Method for predicting wind power generation capacity of wind power plant

Abstract

The invention relates to the technical field of wind power generation, in particular to a wind power generation capacity prediction method for a wind power plant, which comprises the following steps: clustering generators based on environmental data to obtain a plurality of environmental clusters, the effective power generation matrix is used for representing the power generation capacity of the power generator under each piece of wind field information; constructing a prediction model of each environment cluster; and obtaining the predicted generating capacity of each generator according to the prediction model of the environment cluster to which each generator in the wind power plant belongs, and taking the sum of all the predicted generating capacity as the generating capacity prediction result of the wind power plant at the next adjacent moment. According to the technical scheme, the accurate prediction result of the wind power generation can be obtained.

Description

Technical Field

[0001] The present application relates to the technical field of wind power generation, and in particular to a method for predicting wind power generation in a wind farm. Background Art

[0002] With increasing attention paid to energy shortages and environmental issues, wind energy is being continuously developed and utilized as a clean energy source, and wind power generation technology is also developing continuously. However, due to the volatility and randomness of wind, the amount of power generated by wind power is also highly random and uncertain, which has a certain impact on the stable operation of the power grid. In order to improve the utilization rate of wind energy and enhance the level of wind energy absorption, it is necessary to accurately predict the power generation of wind power stations.

[0003] At present, the patent application document with publication number CN111311022A discloses a power generation prediction method, device, equipment and computer-readable storage medium, wherein the method includes: obtaining a satellite image to be predicted and a first shooting parameter corresponding to the satellite image to be predicted, and obtaining first outdoor wind information corresponding to the satellite image to be predicted; the power generation prediction model includes a deep learning model and a power generation regression model, the satellite image to be predicted is input into the deep learning model to obtain multiple first target rectangular frame information corresponding to the satellite image to be predicted, and the probability that the first target frame corresponding to the first target rectangular frame information is a wind turbine; the first shooting parameter, the first outdoor wind information, the first target rectangular frame information and the probability that the first target frame corresponding to the first target rectangular frame information is a wind turbine are input into the power generation regression model to obtain the power generation of the wind power station corresponding to the satellite image to be predicted.

[0004] The above method first obtains the position coordinates of the wind turbine in the satellite image to be predicted, and then predicts the power generation of the wind power station corresponding to the satellite image to be predicted based on the first outdoor wind information, the first shooting parameters and the position coordinates of the wind turbine. However, the above method only obtains the position coordinates of the wind turbine and ignores the influence of information such as the direction of the wind turbine on wind power generation, resulting in inaccurate prediction results of wind power generation. Summary of the Invention

[0005] In order to solve the technical problem of inaccurate prediction results of wind power generation, the present application provides a wind farm wind power generation prediction method, which can obtain accurate wind power generation prediction results.

[0006] In a first aspect, the present application provides a method for predicting wind power generation in a wind farm, the method comprising: clustering generators based on environmental data to obtain multiple environmental clusters, the environmental data comprising terrain category, altitude, and an effective power generation matrix, the effective power generation matrix being used to characterize the power generation capacity of the generator under each wind farm information; constructing a prediction model for each environmental cluster, the input of the prediction model being a time series of wind farm information of each grid area within the neighborhood of any generator, and the output being the predicted power generation of the generator, the wind farm information comprising wind speed and the angle between wind direction and generator yaw angle; obtaining the predicted power generation of each generator in the wind farm according to the prediction model of the environmental cluster to which each generator belongs, and taking the sum of all predicted power generation as the power generation prediction result of the wind farm at the next adjacent moment; wherein, the method for obtaining the effective power generation matrix comprises: constructing an initial matrix with wind speed as rows and the angle between wind direction and yaw angle as columns, all of which have values of 0; calculating the average power generation under each combination of wind speed and angle within a historical time period, and adding the average power generation to the value at the corresponding position in the initial matrix; and normalizing all values in the initial matrix to obtain the effective power generation matrix.

[0007] Taking into account that wind power generation is greatly affected by the environment and the power generation capacity of wind power generation varies in different environments, the generators are divided into multiple environmental clusters based on environmental data, and the power generation capacity of each generator in an environmental cluster is basically the same; further, a prediction model for each environmental cluster is constructed, and the prediction model is used to predict the power generation of the generators in the environmental cluster, eliminating the influence of environmental data on the power generation capacity and improving the accuracy of the power generation prediction; the predicted power generation of each generator in the wind farm is predicted based on the prediction model of the environmental cluster to which each generator belongs, and the sum of all predicted power generation is used as the power generation prediction result of the wind farm to obtain accurate wind power generation prediction results.

[0008] Preferably, obtaining multiple environmental clusters includes: defining a clustering distance and an initial clustering number, and clustering each generator using a K-prototypes algorithm to obtain a clustering result, wherein the clustering distance is the sum of the distances of each environmental data between any two generators; adjusting the initial clustering number multiple times, and determining a target clustering number using an elbow method, wherein the clustering result of the target clustering number corresponds to the multiple environmental clusters.

[0009] Preferably, the prediction model includes a convolution sub-model, a time series sub-model and a regression sub-model; the convolution sub-model is used to extract features of the wind field information of each grid area within the neighborhood at any time to obtain the wind field feature vector at each time; the time series sub-model is used to map the wind field feature vector at each time into a wind field time series vector; the wind field time series vector is input into the regression sub-model to obtain the predicted power generation.

[0010] The convolution sub-model is used to extract the wind field information of each grid area within the neighborhood range (spatial dimension) at any time, and then the time series sub-model is used to extract the time series features of the wind field information (time dimension), obtaining a wind field time series vector that can characterize the changes in wind field information. The predicted power generation of the generator is output by combining the spatial and time dimensions.

[0011] Preferably, the convolution sub-model is a convolutional neural network, the timing sub-model is a recurrent neural network, and the regression sub-model is a fully connected neural network.

[0012] Preferably, the construction of the prediction model for each environmental cluster includes: collecting the historical time series of wind field information of each grid area within the neighborhood of any generator in the environmental cluster as an input sample, and taking the actual power generation at the next adjacent moment of the historical time series as the label of the input sample; inputting the input sample into the prediction model to obtain the output result, and iteratively training the prediction model according to the loss function, where the loss function is the mean square error function between the output result and the label; in response to the loss function being less than the preset loss or the number of iterations being greater than the preset number, completing the construction of the environmental cluster prediction model.

[0013] To ensure that the prediction model of each environmental cluster can accurately predict the power generation of the generator, the historical time series of wind field information of each grid area within the neighborhood of any generator in the environmental cluster is used to train the prediction model of the environmental cluster.

[0014] Preferably, the loss function also includes: calculating the Euclidean distance between the environmental data of each generator in the environmental cluster and the cluster center, and using the normalized Euclidean distance as the attention of the corresponding generator; and using the product of the mean square error function between the output result and the label and the attention as the loss function.

[0015] Since the environmental data of each generator in an environmental cluster may still be different, during the training process, higher attention is given to the generators that deviate from the cluster center in the environmental cluster to enhance the robustness of the prediction model within the environmental cluster, that is, to ensure that the prediction model can accurately predict the power generation of all generators in the environmental cluster.

[0016] Preferably, the loss function for: ; Environmental Cluster Internal generator The Euclidean distance between the environmental data and the cluster center, Environmental Cluster The sum of the Euclidean distances between the environmental data of all generators and the cluster center, Environmental Cluster Internal generator The mean square error function between the output result and the label corresponding to the input sample.

[0017] Preferably, the method for determining the environmental cluster to which a generator belongs includes: collecting environmental data of any generator, calculating the similarity between the environmental data and the cluster centers of each environmental cluster, and taking the environmental cluster corresponding to the maximum similarity as the environmental cluster to which the generator belongs.

[0018] Preferably, the similarity is negatively correlated with the Euclidean distance between the environmental data and the cluster center of each environmental cluster.

[0019] Preferably, the neighborhood range is a rectangular area centered on the generator, the rectangular area includes a plurality of grid areas, and the wind field information of each grid area within the neighborhood range is a two-channel image data.

[0020] The technical solution of this application has the following beneficial technical effects: First, considering that wind power generation is greatly affected by the environment and the power generation capacity of wind power generation varies in different environments, the generators are divided into multiple environmental clusters based on environmental data, and the power generation capacity of each generator in an environmental cluster is basically the same; further, a prediction model for each environmental cluster is constructed, and the prediction model is used to predict the power generation of the generators in the environmental cluster, eliminating the impact of environmental data on the power generation capacity and improving the accuracy of the power generation prediction; the predicted power generation of each generator in the wind farm is predicted based on the prediction model of the environmental cluster to which each generator belongs, and the sum of all predicted power generation is used as the power generation prediction result of the wind farm to obtain accurate wind power generation prediction results.

[0021] Among them, the environmental data includes terrain type, altitude and effective power generation matrix. The terrain type and altitude distinguish the power generation capacity of the generator from the perspective of the geographical environment. The effective power generation matrix distinguishes the power generation capacity of the generator from the perspective of wind field information, realizing the accurate division of environmental clusters and ensuring that the power generation capacity of each generator in an environmental cluster is basically consistent. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a flow chart of a method for predicting wind power generation in a wind farm according to an embodiment of the present application.

[0023] Figure 2 It is a structural diagram of a prediction model according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by technicians in this field without making creative efforts are within the scope of protection of this application.

[0025] According to a first aspect of the present application, the present application provides a method for predicting wind power generation in a wind farm. Figure 1 This is a flow chart of a method for predicting wind power generation in a wind farm according to an embodiment of the present application. Figure 1 As shown, the wind farm wind power generation prediction method includes steps S101 to S103, which are described in detail below.

[0026] S101 , clustering generators based on environmental data to obtain multiple environmental clusters, wherein the environmental data includes terrain type, altitude, and effective power generation matrix, and the effective power generation matrix is used to characterize the power generation capacity of the generators under each wind farm information.

[0027] In one embodiment, since wind power generation is greatly affected by the environment, the power generation capacity of wind power generation varies in different environments. In order to ensure the accuracy of the power generation prediction results, it is necessary to cluster the generators based on the environmental data to obtain multiple environmental clusters. If two generators belong to the same environmental cluster, then the power generation capacity of the two generators is the same. Subsequently, a prediction model will be constructed for each environmental cluster to eliminate the impact of environmental data on the power generation capacity and improve the accuracy of the power generation prediction.

[0028] The environmental data includes terrain category, altitude and effective power generation matrix, wherein the terrain category includes plains, hills and gentle hills. The terrain category and altitude distinguish the power generation capacity of the generator from the perspective of geographical environment, while the effective power generation matrix distinguishes the power generation capacity of the generator from the perspective of wind field information. In other words, the effective power generation matrix can characterize the power generation capacity of the generator under different wind field information.

[0029] Specifically, the method for obtaining the effective power generation matrix includes: constructing an initial matrix with wind speed as the row and wind direction and yaw angle as the column, all with zero values; calculating the average power generated for each combination of wind speed and angle over a historical period, and adding the average power generated to the value at the corresponding position in the initial matrix; and normalizing all values in the initial matrix to obtain the effective power generation matrix. Normalization is achieved by calculating the ratio of any value to the sum of the values in the initial matrix, thus eliminating dimension.

[0030] The wind speed is divided into 30 intervals at 1 meter per second intervals, from 0 to 30 meters per second. This means the initial matrix has 30 rows. Furthermore, the angle between wind direction and yaw angle ranges from 0 to 180 degrees, divided into 36 intervals at 5-degree intervals. This means the initial matrix has 36 rows and columns, resulting in a 30-row, 36-column matrix. The yaw angle represents the generator's orientation and can be directly acquired using the angle sensor deployed on the generator.

[0031] It can be understood that a value in the initial matrix corresponds to a type of wind field information. For example, the 5th row and the 10th column corresponds to wind field information with a wind speed of 5 meters per second and an angle between the wind direction and the yaw angle of 45 degrees to 50 degrees.

[0032] In one embodiment, obtaining multiple environmental clusters includes: defining a clustering distance and an initial cluster number, and clustering each generator using a K-prototypes algorithm to obtain a clustering result, wherein the clustering distance is the sum of the distances of each environmental data between any two generators; adjusting the initial cluster number multiple times, and determining a target cluster number using an elbow method, wherein the clustering result of the target cluster number corresponds to the multiple environmental clusters.

[0033] Among them, the K-prototypes algorithm is a clustering algorithm for processing mixed data, which can simultaneously perform clustering on numerical data and categorical data. In the embodiment of the present application, the terrain category is categorical data, and the altitude and effective power generation matrix are numerical data. Therefore, the K-prototypes algorithm is used for clustering; the initial number of clusters is 2; the number of environmental clusters is equal to the target number of clusters, and an environmental cluster includes multiple generators. The average value of the environmental data of multiple generators in the environmental cluster is used as the clustering center of the environmental cluster, and the clustering center is used to determine the environmental cluster to which each generator in the wind farm belongs.

[0034] In this way, the generators are divided into multiple environmental clusters based on the environmental data, and the power generation capacity of each generator in an environmental cluster is basically the same.

[0035] S102, constructing a prediction model for each environmental cluster, wherein the input of the prediction model is the time series of wind field information of each grid area within the neighborhood of any generator, and the output is the predicted power generation of the generator, wherein the wind field information includes wind speed and the angle between wind direction and generator yaw angle.

[0036] In one embodiment, since the power generation capacity of each generator in an environmental cluster is substantially the same, a prediction model is constructed for each environmental cluster to eliminate the differences in generator power generation capacity caused by the environmental cluster and improve the accuracy of power generation prediction results.

[0037] The input of the prediction model is the time series of wind field information of each grid area within the neighborhood of any generator, and the output is the predicted power generation of the generator. The wind field information of a grid area includes the wind speed, wind direction and the angle between the generator yaw angle within the grid area. The neighborhood range is a rectangular area centered on the generator, and a rectangular area includes multiple grid areas. The wind field information of each grid area within the neighborhood range can be regarded as a two-channel image data.

[0038] For example, if the grid area is 10 meters by 10 meters, a wind speed sensor and a wind direction sensor are deployed within each grid area to collect the wind speed and direction within the grid area. Combined with the generator yaw angle, the wind speed within the grid area and the angle between the wind direction and the generator yaw angle can be obtained. If a rectangular area includes grid areas, the wind field information for each grid area within the neighborhood is a 3×3 two-channel image data. One channel image includes the wind speed data for the grid area where the generator is located and the other eight grid areas within the neighborhood; the other channel image includes the wind direction and the angle between the generator yaw angle and the generator yaw angle within the grid area and the other eight grid areas.

[0039] The same method can be used to obtain the wind field information of each grid area within the neighborhood at each moment. By arranging the wind field information of each grid area within the neighborhood in chronological order, a time series of the wind field information of each grid area within the neighborhood can be obtained. This time series can characterize the changes in the wind field information within the neighborhood of the generator. The time series includes the current moment and multiple historical moments before the current moment.

[0040] Figure 2 This is a structural diagram of a prediction model according to an embodiment of the present application. The prediction model includes a convolution sub-model, a time series sub-model, and a regression sub-model. The convolution sub-model is used to extract features from wind farm information in each grid area within a neighborhood at any time to obtain a wind farm feature vector at each time. The time series sub-model is used to map the wind farm feature vector at each time into a wind farm time series vector. The wind farm time series vector is input into the regression sub-model to obtain predicted power generation.

[0041] Among them, the convolution sub-model can adopt existing convolutional neural networks such as ResNet, VGGNet or ShuffleNet, the timing sub-model can adopt existing recurrent neural networks such as LSTM or Transformer, and the regression sub-model adopts a fully connected neural network.

[0042] In one embodiment, in order to ensure that the prediction model of each environmental cluster can accurately predict the power generation of the generator, it is necessary to train the prediction model of each environmental cluster. The model structure of the prediction model of each environmental cluster is the same, but the training process is different. Specifically, the construction of the prediction model of each environmental cluster includes: collecting the historical time series of wind field information of each grid area within the neighborhood of any generator in the environmental cluster as an input sample, and using the actual power generation at the next adjacent moment of the historical time series as the label of the input sample; inputting the input sample into the prediction model to obtain the output result, and iteratively training the prediction model according to the loss function, where the loss function is the mean square error function between the output result and the label; in response to the loss function being less than the preset loss or the number of iterations being greater than the preset number, the construction of the environmental cluster prediction model is completed.

[0043] Among them, the preset loss is 0.01 and the preset number of times is 500.

[0044] In another embodiment, within an environmental cluster, the environmental data of each generator may still differ. To enhance the robustness of the prediction model within the cluster, that is, to ensure that the prediction model can accurately predict the power generation of all generators within the cluster, generators within the cluster that deviate from the cluster center are given a higher degree of attention during training. The loss function further includes: calculating the Euclidean distance between the environmental data of each generator within the cluster and the cluster center, using the normalized Euclidean distance as the attention level of the corresponding generator; and using the product of the mean square error function between the output result and the label and the attention level as the loss function.

[0045] Specifically, the loss function for: ; Environmental Cluster Internal generator The Euclidean distance between the environmental data and the cluster center, Environmental Cluster The sum of the Euclidean distances between the environmental data of all generators and the cluster center, Environmental Cluster Internal generator The mean square error function between the output result and the label corresponding to the input sample.

[0046] In this way, a prediction model for each environmental cluster is obtained; the prediction model can accurately predict the power generation of all generators in the corresponding environmental cluster.

[0047] S103 , obtaining the predicted power generation of each generator in the wind farm according to the prediction model of the environmental cluster to which each generator belongs, and taking the sum of all the predicted power generation as the power generation prediction result of the wind farm at the next adjacent moment.

[0048] In one embodiment, a method for determining the environmental cluster to which a generator belongs includes: collecting environmental data of any generator, calculating the similarity between the environmental data and the cluster centers of each environmental cluster, and taking the environmental cluster corresponding to the maximum similarity as the environmental cluster to which the generator belongs.

[0049] The similarity is a similarity calculation method based on Euclidean distance. The similarity between environmental data and the cluster center of any environmental cluster is negatively correlated with the Euclidean distance between the environmental data and the cluster center of the environmental cluster.

[0050] A wind farm includes multiple generators. The environmental cluster to which each generator belongs is determined based on the generator's environmental data, and then the predicted power generation of each generator at the next adjacent moment is accurately obtained based on the prediction model of the environmental cluster to which it belongs. The sum of all predicted power generation is used as the power generation prediction result of the wind farm at the next adjacent moment. In this way, the wind power generation of the wind farm can be accurately predicted.

[0051] It should be noted that a person skilled in the art may make a number of modifications and improvements without departing from the concept of the present application, and these modifications and improvements are all within the scope of protection of the present application. Therefore, the scope of protection of the patent application shall be based on the appended claims.

Claims

1. A method for predicting wind power generation in a wind farm, characterized in that: The prediction method comprises: Clustering the generators based on environmental data to obtain multiple environmental clusters, the environmental data including terrain type, altitude, and an effective power generation matrix, which is used to characterize the power generation capacity of the generators under each wind farm information; Construct a prediction model for each environmental cluster. The input of the prediction model is a time series of wind field information for each grid area within the neighborhood of any generator. The output is the predicted power generation of the generator. The wind field information includes wind speed, wind direction, and the angle between the generator yaw angle. Obtain the predicted power generation of each generator in the wind farm based on the prediction model of the environmental cluster to which each generator belongs, and take the sum of all predicted power generation as the power generation prediction result of the wind farm at the next adjacent moment; Among them, the method for obtaining the effective power generation matrix includes: constructing an initial matrix with wind speed as rows and the angle between wind direction and yaw angle as columns, with all values being 0; counting the average power generation power under each combination of wind speed and angle within the historical time, and adding the average power generation power to the value at the corresponding position in the initial matrix; normalizing all values in the initial matrix to obtain the effective power generation matrix.

2. A method for predicting wind power generation in a wind farm according to claim 1, characterized in that: Obtaining multiple environment clusters includes: Define the cluster distance and the initial number of clusters, and use the K-prototypes algorithm to cluster the generators to obtain the clustering results. The cluster distance is the sum of the distances of the environmental data between any two generators. The initial cluster number is adjusted multiple times, and a target cluster number is determined using an elbow method, where a clustering result of the target cluster number corresponds to the multiple environmental clusters.

3. A method for predicting wind power generation in a wind farm according to claim 1, characterized in that: The prediction model includes a convolution sub-model, a time series sub-model and a regression sub-model; The convolution sub-model is used to extract the features of the wind field information of each grid area within the neighborhood at any time, and obtain the wind field feature vector at each time; The time series sub-model is used to map the wind field characteristic vector at each moment into a wind field time series vector; The wind farm time series vector is input into the regression sub-model to obtain the predicted power generation.

4. A method for predicting wind power generation in a wind farm according to claim 3, characterized in that: The convolution sub-model is a convolutional neural network, the timing sub-model is a recurrent neural network, and the regression sub-model is a fully connected neural network.

5. The method for predicting wind power generation in a wind farm according to claim 1, characterized in that: The construction of the prediction model for each environmental cluster includes: Collect the historical time series of wind farm information in each grid area within the neighborhood of any generator in the environmental cluster as input samples, and use the actual power generation at the next adjacent moment of the historical time series as the label of the input sample; Input samples into the prediction model to obtain output results, and iteratively train the prediction model according to the loss function. The loss function is the mean square error function between the output results and the labels. In response to the loss function being less than a preset loss or the number of iterations being greater than a preset number, the construction of the environment cluster prediction model is completed.

6. A method for predicting wind power generation in a wind farm according to claim 5, characterized in that: The loss function also includes: Calculate the Euclidean distance between the environmental data of each generator in the environmental cluster and the cluster center, and use the normalized Euclidean distance as the attention level of the corresponding generator; The product of the mean square error function between the output result and the label and the attention degree is used as the loss function.

7. A method for predicting wind power generation in a wind farm according to claim 6, characterized in that: Loss Function for: ; Environmental Cluster Internal generator The Euclidean distance between the environmental data and the cluster center, Environmental Cluster The sum of the Euclidean distances between the environmental data of all generators and the cluster center, Environmental Cluster Internal generator The mean square error function between the output result and the label corresponding to the input sample.

8. The method for predicting wind power generation in a wind farm according to claim 1, wherein: Methods for determining the environmental cluster to which a generator belongs include: The environmental data of any generator is collected, and the similarity between the environmental data and the cluster centers of each environmental cluster is calculated. The environmental cluster corresponding to the maximum similarity is taken as the environmental cluster to which the generator belongs.

9. A method for predicting wind power generation in a wind farm according to claim 8, characterized in that: The similarity is negatively correlated with the Euclidean distance between the environmental data and the cluster center of each environmental cluster.

10. A method for predicting wind power generation in a wind farm according to claim 1, characterized in that: The neighborhood range is a rectangular area centered on the generator, and the rectangular area includes a plurality of grid areas. The wind field information of each grid area in the neighborhood range is a two-channel image data.

Images