Wind power prediction method and system, electronic equipment and medium

Abstract

The invention relates to the technical field of wind power generation, in particular to a wind power prediction method and system, electronic equipment and a medium, and the method comprises the steps: obtaining influence parameters which influence the wind power of a wind driven generator in a historical time period, and building the matching data of the influence parameters and the corresponding wind power; removing abnormal data in the matched data, and determining a target data set; using the target data set to train an SVM model, and determining a wind power prediction model; and predicting the wind power of the wind driven generator through the wind power prediction model. According to the method, before model training, abnormal data removal is carried out on the training data in advance, so that the influence of the abnormal data on model training is reduced, and the modeling precision of the model is improved.

Description

Background Art

[0002] With the increasing severity of global climate change and the pursuit of sustainable development, the proportion of renewable energy in the global energy structure continues to increase. As a clean, renewable and widely distributed form of energy, wind energy has developed rapidly in recent years and has become an important part of the power supply in many countries. However, due to the natural volatility and unpredictability of wind speed, the output power of wind farms is highly uncertain, which poses a huge challenge to grid dispatching. Accurate and reliable wind power forecasting can not only improve the operating efficiency of the power system, but also effectively reduce wind abandonment and enhance grid stability.

[0003] In the prior art, an artificial intelligence method is used to establish a prediction model, and the prediction model is used to predict the output power of a wind farm. This method can better capture complex nonlinear relationships and spatiotemporal changes. However, during model training, the training data mainly comes from the SCADA system (Supervisory Control and Data Acquisition). Due to sensor anomalies, signal interference in the data transmission process, power grid wind curtailment and power rationing, etc., there are a large number of abnormal data in the training data collected by SCADA.

[0004] In summary, the training data collected by the SCADA system usually contains a large amount of abnormal data, which will directly affect the modeling accuracy of the prediction model and have a certain impact on wind power prediction. Summary of the invention

[0005] In order to overcome the problem that the training data collected by the SCADA system usually contains a large amount of abnormal data, which directly affects the modeling accuracy of the prediction model and has a certain impact on wind power prediction, the present invention provides a wind power prediction method, system, electronic equipment and medium.

[0006] In a first aspect, in order to solve the above technical problems, the present invention provides a wind power prediction method, comprising:

[0007] Obtain influencing parameters that affect the wind power of wind turbines in a historical period, and establish matching data between the influencing parameters and the corresponding wind power;

[0008] Eliminate abnormal data from the matching data and determine the target data set;

[0009] Use the target data set to train the SVM model and determine the wind power prediction model;

[0010] The wind power prediction model is used to predict the wind power of wind turbines.

[0011] In a second aspect, the present invention provides a wind power prediction system, comprising:

[0012] A matching data acquisition module is used to acquire influencing parameters affecting the wind power of the wind turbine in a historical period of time, and to establish matching data between the influencing parameters and the corresponding wind power;

[0013] A target data set determination module is used to remove abnormal data from the matching data and determine the target data set;

[0014] A wind power prediction model determination module is used to train an SVM model using a target data set to determine a wind power prediction model;

[0015] The prediction module is used to predict the wind power of the wind turbine generator set through a wind power prediction model.

[0016] In a third aspect, the present invention provides a computing device, including a memory, a processor, and a program stored in the memory and running on the processor, and when the processor executes the program, the steps of a wind power prediction method as described above are implemented.

[0017] In a fourth aspect, the present invention provides a computer-readable storage medium, in which instructions are stored. When the instructions are executed on a terminal device, the terminal device executes the steps of a wind power prediction method as described above.

[0018] The beneficial effects of the present invention are: obtaining influencing parameters that affect the wind power of a wind turbine generator, thereby establishing matching data between the influencing parameters and the corresponding wind power, and removing abnormal data from the matching data before model training, thereby training the SVM model with a target data set that removes abnormal data to obtain a wind power prediction model, and the wind power prediction model can be used to predict wind power. Before model training, the present application removes abnormal data from the training data in advance, thereby reducing the impact of abnormal data on model training and improving the modeling accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0020] Figure 1 A schematic diagram of a flow chart of a wind power prediction method according to an embodiment of the present invention;

[0021] Figure 2 This is a wind speed-wind power matching data diagram in an embodiment of the present invention;

[0022] Figure 3 It is a sorted series diagram of wind speed-wind power in an embodiment of the present invention;

[0023] Figure 4 This is a sorted sequence diagram after removing abnormal data in an embodiment of the present invention;

[0024] Figure 5 A schematic diagram of a flow chart of a wind power prediction method according to another embodiment of the present invention;

[0025] Figure 6 Iteratively updates the graph with the model's fitness value;

[0026] Figure 7 This is a comparison chart of the prediction results of the training set;

[0027] Figure 8 Comparison chart of predicted values ​​and true values ​​of the training set;

[0028] Fig. 9 This is a comparison chart of the prediction results for the test set;

[0029] Fig.10 Comparison chart of predicted values ​​and true values ​​of the test set;

[0030] Fig.11 The present invention is a schematic diagram of the structure of a wind power prediction system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The following examples are provided to further explain and supplement the present invention and do not constitute any limitation to the present invention.

[0032] A wind power prediction method, system, electronic device and medium according to an embodiment of the present invention are described below with reference to the accompanying drawings.

[0033] like Figure 1 As shown, an embodiment of the present invention provides a wind power prediction method, comprising:

[0034] S1. Obtain influencing parameters that affect the wind power of wind turbines in a historical period, and establish matching data between the influencing parameters and the corresponding wind power.

[0035] S2. Eliminate abnormal data in the matching data and determine the target data set.

[0036] S3. Use the target data set to train the SVM model and determine the wind power prediction model.

[0037] S4. Predict the wind power of the wind turbine using a wind power prediction model.

[0038] In this embodiment, the influencing parameters affecting the wind power of the wind turbine are obtained, so as to establish matching data between the influencing parameters and the corresponding wind power. Before model training, the abnormal data in the matching data is removed, so as to train the SVM model with the target data set from which the abnormal data is removed, and a wind power prediction model is obtained, and the wind power prediction model can be used to predict the wind power. In this embodiment, before model training, the training data is pre-trained to remove abnormal data, so as to reduce the influence of abnormal data on model training and improve the modeling accuracy of the model.

[0039] The SVM model in this embodiment is a regression prediction model of a support vector machine (SVM), which is a supervised learning model widely used in classification and regression analysis. The SVM model is a two-class classifier, and its basic model is defined as a linear classifier with the largest interval in the feature space. Its learning strategy is to maximize the interval, which can eventually be transformed into a solution to a convex quadratic programming problem. In short, the goal of the SVM model is to find a hyperplane so that the two types of data are as far away from the hyperplane as possible, so that the model can classify new data more accurately, that is, the classifier is more stable.

[0040] Optionally, the influencing parameters include unit performance parameters and environmental factors, wherein the unit performance parameters include the rotation speed and blade angle of the wind turbine, and the environmental factors include wind speed and wind direction.

[0041] The selection indexes of the influencing parameters in this embodiment are as follows:

[0042] Relevance: The selected influencing parameters should have a strong correlation with the target variable. The influencing parameters can provide important information about the target variable and help improve the predictive ability of the model.

[0043] Independence: Try to choose influencing parameters that are independent of each other. If there is a high correlation between multiple influencing parameters, it may lead to multicollinearity problems, affecting the stability and interpretability of the model.

[0044] Importance: Based on domain knowledge and experience, give priority to those influencing parameters that have a greater impact on the target variable. This can be determined through expert review, literature review, etc.

[0045] Data quality: Ensure that the data of the selected influencing parameters are accurate and complete. Influencing parameters with too many missing values ​​or abnormal data may affect the effect of model training.

[0046] Dimension simplification: Considering the model complexity and computational efficiency, we should avoid selecting too many influential parameters. We can reduce the feature dimension through methods such as feature selection.

[0047] In summary, in this embodiment, influencing parameters such as wind speed, rotation speed, pitch angle, wind direction, and wind power are selected, and wind speed, rotation speed, pitch angle, and wind direction are used as inputs of the model, and wind power is used as output.

[0048] In this embodiment, wind speed and wind direction are the core parameters that determine wind power generation. They not only directly affect the rotation speed of the wind turbine blades, but also are the basis for calculating wind energy conversion efficiency. Wind speed and wind direction are usually obtained through sensors installed on a wind tower. Considering the impact of terrain and altitude on wind speed, measurements can be performed at different altitudes.

[0049] In this embodiment, the rotation speed of the wind turbine is one of the important indicators for evaluating its operating status. By monitoring the rotation speed, abnormal conditions can be discovered in time, and measures can be taken to adjust it to the optimal operating point.

[0050] In this embodiment, the blade angle refers to the inclination angle of the wind turbine blade relative to the wind direction. When the blade angle increases, the contact area between the blade and the wind decreases, resulting in a decrease in wind energy captured by the wind turbine; conversely, the captured wind energy increases.

[0051] In this embodiment, both the unit performance parameters and environmental factors are decisive factors affecting the wind power of the wind turbine. By collecting the unit performance parameters and environmental factors, it can help users accurately predict the wind power.

[0052] For example, using wind speed as an example, we can establish matching data between wind speed and wind power. This matching data can be displayed in the form of graphics or in the form of a sample table, such as Figure 2 As shown in the figure, based on GB / T 18451.2-2012 "Test of Power Characteristics of Wind Turbine Generator Sets", it can be known that wind speed has the greatest correlation with wind power. By obtaining wind speed and corresponding wind power, a wind turbine generator set power curve can be obtained, where the horizontal axis represents wind speed and the vertical axis represents wind power output. Through this curve, the power generation capacity of a wind turbine generator set under different wind speed conditions can be intuitively understood.

[0053] Optionally, removing abnormal data from the matching data and determining the target data set includes:

[0054] Filter outliers in matching data and delete them;

[0055] The deleted outliers are filled using the first preset values ​​to determine the target data set.

[0056] In this embodiment, the existence of outliers in the data may affect the calculation accuracy to a certain extent. The filloutliers function in the MATLAB software can be used to smooth the data and detect and remove outliers. Selecting valid data can greatly improve the prediction accuracy.

[0057] In addition, in this embodiment, abnormal data can be further cleared based on the following rules, the rules are as follows:

[0058] a) Other external conditions except wind speed exceed the operating range of the wind turbine;

[0059] b) Wind turbine generator set failure causes wind turbine generator set shutdown;

[0060] c) The wind turbine is manually shut down, in test or maintenance mode;

[0061] d) failure or degradation of measuring instruments (e.g. caused by icing);

[0062] e) The wind direction is outside the measurement sector specified in the regulations;

[0063] f) The wind direction is outside the effective sector of the on-site calibration;

[0064] g) Any special atmospheric conditions screened during field calibration should also be screened during the power curve test.

[0065] Optionally, filter out outliers in the matching data and remove outliers, including:

[0066] Sort each matching data from small to large according to the size of the influencing parameter, and determine the sorting sequence of wind power;

[0067] Smooth the sorted series and filter out outliers.

[0068] For example, taking the matching data of wind speed and wind power as an example, the wind speed ranges from 3 to 25 (m·s -1 ) are first sorted from small to large according to the wind speed to obtain the sorted series of wind power. Since wind speed and wind power are proportional in theory, smoothing processing can be used to exclude abnormal points (outliers) and thus eliminate abnormal data.

[0069] Optionally, the sorted series is smoothed to filter out outliers, including:

[0070] Matching data in the sorted sequence that is greater than the third quartile by a second preset value is regarded as an outlier, or matching data in the sorted sequence that is less than the first quartile by a third preset value is regarded as an outlier.

[0071] In this embodiment, the quartiles are the values ​​at three split points when all data are arranged from small to large and divided into four equal parts. The values ​​at these three key split points are respectively called the first quartile (Q1, the smaller quartile), the second quartile (Q2, the median) and the third quartile (Q3, the larger quartile).

[0072] According to the total number of data n, use the formula to determine the position of the quartile. Commonly used formulas are:

[0073] Q1 position = (n+1) × 0.25

[0074] Q2's position = (n+1) × 0.5 (i.e. median)

[0075] Q3 position = (n+1) × 0.75

[0076] If n+1 is not an integer multiple of 4, then the quartile position will be a decimal. In this case, interpolation is required to determine the specific value of the quartile, for example, by interpolating the values ​​at the two integer positions adjacent to the decimal to calculate the value of the quartile.

[0077] Quartiles can clearly show the distribution of data, especially the central tendency and dispersion of the data.

[0078] In this embodiment, the second preset value and the third preset value can be set according to actual conditions. Preferably, in this embodiment, the second preset value is 1.5 third quartiles, that is, the matching data that is 1.5 third quartiles larger than the third quartile is regarded as an outlier, and the third preset value is 1.5 first quartiles, that is, the matching data that is 1.5 first quartiles smaller than the first quartile is regarded as an outlier.

[0079] Optionally, filling the deleted outlier with a first preset value includes:

[0080] The central value in the sorted sequence is used as the first preset value to fill the deleted outliers.

[0081] In this embodiment, outliers are interpolated and filled to maintain the smoothness and continuity of the data, making the predicted data more accurate.

[0082] For example, Figure 3 As shown in the figure, the sorted series (displayed in the form of a curve) obtained from the matching data of wind speed and wind power is smoothed to screen out discrete values ​​including discrete abnormal points, anemometer abnormalities, power limit abnormal points, and wind turbine shutdown points.

[0083] Deleting the above discrete values ​​and filling them with the central values, we can get Figure 4 As shown in the target data set, it can be seen that the target data set has good smoothness.

[0084] Optionally, using the target data set to train an SVM model to determine a wind power prediction model further includes:

[0085] The NGO algorithm is used to optimize the objective function of the SVM model and determine the target parameters of the SVM model.

[0086] In this embodiment, the NGO algorithm simulates the predation behavior of the northern goshawk, converts the optimization problem into the problem of finding the best prey position, and can control the movement position between the goshawk and the prey according to the fitness value, thereby achieving iterative optimization of the prey structure and the optimal solution to the global problem. Using this method, the final optimized SVM model and parameters (optimal values ​​of learning rate, hidden nodes, and regularization coefficient) can be obtained.

[0087] The NGO algorithm mainly includes the following steps:

[0088] Initialize the population: Randomly generate a set of candidate solutions representing different prey locations.

[0089] Exploration phase: The goshawk conducts random searches based on its own sensory information, looking for potential prey.

[0090] Predation stage: When the goshawk finds potential prey, it will adjust its flight path according to the speed and direction of the prey and eventually capture the prey.

[0091] Update the population: According to the predation results, update the individuals in the population and retain the better solutions to obtain the target parameters.

[0092] Optionally, another embodiment is used to illustrate a method for predicting wind power. Figure 5 As shown, specifically including:

[0093] The wind power data (including influencing parameters and wind power) in the historical time period are obtained, and a curve of influencing parameters and wind power is established, for example, a power curve of wind speed and power (wind power) is established.

[0094] Abnormal data is removed based on rules, and then outlier detection and data smoothing are used to remove abnormal data for a second time to obtain preprocessed data.

[0095] The preprocessed data is divided into a training set and a test set, and the training set is input into the SVM model for training.

[0096] During model training, the objective function fitness in the SVM model is calculated, and it is determined whether the objective function fitness meets the conditions. In this embodiment, the condition may be whether the target fitness reaches the target value. If the objective function fitness does not meet the conditions, the first N individuals are used as the initialization goshawk population, and enter the exploration phase (improved prey identification phase) and the predation phase (improved pursuit and escape phase), thereby calculating the new objective function fitness.

[0097] Determine whether the current objective function fitness has reached the target value. If the current objective function fitness has not reached the target value, then enter the iteration, such as Figure 6 As shown, it can be seen that with the increase in the number of iterations, the fitness value tends to be stable, and when the iteration reaches 50 times, the fitness value reaches the target value of 0.002, then the iteration is ended and the optimal parameters are obtained.

[0098] The SVM model with the optimal parameters is tested using the test set to obtain the final wind power prediction model.

[0099] In addition, during the training and testing periods, the RMSE (Root Mean Square Error) of the prediction results was verified. It is a commonly used indicator to measure the accuracy of the prediction model. It is obtained by averaging the square of the difference between the predicted value and the true value and then taking the square root. The smaller the RMSE value, the more accurate the prediction result of the model.

[0100] like Figure 7 As shown in the figure, it can be seen that when the number of training samples increases from 0 to 2000, the prediction results of the prediction samples are almost consistent with the true values, and the RMSE value is 98.394. Figure 8 As shown, it can be seen that the predicted value obtained based on the training set is almost consistent with the true value of the training set, indicating that after the abnormal data is removed, the prediction accuracy of the model is higher.

[0101] like Fig. 9 As shown in the figure, it can be seen that when the number of test samples increases from 0 to 2000, the prediction results of the predicted samples are almost consistent with the true values, and the RMSE value is 140.8997. Fig.10 As shown, it can be seen that the predicted value obtained based on the test set is almost consistent with the true value of the test set, indicating that after the abnormal data is removed, the prediction accuracy of the model is higher.

[0102] like Fig.11 As shown, the present invention provides a wind power prediction system, comprising:

[0103] A matching data acquisition module is used to acquire influencing parameters affecting the wind power of the wind turbine in a historical period of time, and to establish matching data between the influencing parameters and the corresponding wind power;

[0104] A target data set determination module is used to remove abnormal data from the matching data and determine the target data set;

[0105] A wind power prediction model determination module is used to train an SVM model using a target data set to determine a wind power prediction model;

[0106] The prediction module is used to predict the wind power of the wind turbine generator through a wind power prediction model.

[0107] Optionally, the target data set determination module is specifically used to:

[0108] Filter outliers in matching data and delete them;

[0109] The deleted outliers are filled using the first preset values ​​to determine the target data set.

[0110] Optionally, the target data set determination module is specifically used to:

[0111] Sort each matching data from small to large according to the size of the influencing parameter, and determine the sorting sequence of wind power;

[0112] Smooth the sorted series and filter out outliers.

[0113] Optionally, the target data set determination module is specifically used to:

[0114] Matching data in the sorted sequence that is greater than the third quartile by a second preset value is regarded as an outlier, or matching data in the sorted sequence that is less than the first quartile by a third preset value is regarded as an outlier.

[0115] Optionally, the target data set determination module is specifically used to:

[0116] The central value in the sorted sequence is used as the first preset value to fill the deleted outliers.

[0117] Optionally, the wind power prediction model determination module is further used to:

[0118] The NGO algorithm is used to optimize the objective function of the SVM model and determine the target parameters of the SVM model.

[0119] An embodiment of the present invention also provides a computing device, including a memory, a manager, and a program stored in the memory and running on the manager. When the manager executes the program, some or all steps of the above-mentioned wind power prediction method are implemented.

[0120] Among them, the computing device can be a computer, and correspondingly, its program is computer software. The above-mentioned parameters and steps in the computing device of the present invention can refer to the parameters and steps in the embodiment of a wind power prediction method mentioned above, and will not be repeated here.

[0121] Those skilled in the art know that the present invention can be implemented as a system, method or computer program product. Therefore, the present disclosure can be specifically implemented in the following forms, namely: it can be complete hardware, it can be complete software (including firmware, resident software, microcode, etc.), or it can be a combination of hardware and software, generally referred to as "circuit", "module" or "system" herein. In addition, in some embodiments, the present invention can also be implemented in the form of a computer program product in one or more computer-readable media, and the computer-readable medium contains computer-readable program code. Computer-readable storage media can be, for example, but not limited to - electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any combination of the above.

[0122] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means 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 may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0123] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A wind power prediction method, characterized in that: include: Acquire influencing parameters affecting the wind power of the wind turbine in a historical period, and establish matching data between the influencing parameters and the corresponding wind power; Eliminate abnormal data from the matching data to determine a target data set; Using the target data set to train an SVM model to determine a wind power prediction model; The wind power prediction model is used to predict the wind power of the wind turbine.

2. The method according to claim 1, characterized in that The influencing parameters include unit performance parameters and environmental factors, wherein the unit performance parameters include the rotation speed and blade angle of the wind turbine, and the environmental factors include wind speed and wind direction.

3. The method according to claim 2, characterized in that Eliminating abnormal data in the matching data and determining a target data set includes: Screening out outliers in the matching data and deleting the outliers; The deleted outliers are filled using the first preset values ​​to determine the target data set.

4. The method according to claim 3, characterized in that The step of screening out outliers in the matching data and deleting the outliers includes: Sort the matching data from small to large according to the magnitude of the influencing parameters to determine a sorted sequence of wind power; The sorted sequence is smoothed to filter out outliers.

5. The method according to claim 4, characterized in that Smoothing the sorted sequence to filter out outliers includes: Matching data in the sorted sequence that is greater than the third quartile by a second preset value is regarded as an outlier, or matching data in the sorted sequence that is less than the first quartile by a third preset value is regarded as an outlier.

6. The method according to claim 4, characterized in that The step of filling the deleted outlier with the first preset value comprises: The central value in the sorted sequence is used as the first preset value to fill the deleted outliers.

7. The method according to any one of claims 1 to 6, characterized in that: Using the target data set to train an SVM model to determine a wind power prediction model also includes: The NGO algorithm is used to optimize the objective function of the SVM model and determine the target parameters of the SVM model.

8. A wind power prediction system, characterized in that: include: A matching data acquisition module is used to acquire influencing parameters affecting the wind power of the wind turbine in a historical period of time, and to establish matching data between the influencing parameters and the corresponding wind power; A target data set determination module, used to remove abnormal data from the matching data and determine the target data set; A wind power prediction model determination module, used to train an SVM model using the target data set to determine a wind power prediction model; The prediction module is used to predict the wind power of the wind turbine generator by using the wind power prediction model.

9. A computing device comprising a memory, a processor, and a program stored in the memory and running on the processor, characterized in that: When the processor executes the program, the steps of a wind power prediction method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: Instructions are stored in the computer-readable storage medium. When the instructions are executed on the terminal device, the terminal device executes the steps of a wind power prediction method as described in any one of claims 1 to 7.

Images