological data set, where the target meteorological data set is the target wind speed at each moment within a preset time period; A division module for dividing each of the target wind speeds according to a wind speed fluctuation division model to obtain each target wind speed group corresponding to each wind speed fluctuation process; A determination module, configured to input a target wind speed group of each wind speed fluctuation process into a wind power prediction model corresponding to the wind speed fluctuation process, and determine a target predicted wind power group corresponding to the fluctuation process; A sorting module, configured to sort each of the target predicted wind powers according to a time series of target wind speeds corresponding to the target predicted wind powers, to obtain predicted wind powers; Wherein, the wind power prediction model is trained by the training method of the wind power prediction model according to any one of claims 1 to 3.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 3 or claim 4 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 3 or claim 4 are implemented.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 3 or claim 4 are implemented.

Images