A wind power prediction integrated optimization method and device based on wind process identification

Abstract

The application relates to a wind power prediction integrated optimization method and device based on wind process identification, which comprises the following steps: substituting numerical weather forecast data at a prediction time into each pre-trained wind power prediction model to obtain power data at the prediction time output by each pre-trained wind power prediction model; and determining optimal power data at the prediction time by using linear regression coefficients corresponding to each pre-trained wind power prediction model at the prediction time and the power data at the prediction time output by each pre-trained wind power prediction model; wherein the linear regression coefficients corresponding to each pre-trained wind power prediction model at the prediction time are obtained based on training data of the pre-trained wind power prediction model. The technical scheme provided by the application can quickly and effectively match and identify similar vectors of wind processes at the prediction time and simultaneously perform integrated optimization on wind power, thereby improving the accuracy of wind power prediction and the prediction method is simple and fast.

Description

Technical Field

[0001] This invention relates to the field of wind power prediction, and specifically to an integrated optimization method and apparatus for wind power prediction based on wind process identification. Background Technology

[0002] With the large-scale grid connection of wind power, the randomness and volatility of wind power output have brought a huge impact on the safe and economical operation of the power system. Wind power forecasting can provide future wind power output in advance, thus serving as an important basis for power system decision optimization. Therefore, wind power forecasting is an important technical means to improve the level of wind power absorption, ensure the safety and stability of the power system, and improve the economic efficiency of grid operation.

[0003] Currently, the wind power prediction methods used in engineering applications are mainly statistical methods. These methods establish a mapping relationship between explanatory variables (such as numerical weather prediction, historical power, meteorological observation data, etc.) and the explained variable (wind power) through statistical models or machine learning (artificial intelligence) models. However, due to the limitations of human cognition and technical level, as well as the complexity of the wind power prediction problem itself, wind power prediction has a large error. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide an integrated optimization method and apparatus for wind power prediction based on wind process identification. This method involves substituting numerical weather forecast data at the prediction time into pre-trained wind power prediction models to obtain the prediction time power data output by each model. The optimal prediction time power data is then determined using the linear regression coefficients of each pre-trained model at the prediction time and the prediction time power data output by each model. This method improves the accuracy of wind power prediction, and its simple and rapid approach allows for widespread application.

[0005] The objective of this invention is achieved through the following technical solution:

[0006] This invention provides an integrated optimization method for wind power prediction based on wind process identification, the improvement of which is that the method includes:

[0007] The numerical weather forecast data at the predicted time is substituted into each pre-trained wind power prediction model to obtain the predicted power data output by each pre-trained wind power prediction model.

[0008] The optimal power data at the prediction time is determined by using the linear regression coefficients of each pre-trained wind power prediction model at the prediction time and the power data at the prediction time output by each pre-trained wind power prediction model.

[0009] The linear regression coefficients of each pre-trained wind power prediction model at the prediction time are obtained based on the training data of each pre-trained wind power prediction model.

[0010] Preferably, the training process for each pre-trained wind power prediction model includes:

[0011] The dataset formed by numerical weather forecast data and corresponding wind power data at each moment in the historical period is randomly resampled using the bootstrap method to obtain y sets of training data.

[0012] The numerical weather forecast data in the y training data are used as the input layer data of each wind power prediction model, and the wind power data corresponding to the numerical weather forecast data in the y training data are used as the output layer data of each wind power prediction model for training, so as to obtain m pre-trained wind power prediction models.

[0013] Each training data set includes q sample data consisting of q random resamplings with replacement, and the number of each wind power prediction model is g, where m = y × g.

[0014] Furthermore, the wind power prediction models include: BP neural network model, support vector regression model, decision tree regression model, and k-nearest neighbor regression model.

[0015] Preferably, the step of determining the optimal prediction time power data using the pre-acquired linear regression matrix for the prediction time and the prediction time power data output by each pre-trained wind power prediction model includes:

[0016] The optimal predicted power data is determined by the following formula.

[0017]

[0018] In the above formula, β represents the predicted power data at the time of the c-th wind power prediction model. c Let β0 be the linear regression coefficient of the c-th wind power prediction model at the prediction time, where β0 = 1.

[0019] Preferably, the process of obtaining the linear regression coefficients of each pre-trained wind power prediction model at the prediction time based on the training data of each pre-trained wind power prediction model includes:

[0020] Select k similar vectors from the wind process vectors at each time point in the historical period that correspond to the wind process vector at the predicted time.

[0021] Substitute the historical weather forecast data corresponding to the k similar vectors into m pre-trained wind power prediction models to obtain the similar power vectors at the prediction time output by each pre-trained wind power prediction model.

[0022] The linear regression coefficients of each pre-trained wind power prediction model at the prediction time are determined by using the similar power vectors at the prediction time output by each pre-trained wind power prediction model.

[0023] Furthermore, selecting k similar vectors from the wind process vectors at each time point within the historical period that correspond to the wind process vector at the predicted time includes:

[0024] Obtain the wind process vectors at each time point within the historical time period, and determine the Euclidean distance between the wind process vector at the predicted time and the wind process vectors at each time point within the historical time period.

[0025] The Euclidean distances between the wind process vector at the predicted time and the wind process vectors at each time in the historical period are sorted in descending order, and the wind process vectors corresponding to the first k Euclidean distances in the descending sequence are selected as the k similar vectors of the wind process vector at the predicted time.

[0026] Furthermore, the step of obtaining the wind process vectors at each moment within the historical time period and determining the Euclidean distance between the wind process vector at the predicted moment and the wind process vectors at each moment within the historical time period includes:

[0027] The Euclidean distance d between the wind process vector at prediction time i and the wind process vector at time j within the historical time period is determined by the following formula. i,j :

[0028]

[0029] In the above formula, j∈(1~N), v i =[v i-b ,v i-b+1 ,...,v i ,...,v i+b-1 ,v i+b ] T v i Let v be the wind process vector at prediction time i. i+b Let v be the wind speed at time i+b, b be the time bandwidth, and v be the wind speed at time i+b. j =[v j-b ,v j-b+1 ,...,v j ,...,v j+b-1 ,v j+b ] T v j Let be the wind process vector at the j-th moment within the historical time period.

[0030] Furthermore, the step of substituting the historical weather forecast data corresponding to the k similar vectors into m pre-trained wind power prediction models to obtain the similar power vectors at the prediction time output by each pre-trained wind power prediction model includes:

[0031] The following formula is used to determine the similar power vector at the prediction time obtained by substituting the historical weather forecast data corresponding to the k similarity vectors into the c-th wind power prediction model.

[0032]

[0033] In the above formula, p cr The wind power data is output after substituting the numerical weather forecast data corresponding to the r-th similar vector among k similar vectors into the c-th wind power prediction model.

[0034] Furthermore, determining the linear regression coefficients of each pre-trained wind power prediction model at the prediction time using the similar power vectors output by each pre-trained wind power prediction model at the prediction time includes:

[0035] The linear regression coefficients [β1...β1] of the first to m-th pre-trained wind power prediction models at the prediction time are determined by the following formula. c ...β m ]:

[0036]

[0037] In the above formula, Let p be the similar power vectors at the prediction time output by the c-th wind power prediction model after substituting the historical weather forecast data corresponding to the k similar vectors into the c-th wind power prediction model. k =[p1...p r ...p k ], p k Let p be the actual wind power vector corresponding to k similar vectors. r Let β be the actual wind power corresponding to the r-th similar vector among k similar vectors. c denoted as the linear regression coefficient of the c-th wind power prediction model at the prediction time.

[0038] This invention provides an integrated optimization device for wind power prediction based on wind process identification, the improvement of which is that the device includes:

[0039] The acquisition module is used to input the numerical weather forecast data at the prediction time into each pre-trained wind power prediction model and obtain the prediction time power data output by each pre-trained wind power prediction model.

[0040] The determination module is used to determine the optimal power data at the prediction time by using the linear regression coefficients of each pre-trained wind power prediction model at the prediction time and the power data at the prediction time output by each pre-trained wind power prediction model.

[0041] The linear regression coefficients of each pre-trained wind power prediction model at the prediction time are obtained based on the training data of each pre-trained wind power prediction model.

[0042] Preferably, the training process for each pre-trained wind power prediction model includes:

[0043] The dataset formed by numerical weather forecast data and corresponding wind power data at each moment in the historical period is randomly resampled using the bootstrap method to obtain y sets of training data.

[0044] The numerical weather forecast data in the y training data are used as the input layer data of each wind power prediction model, and the wind power data corresponding to the numerical weather forecast data in the y training data are used as the output layer data of each wind power prediction model for training, so as to obtain m pre-trained wind power prediction models.

[0045] Each training data set includes q sample data consisting of q random resamplings with replacement, and the number of each wind power prediction model is g, where m = y × g.

[0046] Furthermore, the wind power prediction models include: BP neural network model, support vector regression model, decision tree regression model, and k-nearest neighbor regression model.

[0047] Preferably, the determining module is used for:

[0048] The optimal predicted power data is determined by the following formula.

[0049]

[0050] In the above formula, β represents the predicted power data at the time of the c-th wind power prediction model. c Let β0 be the linear regression coefficient of the c-th wind power prediction model at the prediction time, where β0 = 1.

[0051] Preferably, the process of obtaining the linear regression coefficients of each pre-trained wind power prediction model at the prediction time based on the training data of each pre-trained wind power prediction model includes:

[0052] Select k similar vectors from the wind process vectors at each time point in the historical period that correspond to the wind process vector at the predicted time.

[0053] Substitute the historical weather forecast data corresponding to the k similar vectors into m pre-trained wind power prediction models to obtain the similar power vectors at the prediction time output by each pre-trained wind power prediction model.

[0054] The linear regression coefficients of each pre-trained wind power prediction model at the prediction time are determined by using the similar power vectors at the prediction time output by each pre-trained wind power prediction model.

[0055] Furthermore, selecting k similar vectors from the wind process vectors at each time point within the historical period that correspond to the wind process vector at the predicted time includes:

[0056] Obtain the wind process vectors at each time point within the historical time period, and determine the Euclidean distance between the wind process vector at the predicted time and the wind process vectors at each time point within the historical time period.

[0057] The Euclidean distances between the wind process vector at the predicted time and the wind process vectors at each time in the historical period are sorted in descending order, and the wind process vectors corresponding to the first k Euclidean distances in the descending sequence are selected as the k similar vectors of the wind process vector at the predicted time.

[0058] Furthermore, the step of obtaining the wind process vectors at each moment within the historical time period and determining the Euclidean distance between the wind process vector at the predicted moment and the wind process vectors at each moment within the historical time period includes:

[0059] The Euclidean distance d between the wind process vector at prediction time i and the wind process vector at time j within the historical time period is determined by the following formula. i,j :

[0060]

[0061] In the above formula, j∈(1~N), v i =[v i-b ,v i-b+1 ,...,v i ,...,v i+b-1 ,v i+b ] T v i Let v be the wind process vector at prediction time i. i+bLet v be the wind speed at time i+b, b be the time bandwidth, and v be the wind speed at time i+b. j =[v j-b ,v j-b+1 ,...,v j ,...,v j+b-1 ,v j+b ] T v j Let be the wind process vector at the j-th moment within the historical time period.

[0062] Furthermore, the step of substituting the historical weather forecast data corresponding to the k similar vectors into m pre-trained wind power prediction models to obtain the similar power vectors at the prediction time output by each pre-trained wind power prediction model includes:

[0063] The following formula is used to determine the similar power vector at the prediction time obtained by substituting the historical weather forecast data corresponding to the k similarity vectors into the c-th wind power prediction model.

[0064]

[0065] In the above formula, p cr The wind power data is output after substituting the numerical weather forecast data corresponding to the r-th similar vector among k similar vectors into the c-th wind power prediction model.

[0066] Furthermore, determining the linear regression coefficients of each pre-trained wind power prediction model at the prediction time using the similar power vectors output by each pre-trained wind power prediction model at the prediction time includes:

[0067] The linear regression coefficients [β1...β1] of the first to m-th pre-trained wind power prediction models at the prediction time are determined by the following formula. c ...β m ]:

[0068]

[0069] In the above formula, Let p be the similar power vectors at the prediction time output by the c-th wind power prediction model after substituting the historical weather forecast data corresponding to the k similar vectors into the c-th wind power prediction model. k =[p1...p r ...p k ], p k Let p be the actual wind power vector corresponding to k similar vectors. rLet β be the actual wind power corresponding to the r-th similar vector among k similar vectors. c denoted as the linear regression coefficient of the c-th wind power prediction model at the prediction time.

[0070] Compared with the closest existing technology, the present invention has the following advantages:

[0071] The technical solution provided by this invention involves substituting numerical weather forecast data at the predicted time into each pre-trained wind power prediction model to obtain the predicted power data output by each pre-trained wind power prediction model. The optimal predicted power data is determined by using the linear regression coefficients of each pre-trained wind power prediction model at the predicted time and the predicted power data output by each pre-trained wind power prediction model. This method can quickly and effectively match and identify similar vectors of the wind process at the predicted time while simultaneously optimizing wind power, thereby improving the accuracy of wind power prediction. The prediction method is simple and fast and can be widely applied. Attached Figure Description

[0072] Figure 1 This is a flowchart of an integrated optimization method for wind power prediction based on wind process identification provided by the present invention;

[0073] Figure 2 This is a diagram showing the results of similarity vector identification and matching for four wind processes provided by this invention.

[0074] Figure 3 This is a time series diagram of the predicted and actual power of a wind farm provided by the present invention;

[0075] Figure 4 This is a structural diagram of a wind power prediction integrated optimization device based on wind process identification provided by the present invention. Detailed Implementation

[0076] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0077] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0078] This invention provides an integrated optimization method for wind power prediction based on wind process identification, such as... Figure 1 As shown, the method includes:

[0079] The numerical weather forecast data at the predicted time is substituted into each pre-trained wind power prediction model to obtain the predicted power data output by each pre-trained wind power prediction model.

[0080] The optimal power data at the prediction time is determined by using the linear regression coefficients of each pre-trained wind power prediction model at the prediction time and the power data at the prediction time output by each pre-trained wind power prediction model.

[0081] The linear regression coefficients of each pre-trained wind power prediction model at the prediction time are obtained based on the training data of each pre-trained wind power prediction model.

[0082] In the preferred embodiment of the present invention, the training process of each pre-trained wind power prediction model includes:

[0083] The dataset formed by numerical weather forecast data and corresponding wind power data at each moment in the historical period is randomly resampled using the bootstrap method to obtain y sets of training data.

[0084] The numerical weather forecast data in the y training data are used as the input layer data of each wind power prediction model, and the wind power data corresponding to the numerical weather forecast data in the y training data are used as the output layer data of each wind power prediction model for training, so as to obtain m pre-trained wind power prediction models.

[0085] Each training data set includes q sample data consisting of q random resamplings with replacement, and the number of each wind power prediction model is g, where m = y × g.

[0086] The wind power prediction models include: BP neural network model, support vector regression model, decision tree regression model and k-nearest neighbor regression model.

[0087] In the preferred embodiment of the present invention, determining the optimal prediction time power data by using the linear regression matrix of the prediction time obtained in advance and the prediction time power data output by each pre-trained wind power prediction model includes:

[0088] The optimal predicted power data is determined by the following formula.

[0089]

[0090] In the above formula, β represents the predicted power data at the time of the c-th wind power prediction model. c Let β0 be the linear regression coefficient of the c-th wind power prediction model at the prediction time, where β0 = 1.

[0091] The specific process of obtaining the linear regression coefficients of each pre-trained wind power prediction model at the prediction time based on the training data of each pre-trained wind power prediction model includes:

[0092] Select k similar vectors from the wind process vectors at each time point in the historical period that correspond to the wind process vector at the predicted time.

[0093] Substitute the historical weather forecast data corresponding to the k similar vectors into m pre-trained wind power prediction models to obtain the similar power vectors at the prediction time output by each pre-trained wind power prediction model.

[0094] The linear regression coefficients of each pre-trained wind power prediction model at the prediction time are determined by using the similar power vectors at the prediction time output by each pre-trained wind power prediction model.

[0095] The training data refers to numerical weather forecast data for more than one year and its corresponding wind power data.

[0096] In the preferred embodiment of the present invention, such as Figure 2 As shown, (a), (b), (c) and (d) are the results of the identification and matching of similar vectors for four wind processes. The thick solid line represents the target wind process, and the other thin curves represent the matched similar wind processes. In the examples, the time bandwidth b is 4 hours and the time resolution is 15 minutes. It can be seen from the examples that the matching degree of the wind processes is good.

[0097] The step of selecting k similar vectors from the wind process vectors at each moment in the historical time period that correspond to the wind process vector at the predicted moment includes:

[0098] Obtain the wind process vectors at each time point within the historical time period, and determine the Euclidean distance between the wind process vector at the predicted time and the wind process vectors at each time point within the historical time period.

[0099] The Euclidean distances between the wind process vector at the predicted time and the wind process vectors at each time in the historical period are sorted in descending order, and the wind process vectors corresponding to the first k Euclidean distances in the descending sequence are selected as the k similar vectors of the wind process vector at the predicted time.

[0100] The step of obtaining the wind process vectors at each moment within the historical time period and determining the Euclidean distance between the wind process vector at the predicted moment and the wind process vectors at each moment within the historical time period includes:

[0101] The Euclidean distance d between the wind process vector at prediction time i and the wind process vector at time j within the historical time period is determined by the following formula. i,j :

[0102]

[0103] In the above formula, j∈(1~N), v i =[v i-b ,v i-b+1 ,...,v i ,...,v i+b-1 ,v i+b ] T v i Let v be the wind process vector at prediction time i. i+b Let v be the wind speed at time i+b, b be the time bandwidth, and v be the wind speed at time i+b. j =[v j-b ,v j-b+1 ,...,v j ,...,v j+b-1 ,v j+b ] T v j Let be the wind process vector at the j-th moment within the historical time period.

[0104] Furthermore, the step of substituting the historical weather forecast data corresponding to the k similar vectors into m pre-trained wind power prediction models to obtain the similar power vectors at the prediction time output by each pre-trained wind power prediction model includes:

[0105] The following formula is used to determine the similar power vector at the prediction time obtained by substituting the historical weather forecast data corresponding to the k similarity vectors into the c-th wind power prediction model.

[0106]

[0107] In the above formula, p cr The wind power data is output after substituting the numerical weather forecast data corresponding to the r-th similar vector among k similar vectors into the c-th wind power prediction model.

[0108] Furthermore, determining the linear regression coefficients of each pre-trained wind power prediction model at the prediction time using the similar power vectors output by each pre-trained wind power prediction model at the prediction time includes:

[0109] The linear regression coefficients [β1...β1] of the first to m-th pre-trained wind power prediction models at the prediction time are determined by the following formula. c ...β m ]:

[0110]

[0111] In the above formula, Let p be the similar power vectors at the prediction time output by the c-th wind power prediction model after substituting the historical weather forecast data corresponding to the k similar vectors into the c-th wind power prediction model. k =[p1...p r ...p k ], p k Let p be the actual wind power vector corresponding to k similar vectors. r Let β be the actual wind power corresponding to the r-th similar vector among k similar vectors. c denoted as the linear regression coefficient of the c-th wind power prediction model at the prediction time.

[0112] In the optimal embodiment of this invention, Tables 1-3 present the prediction results for three wind farms in a certain province. The wind power prediction model provides prediction results for 96 points (at a 15-minute resolution) from 00:00 to 23:45 the following day at 8:00 every day. For each wind farm, three differentiated models are provided, namely Models 1-3. The "proposed optimized model" determines the optimal prediction time power data by using the linear regression coefficients of each pre-trained wind power prediction model at the prediction time and the power data output by each pre-trained wind power prediction model at the prediction time. The results in the tables show that the proposed optimized model can significantly improve prediction accuracy. To illustrate the effect more intuitively, as shown... Figure 3 As shown, the time series plot corresponding to wind farm 3 is given. It can be found that the optimization model can utilize the complementary effect of the differences between prediction models to optimize the prediction results.

[0113] Table 1. Wind Farm Prediction Indicators

[0114]

[0115]

[0116] Table 2 Wind Farm Prediction Indicators

[0117]

[0118] Table 3. Wind Farm Prediction Indicators

[0119]

[0120] This invention provides an integrated optimization device for wind power prediction based on wind process identification, such as... Figure 4 As shown, the device includes:

[0121] The acquisition module is used to input the numerical weather forecast data at the prediction time into each pre-trained wind power prediction model and obtain the prediction time power data output by each pre-trained wind power prediction model.

[0122] The determination module is used to determine the optimal power data at the prediction time by using the linear regression coefficients of each pre-trained wind power prediction model at the prediction time and the power data at the prediction time output by each pre-trained wind power prediction model.

[0123] The linear regression coefficients of each pre-trained wind power prediction model at the prediction time are obtained based on the training data of each pre-trained wind power prediction model.

[0124] In the preferred embodiment of the present invention, the training process of each pre-trained wind power prediction model includes:

[0125] The dataset formed by numerical weather forecast data and corresponding wind power data at each moment in the historical period is randomly resampled using the bootstrap method to obtain y sets of training data.

[0126] The numerical weather forecast data in the y training data are used as the input layer data of each wind power prediction model, and the wind power data corresponding to the numerical weather forecast data in the y training data are used as the output layer data of each wind power prediction model for training, so as to obtain m pre-trained wind power prediction models.

[0127] Each training data set includes q sample data consisting of q random resamplings with replacement, and the number of each wind power prediction model is g, where m = y × g.

[0128] The wind power prediction models include: BP neural network model, support vector regression model, decision tree regression model and k-nearest neighbor regression model.

[0129] Specifically, the determining module is used for:

[0130] The optimal predicted power data is determined by the following formula.

[0131]

[0132] In the above formula, β represents the predicted power data at the time of the c-th wind power prediction model. c Let β0 be the linear regression coefficient of the c-th wind power prediction model at the prediction time, where β0 = 1.

[0133] Specifically, the process of obtaining the linear regression coefficients of each pre-trained wind power prediction model at the prediction time based on the training data of each pre-trained wind power prediction model includes:

[0134] Select k similar vectors from the wind process vectors at each time point in the historical period that correspond to the wind process vector at the predicted time.

[0135] Substitute the historical weather forecast data corresponding to the k similar vectors into m pre-trained wind power prediction models to obtain the similar power vectors at the prediction time output by each pre-trained wind power prediction model.

[0136] The linear regression coefficients of each pre-trained wind power prediction model at the prediction time are determined by using the similar power vectors at the prediction time output by each pre-trained wind power prediction model.

[0137] Furthermore, selecting k similar vectors from the wind process vectors at each time point within the historical period that correspond to the wind process vector at the predicted time includes:

[0138] Obtain the wind process vectors at each time point within the historical time period, and determine the Euclidean distance between the wind process vector at the predicted time and the wind process vectors at each time point within the historical time period.

[0139] The Euclidean distances between the wind process vector at the predicted time and the wind process vectors at each time in the historical period are sorted in descending order, and the wind process vectors corresponding to the first k Euclidean distances in the descending sequence are selected as the k similar vectors of the wind process vector at the predicted time.

[0140] The step of obtaining the wind process vectors at each moment within the historical time period and determining the Euclidean distance between the wind process vector at the predicted moment and the wind process vectors at each moment within the historical time period includes:

[0141] The Euclidean distance d between the wind process vector at prediction time i and the wind process vector at time j within the historical time period is determined by the following formula. i,j :

[0142]

[0143] In the above formula, j∈(1~N), v i =[v i-b ,v i-b+1 ,...,v i ,...,v i+b-1 ,v i+b ] T v i Let v be the wind process vector at prediction time i. i+bLet v be the wind speed at time i+b, b be the time bandwidth, and v be the wind speed at time i+b. j =[v j-b ,v j-b+1 ,...,v j ,...,v j+b-1 ,v j+b ] T v j Let be the wind process vector at the j-th moment within the historical time period.

[0144] Furthermore, the step of substituting the historical weather forecast data corresponding to the k similar vectors into m pre-trained wind power prediction models to obtain the similar power vectors at the prediction time output by each pre-trained wind power prediction model includes:

[0145] The following formula is used to determine the similar power vector at the prediction time obtained by substituting the historical weather forecast data corresponding to the k similarity vectors into the c-th wind power prediction model.

[0146]

[0147] In the above formula, p cr The wind power data is output after substituting the numerical weather forecast data corresponding to the r-th similar vector among k similar vectors into the c-th wind power prediction model.

[0148] Furthermore, determining the linear regression coefficients of each pre-trained wind power prediction model at the prediction time using the similar power vectors output by each pre-trained wind power prediction model at the prediction time includes:

[0149] The linear regression coefficients [β1...β1] of the first to m-th pre-trained wind power prediction models at the prediction time are determined by the following formula. c ...β m ]:

[0150]

[0151] In the above formula, Let p be the similar power vectors at the prediction time output by the c-th wind power prediction model after substituting the historical weather forecast data corresponding to the k similar vectors into the c-th wind power prediction model. k =[p1...p r ...p k ], p k Let p be the actual wind power vector corresponding to k similar vectors. rLet β be the actual wind power corresponding to the r-th similar vector among k similar vectors. c denoted as the linear regression coefficient of the c-th wind power prediction model at the prediction time.

[0152] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0153] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0154] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0155] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0156] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A wind power prediction integrated optimization method based on wind process identification, characterized in that, The method includes: The numerical weather forecast data at the predicted time is substituted into each pre-trained wind power prediction model to obtain the predicted power data output by each pre-trained wind power prediction model. The optimal power data at the prediction time is determined by using the linear regression coefficients of each pre-trained wind power prediction model at the prediction time and the power data at the prediction time output by each pre-trained wind power prediction model. The linear regression coefficients of each pre-trained wind power prediction model at the prediction time are obtained based on the training data of each pre-trained wind power prediction model. The optimal predicted power data is determined by the following formula. : In the above formula, For the first Power data at the predicted time output by a wind power prediction model; For the first The linear regression coefficients of a wind power prediction model at the prediction time. ; The process of obtaining the linear regression coefficients of each pre-trained wind power prediction model at the prediction time based on the training data of each pre-trained wind power prediction model includes: Select the wind process vector corresponding to the wind process vector at the prediction time from the wind process vectors at each moment within the historical time period. 1 similar vector; Will Substituting the historical numerical weather forecast data corresponding to each similar vector into... In each pre-trained wind power prediction model, obtain the similar power vectors at the prediction time output by each pre-trained wind power prediction model. The linear regression coefficients of each pre-trained wind power prediction model at the prediction time are determined by using the similar power vectors at the prediction time output by each pre-trained wind power prediction model.

2. The method as described in claim 1, characterized in that, The training process for each pre-trained wind power prediction model includes: The dataset is formed by numerical weather forecast data and corresponding wind power data at various times within a historical period. Group bootstrap random resampling to obtain Group training data; Will In the training dataset, numerical weather prediction data were used as the input layer data for each wind power prediction model. The wind power data corresponding to the numerical weather prediction data in the training dataset were used as the output layer data for training each wind power prediction model to obtain... A pre-trained wind power prediction model; Each set of training data includes Composed of random resampling with replacement The number of sample data and the number of wind power prediction models are [number] times. , .

3. The method as described in claim 2, characterized in that, The wind power prediction models include: BP neural network model, support vector regression model, decision tree regression model, and... k Nearest neighbor regression model.

4. The method as described in claim 1, characterized in that, The step involves selecting the wind process vector corresponding to the wind process vector at the predicted time from the wind process vectors at various times within the historical period. Similar vectors include: Obtain the wind process vectors at each time point within the historical time period, and determine the Euclidean distance between the wind process vector at the predicted time and the wind process vectors at each time point within the historical time period. The wind process vector at the predicted time is sorted in descending order by the Euclidean distance between it and the wind process vectors at all times within the historical time period, and the first wind process vector in the descending order is selected. The wind process vector corresponding to each Euclidean distance is used as the wind process vector at the prediction time. There are 10 similar vectors.

5. The method as described in claim 4, characterized in that, The step of obtaining wind process vectors at each moment within a historical time period and determining the Euclidean distance between the wind process vector at the predicted moment and the wind process vectors at each moment within the historical time period includes: The Euclidean distance between the wind process vector at prediction time i and the wind process vector at time j within the historical time period is determined by the following formula. : In the above formula, , , Let i be the wind process vector at prediction time i. For the first Wind speed at that moment, For time bandwidth, , For the first time in the historical period The wind process vector at each moment.

6. The method as described in claim 1, characterized in that, The Substituting the historical numerical weather forecast data corresponding to each similar vector into... In each pre-trained wind power prediction model, the similar power vectors at the prediction time output by each pre-trained wind power prediction model are obtained, including: Determine by the following formula Substituting the historical weather forecast data corresponding to the nth similar vector into the nth... The first wind power prediction model obtained The similar power vectors output by each wind power prediction model at the prediction time. : In the above formula, for Among the similar vectors, the th... Substituting the numerical weather forecast data corresponding to the nth similarity vector into the nth... The wind power data output after a wind power prediction model.

7. The method as described in claim 1, characterized in that, The step of determining the linear regression coefficients of each pre-trained wind power prediction model at the prediction time using the similar power vectors output by each pre-trained wind power prediction model at the prediction time includes: Determine the first to the second according to the following formula The linear regression coefficients of a pre-trained wind power prediction model at the prediction time. : In the above formula, , for Substituting the historical weather forecast data corresponding to the nth similar vector into the nth... After the first wind power prediction model The similar power vectors output by each wind power prediction model at the prediction time. , for The actual wind power vector corresponding to each similar vector. for Among the similar vectors, the th... The actual wind power corresponding to each similar vector For the first The linear regression coefficients of a wind power prediction model at the prediction time.

8. A wind power prediction integrated optimization device based on wind process identification, characterized in that, The device includes: The acquisition module is used to input the numerical weather forecast data at the prediction time into each pre-trained wind power prediction model and obtain the prediction time power data output by each pre-trained wind power prediction model. The determination module is used to determine the optimal power data at the prediction time by using the linear regression coefficients of each pre-trained wind power prediction model at the prediction time and the power data at the prediction time output by each pre-trained wind power prediction model. The linear regression coefficients of each pre-trained wind power prediction model at the prediction time are obtained based on the training data of each pre-trained wind power prediction model. The determining module is used for: The optimal predicted power data is determined by the following formula. : In the above formula, For the first Power data at the predicted time output by a wind power prediction model; For the first The linear regression coefficients of a wind power prediction model at the prediction time. ; The process of obtaining the linear regression coefficients of each pre-trained wind power prediction model at the prediction time based on the training data of each pre-trained wind power prediction model includes: Select the wind process vector corresponding to the wind process vector at the prediction time from the wind process vectors at each moment within the historical time period. 1 similar vector; Will Substituting the historical numerical weather forecast data corresponding to each similar vector into... In each pre-trained wind power prediction model, obtain the similar power vectors at the prediction time output by each pre-trained wind power prediction model. The linear regression coefficients of each pre-trained wind power prediction model at the prediction time are determined by using the similar power vectors at the prediction time output by each pre-trained wind power prediction model.

9. The apparatus as claimed in claim 8, characterized in that, The training process for each pre-trained wind power prediction model includes: The dataset is formed by numerical weather forecast data and corresponding wind power data at various times within a historical period. Group bootstrap random resampling to obtain Group training data; Will In the training dataset, numerical weather prediction data were used as the input layer data for each wind power prediction model. The wind power data corresponding to the numerical weather prediction data in the training dataset were used as the output layer data for training each wind power prediction model to obtain... A pre-trained wind power prediction model; Each set of training data includes Secondary random resampling with replacement The number of sample data and the number of wind power prediction models are [number] times. , .

10. The apparatus as claimed in claim 9, characterized in that, The wind power prediction models include: BP neural network model, support vector regression model, decision tree regression model, and... k Nearest neighbor regression model.

11. The apparatus as claimed in claim 8, characterized in that, The step involves selecting the wind process vector corresponding to the wind process vector at the predicted time from the wind process vectors at various times within the historical period. Similar vectors include: Obtain the wind process vectors at each time point within the historical time period, and determine the Euclidean distance between the wind process vector at the predicted time and the wind process vectors at each time point within the historical time period. The wind process vector at the predicted time is sorted in descending order by the Euclidean distance between it and the wind process vectors at all times within the historical time period, and the first wind process vector in the descending order is selected. The wind process vector corresponding to each Euclidean distance is used as the wind process vector at the prediction time. There are 10 similar vectors.

12. The apparatus as claimed in claim 11, characterized in that, The step of obtaining wind process vectors at each moment within a historical time period and determining the Euclidean distance between the wind process vector at the predicted moment and the wind process vectors at each moment within the historical time period includes: The Euclidean distance between the wind process vector at prediction time i and the wind process vector at time j within the historical time period is determined by the following formula. : In the above formula, , , Let i be the wind process vector at prediction time i. For the first Wind speed at that moment, For time bandwidth, , For the first time in the historical period The wind process vector at each moment.

13. The apparatus as claimed in claim 8, characterized in that, The Substituting the historical numerical weather forecast data corresponding to each similar vector into... In each pre-trained wind power prediction model, the similar power vectors at the prediction time output by each pre-trained wind power prediction model are obtained, including: Determine by the following formula Substituting the historical weather forecast data corresponding to the nth similar vector into the nth... The first wind power prediction model obtained The similar power vectors output by each wind power prediction model at the prediction time. : In the above formula, for Among the similar vectors, the th... Substituting the numerical weather forecast data corresponding to the nth similarity vector into the nth... The wind power data output after a wind power prediction model.

14. The apparatus as claimed in claim 8, characterized in that, The step of determining the linear regression coefficients of each pre-trained wind power prediction model at the prediction time using the similar power vectors output by each pre-trained wind power prediction model at the prediction time includes: Determine the first to the second according to the following formula The linear regression coefficients of a pre-trained wind power prediction model at the prediction time. : In the above formula, , for Substituting the historical weather forecast data corresponding to the nth similar vector into the nth... After the first wind power prediction model The similar power vectors output by each wind power prediction model at the prediction time. , for The actual wind power vector corresponding to each similar vector. for Among the similar vectors, the th... The actual wind power corresponding to each similar vector For the first The linear regression coefficients of a wind power prediction model at the prediction time.

Images