Wind power prediction method and system based on machine learning

Abstract

The invention discloses a wind power prediction method and system based on machine learning, and the method comprises the steps: obtaining historical wind power data and historical gas phase data of a wind power station, carrying out the preprocessing of the data, determining a gas phase impact factor which affects the wind power from the historical gas phase data after the preprocessing, and carrying out the prediction of the wind power. Extracting a feature vector of each gas phase influence factor, constructing and training a wind power prediction sub-model based on the feature vector of each gas phase influence factor and a preset machine learning algorithm model, and performing prediction through each wind power prediction sub-model to obtain each wind power prediction value, and performing weighted averaging on each wind power prediction value to obtain a wind power prediction total value. According to the method, the wind power prediction sub-models are respectively constructed for different gas phase influence factors based on a machine learning mode to carry out real-time prediction, and the total wind power prediction value is calculated by combining the prediction values of the plurality of wind power prediction sub-models, so that the accuracy of wind power prediction of the wind power station is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation, and in particular to a wind power prediction method and system based on machine learning. Background Art

[0002] Wind power forecasting refers to the use of various meteorological data and historical power data of wind farms to predict the power generation of wind farms in a certain period of time in the future by establishing mathematical models or using machine learning algorithms. This prediction is very important for the operation and scheduling of wind farms, and can help achieve the effective utilization of wind power resources and the stable operation of the power grid.

[0003] However, traditional wind power prediction methods usually rely on the establishment of complex physical models to describe the relationship between wind farms and wind turbines. These models need to consider multiple factors such as wind turbine characteristics, meteorological conditions, and terrain. Therefore, the establishment and solution of the model are relatively complex. At the same time, traditional methods have limited processing capabilities for multi-variable relationships. Since wind power is affected by a variety of meteorological factors, traditional methods have certain limitations when dealing with complex relationships between multiple variables, and it is difficult to fully consider the impact of various factors on power. In addition, traditional methods often find it difficult to update models in a timely manner and adapt to rapidly changing meteorological conditions in wind farm power predictions with high real-time requirements. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a wind power prediction method and system based on machine learning.

[0005] A first aspect of the present invention provides a wind power prediction method based on machine learning, comprising:

[0006] Obtain the historical wind power data of the wind power station and the historical gas phase data corresponding to the time, and pre-process the historical wind power data and the historical gas phase data;

[0007] Analyze the pre-processed historical wind power data and historical gas phase data to determine the gas phase influencing factors that affect wind power in the historical gas phase data;

[0008] Extract features of each gas phase influencing factor to obtain a feature vector of each gas phase influencing factor;

[0009] Based on the characteristic vectors of each gas phase influencing factor and the preset machine learning algorithm model, the wind power prediction sub-models of the wind power station are respectively constructed, and each wind power prediction sub-model is trained and tested;

[0010] The wind power of the wind power station is predicted by each trained wind power prediction sub-model to obtain each wind power prediction value of the wind power station;

[0011] The wind power forecast values ​​are comprehensively calculated using the weighted average method to obtain the total wind power forecast value of the wind power station.

[0012] Furthermore, the preprocessing of the historical wind power data and the historical gas phase data includes:

[0013] Perform data cleaning on historical wind power data and historical gas phase data to remove, delete or correct erroneous values, outliers, missing values ​​and duplicate values ​​in the data;

[0014] The cleaned historical wind power data and historical gas phase data are smoothed and normalized.

[0015] Furthermore, the pre-processed historical wind power data and historical gas phase data are analyzed to determine the gas phase influencing factors affecting wind power in the historical gas phase data, including:

[0016] Obtain pre-processed historical wind power data and historical gas phase data, and divide the historical gas phase data into a number of gas phase parameter data according to data parameter types;

[0017] Draw trend graphs between each gas phase parameter data and historical wind power data respectively, obtain the slope of the trend line in each trend graph, and determine the gas phase parameters whose trend line slopes are within a preset range as candidate gas phase influencing factors;

[0018] Obtain candidate gas phase influencing factor data corresponding to the candidate gas phase influencing factor, and use the correlation analysis method to calculate the correlation coefficients between the candidate gas phase influencing factor data and the historical wind power data;

[0019] Candidate gas-phase influencing factors whose correlation coefficients exceed a preset threshold are screened out and determined as gas-phase influencing factors.

[0020] Furthermore, the feature extraction of each gas phase influencing factor to obtain the feature vector of each gas phase influencing factor includes:

[0021] Obtain gas phase influencing factor data corresponding to each gas phase influencing factor, use statistical feature extraction methods to extract features from each meteorological influencing factor data, and obtain corresponding statistical features of each gas phase influencing factor by calculation;

[0022] The corresponding statistical features of each gas phase influencing factor are combined to obtain the characteristic vector of each gas phase influencing factor.

[0023] Furthermore, the wind power prediction sub-models of the wind power station are respectively constructed based on the characteristic vectors of each gas phase influencing factor and the preset machine learning algorithm model, and each wind power prediction sub-model is trained and tested, including:

[0024] Selecting a machine learning algorithm model that is compatible with the characteristics of the characteristic vectors of each gas phase influencing factor as a preset machine learning algorithm model;

[0025] Based on each eigenvector and the machine learning algorithm model, a wind power prediction sub-model corresponding to each eigenvector is constructed;

[0026] Divide each feature vector into a training set and a test set according to a certain ratio;

[0027] The training set and the test set are input into each wind power prediction sub-model, and each wind power prediction sub-model is trained and tested until each wind power prediction sub-model meets the preset convergence conditions.

[0028] Furthermore, the wind power prediction sub-models completed through training predict the wind power of the wind power station to obtain the wind power prediction values ​​of the wind power station, including:

[0029] Acquire the current real-time gas phase data of the wind power station, and select the gas phase parameter data corresponding to the gas phase influencing factor from the real-time gas phase data;

[0030] Inputting the gas phase parameter data corresponding to each gas phase influencing factor into the corresponding wind power prediction sub-models;

[0031] The wind power prediction sub-models are used to output the wind power prediction values ​​of the wind power stations.

[0032] Furthermore, the wind power forecast values ​​are calculated comprehensively by weighted average method to obtain the total wind power forecast value of the wind power station, including:

[0033] The wind power prediction value of each wind power prediction sub-model for the wind power station and the gas phase influencing factors and their correlation coefficients corresponding to each wind power prediction sub-model;

[0034] Determine the range of the correlation coefficient of each gas phase influencing factor, and determine the weight coefficient of each wind power prediction sub-model according to the preset value corresponding to the range;

[0035] The weight coefficients of each gas phase influencing factor are normalized so that the sum of the weight coefficients of all wind power prediction sub-models is equal to 1;

[0036] Based on the wind power prediction value of the wind power station by each wind power prediction sub-model and the weight coefficient of each wind power prediction sub-model, the total wind power prediction value of the wind power station is calculated by weighted average method. The expression is as follows:

[0037]

[0038] Among them, S is the total wind power prediction value of the wind power station, Si represents the wind power prediction value of the wind power station by the i-th wind power prediction sub-model, xi represents the weight coefficient of the i-th wind power prediction sub-model, and n represents the number of wind power prediction sub-models.

[0039] A second aspect of the present invention provides a wind power prediction system based on machine learning, comprising:

[0040] An acquisition module is used to acquire the historical wind power data of the wind power station and the historical gas phase data corresponding to the time, and pre-process the historical wind power data and the historical gas phase data;

[0041] A determination module is used to analyze the pre-processed historical wind power data and historical gas phase data to determine the gas phase influencing factors that affect wind power in the historical gas phase data;

[0042] An extraction module is used to extract features of each gas phase influencing factor to obtain a feature vector of each gas phase influencing factor;

[0043] A modeling module, used to construct wind power prediction sub-models of wind power stations based on the characteristic vectors of each gas phase influencing factor and a preset machine learning algorithm model, and to train and test each wind power prediction sub-model;

[0044] A prediction module is used to predict the wind power of the wind power station through each trained wind power prediction sub-model to obtain each wind power prediction value of the wind power station;

[0045] The calculation module is used to perform comprehensive calculation on each wind power prediction value by weighted average method to obtain the total wind power prediction value of the wind power station.

[0046] Compared with the prior art, the wind power prediction method and system based on machine learning provided by the present invention have the following beneficial effects:

[0047] The present invention constructs wind power prediction sub-models for different gas phase influencing factors based on machine learning to perform real-time prediction, and calculates the total wind power prediction value by combining the prediction values ​​of multiple wind power prediction sub-models, thereby improving the accuracy of wind power prediction in wind power stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a schematic diagram of the flow structure of a wind power prediction method based on machine learning in an embodiment of the present invention;

[0049] Figure 2 It is a schematic diagram of the composition of a wind power prediction system based on machine learning in an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The specific implementation methods of the present application are further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0051] In the description of the present application, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the platform or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0052] The terms "second" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, a feature defined with "second" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "multiple" means two or more.

[0053] In the description of this application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0054] like Figure 1As shown, in an embodiment of the present application, a wind power prediction method based on machine learning is provided, including: S100: obtaining the past historical wind power data of the wind power station and the historical gas phase data corresponding to the time, and preprocessing the historical wind power data and the historical gas phase data; S200: analyzing the preprocessed historical wind power data and the historical gas phase data, and determining the gas phase influencing factors affecting the wind power in the historical gas phase data; S300: extracting features of each gas phase influencing factor to obtain the feature vector of each gas phase influencing factor; S400: constructing a wind power prediction sub-model of the wind power station based on the feature vector of each gas phase influencing factor and a preset machine learning algorithm model, and training and testing each wind power prediction sub-model; S500: predicting the wind power of the wind power station through each trained wind power prediction sub-model to obtain each wind power prediction value of the wind power station; S600: comprehensively calculating each wind power prediction value through the weighted average method to obtain the total wind power prediction value of the wind power station.

[0055] Furthermore, the present invention constructs wind power prediction sub-models for different gas phase influencing factors based on machine learning to perform real-time prediction, and calculates the total wind power prediction value by combining the prediction values ​​of multiple wind power prediction sub-models, thereby improving the accuracy of wind power prediction of wind power stations.

[0056] In an embodiment of the present application, a wind power prediction method based on machine learning is provided, wherein the preprocessing of historical wind power data and historical gas phase data includes: data cleaning of the historical wind power data and historical gas phase data, removing, deleting or correcting erroneous values, abnormal values, missing values ​​and duplicate values ​​in the data; and smoothing and normalizing the cleaned historical wind power data and historical gas phase data.

[0057] Specifically, the historical wind power data and historical meteorological data are checked to identify and process erroneous values, outliers and missing values. Erroneous values ​​and outliers may have a negative impact on analysis and modeling, so they need to be identified and corrected. Missing values ​​need to be processed by strategies such as filling or deletion. Duplicate values ​​need to be deduplicated to ensure the uniqueness of the data. The historical wind power data and historical meteorological data are smoothed using techniques such as moving average and exponential smoothing to reduce data volatility and make the data easier to analyze and model. The historical wind power data and historical meteorological data are normalized and scaled to a similar numerical range to eliminate the dimensional effects between different variables and help improve the convergence speed and performance of the model.

[0058] In an embodiment of the present application, a wind power prediction method based on machine learning is provided, wherein the preprocessed historical wind power data and historical gas phase data are analyzed to determine the gas phase influencing factors in the historical gas phase data that affect the wind power, including: obtaining the preprocessed historical wind power data and historical gas phase data, and dividing the historical gas phase data into a plurality of gas phase parameter data according to the data parameter type; respectively drawing trend graphs between each gas phase parameter data and the historical wind power data, and obtaining the slope of the trend line in each trend graph, and determining the gas phase parameters whose trend line slopes are within a preset range as candidate gas phase influencing factors; obtaining candidate gas phase influencing factor data corresponding to the candidate gas phase influencing factors, and using the correlation analysis method to respectively calculate the correlation coefficients of the candidate gas phase influencing factor data and the historical wind power data; screening out the candidate gas phase influencing factors whose correlation coefficients exceed a preset threshold value, and determining them as gas phase influencing factors.

[0059] Specifically, the pre-processed historical wind power data and historical meteorological data are obtained to ensure the quality and integrity of the data. The historical meteorological data are divided into several meteorological parameter data according to the data parameter type. The historical meteorological data can be divided and sorted according to different meteorological parameters (such as temperature, humidity, wind speed, etc.) to obtain data sets of various meteorological parameters; for each meteorological parameter data, a trend graph is drawn to show the relationship between it and the historical wind power data, and then the slope of the trend line is calculated to quantify the degree of correlation between the meteorological parameter and wind power. According to the preset slope range, the meteorological parameters with the slope of the trend line within the range are determined as candidate meteorological influencing factors; for the candidate meteorological influencing factor data, the correlation coefficient with the historical wind power data is calculated using a correlation analysis method (such as the Pearson correlation coefficient), and the candidate meteorological influencing factors with correlation coefficients exceeding the preset threshold are screened out and determined as meteorological influencing factors.

[0060] In an embodiment of the present application, a wind power prediction method based on machine learning is provided, wherein feature extraction is performed on each gas phase influencing factor to obtain a feature vector of each gas phase influencing factor, including: acquiring gas phase influencing factor data corresponding to each gas phase influencing factor, taking a statistical feature extraction method to perform feature extraction on each meteorological influencing factor data, and obtaining the corresponding statistical features of each gas phase influencing factor by calculation; and combining the corresponding statistical features of each gas phase influencing factor to obtain the feature vector of each gas phase influencing factor.

[0061] Specifically, according to the meteorological influencing factors screened out previously, the data corresponding to these factors are extracted from the original historical meteorological data to form a data set of each meteorological influencing factor; for each data set of meteorological influencing factors, various statistical feature extraction methods are used, such as mean, variance, maximum value, minimum value, median, skewness, kurtosis and other statistical indicators to describe the distribution characteristics and degree of variation of the data, and for each data set of meteorological influencing factors, the corresponding statistical characteristic values ​​are calculated respectively; the corresponding statistical characteristic values ​​of each meteorological influencing factor are combined to obtain the characteristic vector of each meteorological influencing factor. These characteristic vectors serve as abstract representations to describe the impact of meteorological influencing factors on wind power, providing a data basis for subsequent modeling and analysis.

[0062] In an embodiment of the present application, a wind power prediction method based on machine learning is provided, wherein a wind power prediction sub-model of a wind power station is respectively constructed based on the characteristic vector of each gas phase influencing factor and a preset machine learning algorithm model, and each wind power prediction sub-model is trained and tested, including: selecting a machine learning algorithm model that is adapted to the characteristics of the characteristic vector of each gas phase influencing factor as the preset machine learning algorithm model; constructing a wind power prediction sub-model corresponding to each characteristic vector based on each characteristic vector and the machine learning algorithm model; dividing each characteristic vector into a training set and a test set according to a certain ratio; inputting the training set and the test set into each wind power prediction sub-model, and training and testing each wind power prediction sub-model until each wind power prediction sub-model meets the preset convergence condition.

[0063] Specifically, according to the extracted eigenvectors of each meteorological influencing factor, a suitable machine learning algorithm model is selected. When selecting the model, the model is selected according to the characteristics of the eigenvector, the distribution of the data and the applicable scope of the model; for each eigenvector, the selected machine learning algorithm model is used to build the corresponding wind power prediction sub-model; each eigenvector is divided into a training set and a test set according to a certain ratio, usually 70% of the data is used as the training set and 30% of the data is used as the test set; the training set and the test set are input into each wind power prediction sub-model, and each sub-model is trained and tested. During the training process, the model will adjust the parameters according to the data of the training set until the preset convergence conditions are met. During the test process, the performance indicators of the model are evaluated, such as the root mean square error (RMSE), the coefficient of determination (R2), etc., to verify the prediction effect of the model.

[0064] In an embodiment of the present application, a wind power prediction method based on machine learning is provided, wherein each wind power prediction sub-model that has been trained predicts the wind power of a wind power station to obtain each wind power prediction value of the wind power station, including: acquiring the current real-time gas phase data of the wind power station, and selecting gas phase parameter data corresponding to the gas phase influencing factor from the real-time gas phase data; inputting the gas phase parameter data corresponding to each gas phase influencing factor into the corresponding wind power prediction sub-models; and outputting by each wind power prediction sub-model to obtain the wind power prediction value of the wind power station for each wind power prediction sub-model.

[0065] Specifically, based on the previously determined meteorological influencing factors, the latest data corresponding to these factors are extracted from the real-time meteorological data; the meteorological parameter data obtained in real time is input into the previously constructed wind power prediction sub-models for prediction. Through the prediction of each wind power prediction sub-model, the wind power prediction value under the current meteorological conditions can be obtained.

[0066] In an embodiment of the present application, a method for predicting wind power based on machine learning is provided, wherein the wind power prediction values ​​are comprehensively calculated by a weighted average method to obtain a total wind power prediction value of a wind power station, including: the wind power prediction value of the wind power station by each wind power prediction sub-model and each gas phase influencing factor and its correlation coefficient corresponding to each wind power prediction sub-model; judging the range interval of the correlation coefficient of each gas phase influencing factor, and determining the weight coefficient of each wind power prediction sub-model according to the preset value corresponding to the range interval; normalizing the weight coefficient of each gas phase influencing factor so that the sum of the weight coefficients of all wind power prediction sub-models is equal to 1; based on the wind power prediction value of the wind power station by each wind power prediction sub-model and the weight coefficient of each wind power prediction sub-model, the total wind power prediction value of the wind power station is calculated by a weighted average method, and the expression is as follows:

[0067]

[0068] Among them, S is the total wind power prediction value of the wind power station, Si represents the wind power prediction value of the wind power station by the i-th wind power prediction sub-model, xi represents the weight coefficient of the i-th wind power prediction sub-model, and n represents the number of wind power prediction sub-models.

[0069] Specifically, the prediction results of each wind power prediction sub-model and the meteorological influencing factors used in each model and their correlation coefficients are obtained; the range interval of the correlation coefficient of each meteorological influencing factor is determined, and the weight coefficient of each wind power prediction sub-model is determined according to the preset value corresponding to the range interval, wherein the correlation coefficient is divided into different range intervals according to the size of the correlation coefficient, and then a preset weight value is determined for each interval to indicate the importance of the factor to the prediction; the weight coefficient of each meteorological influencing factor is normalized so that the sum of the weight coefficients of all wind power prediction sub-models is equal to 1, wherein the normalization process ensures that the sum of the weight coefficients is equal to 1, so as to ensure the effectiveness of the weighted average; based on the wind power prediction value of the wind power station for each wind power prediction sub-model and the weight coefficient of each wind power prediction sub-model, the total wind power prediction value of the wind power station is calculated by the weighted average method, wherein the prediction results of each model are weighted averaged using the normalized weight coefficient to obtain the final total wind power prediction value.

[0070] like Figure 2 As shown, in an embodiment of the present application, a wind power prediction system based on machine learning is provided, including: an acquisition module, used to acquire the past historical wind power data of the wind power station and the historical gas phase data corresponding to the time, and pre-process the historical wind power data and the historical gas phase data; a determination module, used to analyze the pre-processed historical wind power data and the historical gas phase data, and determine the gas phase influencing factors affecting the wind power in the historical gas phase data; an extraction module, used to extract features of each gas phase influencing factor, and obtain the feature vector of each gas phase influencing factor; a modeling module, used to construct a wind power prediction sub-model of the wind power station based on the feature vector of each gas phase influencing factor and a preset machine learning algorithm model, and train and test each wind power prediction sub-model; a prediction module, used to predict the wind power of the wind power station through each trained wind power prediction sub-model, and obtain each wind power prediction value of the wind power station; a calculation module, used to comprehensively calculate each wind power prediction value through the weighted average method, and obtain the total wind power prediction value of the wind power station.

[0071] In summary, the embodiment of the present invention provides a method and system for predicting wind power based on machine learning, which includes: obtaining historical wind power data and historical gas phase data of a wind power station, preprocessing these data, determining the gas phase influencing factors that affect wind power from the historical gas phase data after preprocessing, extracting the characteristic vectors of each gas phase influencing factor, constructing and training wind power prediction sub-models based on the characteristic vectors of each gas phase influencing factor and a preset machine learning algorithm model, predicting each wind power prediction value through each wind power prediction sub-model, and weighted averaging each wind power prediction value to obtain a total wind power prediction value. The present invention improves the accuracy of wind power prediction of wind power stations by constructing wind power prediction sub-models for different gas phase influencing factors based on machine learning to perform real-time prediction, and calculating the total wind power prediction value by combining the prediction values ​​of multiple wind power prediction sub-models.

[0072] Finally, it should be noted that it is obvious that those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

[0073] The above is only an example of implementation of the present invention, but it cannot be used to limit the scope of the present invention. Any structural changes made according to the present invention, as long as they do not lose the essence of the present invention, should be regarded as falling within the scope of protection of the present invention and being restricted. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process and related instructions of the platform described above can refer to the corresponding process in the aforementioned platform embodiment, and will not be repeated here.

[0074] The term "comprises" or any other similar term is intended to cover a non-exclusive inclusion such that a process, platform, article, or apparatus / platform that includes a list of elements includes not only those elements but also other elements not expressly listed or inherent to such process, platform, article, or apparatus / platform.

[0075] So far, the technical solutions of the present invention have been described in conjunction with the further embodiments shown in the accompanying drawings, but it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0076] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention.

Claims

1. A wind power prediction method based on machine learning, characterized in that: include: Obtain the historical wind power data of the wind power station and the historical gas phase data corresponding to the time, and pre-process the historical wind power data and the historical gas phase data; Analyze the pre-processed historical wind power data and historical gas phase data to determine the gas phase influencing factors that affect wind power in the historical gas phase data; Extract features of each gas phase influencing factor to obtain a feature vector of each gas phase influencing factor; Based on the characteristic vectors of each gas phase influencing factor and the preset machine learning algorithm model, the wind power prediction sub-models of the wind power station are respectively constructed, and each wind power prediction sub-model is trained and tested; The wind power of the wind power station is predicted by each trained wind power prediction sub-model to obtain each wind power prediction value of the wind power station; The wind power forecast values ​​are comprehensively calculated using the weighted average method to obtain the total wind power forecast value of the wind power station.

2. The method for predicting wind power based on machine learning according to claim 1, characterized in that: The preprocessing of the historical wind power data and the historical gas phase data includes: Perform data cleaning on historical wind power data and historical gas phase data to remove, delete or correct erroneous values, outliers, missing values ​​and duplicate values ​​in the data; The cleaned historical wind power data and historical gas phase data are smoothed and normalized.

3. The method for predicting wind power based on machine learning according to claim 2, characterized in that: The analysis of the pre-processed historical wind power data and historical gas phase data to determine the gas phase influencing factors affecting wind power in the historical gas phase data includes: Obtain pre-processed historical wind power data and historical gas phase data, and divide the historical gas phase data into a number of gas phase parameter data according to data parameter types; Draw trend graphs between each gas phase parameter data and historical wind power data respectively, obtain the slope of the trend line in each trend graph, and determine the gas phase parameters whose trend line slopes are within a preset range as candidate gas phase influencing factors; Obtain candidate gas phase influencing factor data corresponding to the candidate gas phase influencing factor, and use the correlation analysis method to calculate the correlation coefficients between the candidate gas phase influencing factor data and the historical wind power data; Candidate gas-phase influencing factors whose correlation coefficients exceed a preset threshold are screened out and determined as gas-phase influencing factors.

4. The method for predicting wind power based on machine learning according to claim 3, characterized in that: The feature extraction of each gas phase influencing factor to obtain a feature vector of each gas phase influencing factor includes: Obtain gas phase influencing factor data corresponding to each gas phase influencing factor, use statistical feature extraction methods to extract features from each meteorological influencing factor data, and obtain corresponding statistical features of each gas phase influencing factor by calculation; The corresponding statistical features of each gas phase influencing factor are combined to obtain the characteristic vector of each gas phase influencing factor.

5. The method for predicting wind power based on machine learning according to claim 1, characterized in that: The wind power prediction sub-models of the wind power station are respectively constructed based on the characteristic vectors of each gas phase influencing factor and the preset machine learning algorithm model, and each wind power prediction sub-model is trained and tested, including: Selecting a machine learning algorithm model that is compatible with the characteristics of the characteristic vectors of each gas phase influencing factor as a preset machine learning algorithm model; Based on each eigenvector and the machine learning algorithm model, a wind power prediction sub-model corresponding to each eigenvector is constructed; Divide each feature vector into a training set and a test set according to a certain ratio; The training set and the test set are input into each wind power prediction sub-model, and each wind power prediction sub-model is trained and tested until each wind power prediction sub-model meets the preset convergence conditions.

6. The method for predicting wind power based on machine learning according to claim 5, characterized in that: The wind power prediction sub-models completed through training predict the wind power of the wind power station to obtain the wind power prediction values ​​of the wind power station, including: Acquire the current real-time gas phase data of the wind power station, and select the gas phase parameter data corresponding to the gas phase influencing factor from the real-time gas phase data; Inputting the gas phase parameter data corresponding to each gas phase influencing factor into the corresponding wind power prediction sub-models; The wind power prediction sub-models are used to output the wind power prediction values ​​of the wind power stations.

7. The method for predicting wind power based on machine learning according to claim 6, characterized in that: The wind power forecast value of the wind power station is obtained by comprehensively calculating each wind power forecast value by weighted average method, including: The wind power prediction value of each wind power prediction sub-model for the wind power station and the gas phase influencing factors and their correlation coefficients corresponding to each wind power prediction sub-model; Determine the range of the correlation coefficient of each gas phase influencing factor, and determine the weight coefficient of each wind power prediction sub-model according to the preset value corresponding to the range; The weight coefficients of each gas phase influencing factor are normalized so that the sum of the weight coefficients of all wind power prediction sub-models is equal to 1; Based on the wind power prediction value of the wind power station by each wind power prediction sub-model and the weight coefficient of each wind power prediction sub-model, the total wind power prediction value of the wind power station is calculated by weighted average method. The expression is as follows: Among them, S is the total wind power prediction value of the wind power station, Si represents the wind power prediction value of the wind power station by the i-th wind power prediction sub-model, xi represents the weight coefficient of the i-th wind power prediction sub-model, and n represents the number of wind power prediction sub-models.

8. A wind power prediction system based on machine learning, characterized in that: include: An acquisition module is used to acquire the historical wind power data of the wind power station and the historical gas phase data corresponding to the time, and pre-process the historical wind power data and the historical gas phase data; A determination module is used to analyze the pre-processed historical wind power data and historical gas phase data to determine the gas phase influencing factors that affect wind power in the historical gas phase data; An extraction module is used to extract features of each gas phase influencing factor to obtain a feature vector of each gas phase influencing factor; A modeling module, used to construct wind power prediction sub-models of wind power stations based on the characteristic vectors of each gas phase influencing factor and a preset machine learning algorithm model, and to train and test each wind power prediction sub-model; A prediction module is used to predict the wind power of the wind power station through each trained wind power prediction sub-model to obtain each wind power prediction value of the wind power station; The calculation module is used to perform comprehensive calculation on each wind power prediction value by weighted average method to obtain the total wind power prediction value of the wind power station.

Images