Wind field forecasting method and device, storage medium and computer program product

Abstract

The invention discloses a wind field forecasting method and device, a storage medium and a computer program product, and relates to the field of artificial intelligence, and the wind field forecasting method comprises the steps: inputting the historical data of a wind field in a historical time period and the historical data of a target meteorological element in the historical time period into a time sequence network, the wind field output of each time period and the target meteorological element output of each time period in the historical time period are obtained; fusing the wind field output of the first time period in the historical time period and the target meteorological element output of the first time period to determine a first fusion result of the first time period; and according to the wind field output of the second time period in the historical time period, the target meteorological element output of the second time period, and the first fusion result, performing wind field forecasting of the third time period. According to the technical scheme, the problem that wind field forecasting cannot be accurately carried out due to the fact that useful information extracted from wind field historical data is relatively limited is solved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and more specifically, to a wind farm forecasting method and device, a storage medium, and a computer program product. Background Art

[0002] When relevant technologies are used to forecast wind fields, the useful information extracted from historical wind field data is relatively limited. Traditional models find it difficult to capture the nonlinear relationship between wind fields and other relevant meteorological elements (such as temperature, humidity, air pressure, etc.), further limiting the accuracy of wind field forecasts.

[0003] In the related technology, due to the relatively limited useful information extracted from the historical data of the wind farm, the wind farm forecast cannot be accurately carried out. At present, no effective solution has been proposed.

[0004] Therefore, it is necessary to improve the related technology to overcome the above-mentioned defects in the related technology. Summary of the invention

[0005] The embodiments of the present application provide a wind farm forecasting method and device, a storage medium, and a computer program product to at least solve the problem in the related art that the wind farm forecast cannot be accurately performed due to the relatively limited useful information extracted from the historical data of the wind farm.

[0006] According to one aspect of an embodiment of the present application, a wind field forecasting method is provided, comprising: inputting historical data of wind fields within a historical time period and historical data of target meteorological elements within the historical time period into a time series network to obtain wind field outputs for each time period within the historical time period and target meteorological element outputs for each time period, wherein the target meteorological elements include at least one of the following: temperature, humidity, air pressure, and precipitation; fusing the wind field outputs for a first time period within the historical time period and the target meteorological element outputs for the first time period to determine a first fusion result for the first time period; forecasting the wind field for a third time period based on the wind field outputs for a second time period within the historical time period and the target meteorological element outputs for the second time period, as well as the first fusion result, wherein each time period includes: the first time period, the second time period, and the third time period; the first time period is earlier than the second time period, and the second time period is earlier than the third time period, and the first time period, the second time period, and the third time period are adjacent in sequence.

[0007] In an exemplary embodiment, before obtaining the wind field output of each time period within the historical time period and the target meteorological element output of each time period, the method further includes: obtaining a one-dimensional feature vector of the wind field historical data in the first time period, and a one-dimensional feature vector of the historical data of the target meteorological element in the first time period; determining the wind field output of the first time period, the target meteorological element output of the first time period, the conditional feature matrix of the wind field historical data output through the timing network in the first time period, and the conditional feature matrix of the historical data of the target meteorological element output through the timing network in the first time period based on the one-dimensional feature vector of the wind field historical data in the first time period and the one-dimensional feature vector of the historical data of the target meteorological element in the first time period.

[0008] In an exemplary embodiment, the wind farm output of each time period within the historical time period and the target meteorological element output of each time period are obtained, including: obtaining a one-dimensional feature vector of the wind farm historical data in each time period, a one-dimensional feature vector of the target meteorological element historical data in each time period, the wind farm output of the first time period, the target meteorological element output of the first time period, a conditional feature matrix of the wind farm historical data output through the timing network in the first time period, and a conditional feature matrix of the target meteorological element historical data output through the timing network in the first time period; when j is greater than 1, the wind farm output of each time period is determined according to the following formula And determine the target meteorological element output for each time period according to the following formula: Where Tj is the jth time period, Tj-1 is the j-1th time period, j is a positive integer, and WD Tj is the one-dimensional feature vector of the wind farm historical data in the Tj time period, WV Tj is the one-dimensional feature vector of the historical data of the target meteorological element in the Tj time period, is the wind farm output in the Tj time period, is the target meteorological element output in the Tj time period, is the wind farm output in the Tj-1 time period, is the target meteorological element output in the Tj-1 time period, is the conditional characteristic matrix of the wind farm historical data output through the timing network in the Tj time period, is the conditional feature matrix of the historical data of the target meteorological element output by the time series network in the Tj time period, is the conditional characteristic matrix of the wind farm historical data output by the timing network in the Tj-1 time period, It is the conditional feature matrix output by the time series network of the historical data of the target meteorological element in the Tj-1 time period.

[0009] In an exemplary embodiment, the wind farm output of the first time period within the historical time period and the target meteorological element output of the first time period are fused to determine a first fusion result of the first time period, including: obtaining the wind farm output of the first time period and the target meteorological element output of the first time period; determining the first fusion result according to the following formula

[0010] and Among them, σ, tanh and Softmax represent three different activation functions, CINF T0 is a normally distributed random matrix or a zero matrix, is the wind farm output in the first time period, Output of target meteorological elements for the first time period.

[0011] In an exemplary embodiment, the wind farm forecast for a third time period is performed based on the wind farm output of a second time period within the historical time period and the target meteorological element output of the second time period, as well as the first fusion result, including: obtaining a conditional feature matrix of the wind farm historical data output through the time series network in the second time period and a conditional feature matrix of the target meteorological element historical data output through the time series network in the second time period; when j=2, determining the result of the wind farm forecast according to the following formula Where W and b represent training parameters, ReLU represents activation function, α, β, γ are parameters, 0≤α≤1, 0≤β≤1, 0≤γ≤1, is the wind farm output in the Tj time period, is the target meteorological element output in the Tj time period, CINF Tj-1 is the j-1th fusion result in time period Tj-1, is the conditional characteristic matrix of the wind farm historical data output through the timing network in the Tj time period, It is the conditional feature matrix output by the time series network of the historical data of the target meteorological element in the Tj time period, where j is a positive integer.

[0012] In an exemplary embodiment, before inputting the historical data of the wind farm within a historical time period and the historical data of the target meteorological elements within the historical time period into the time series network, the method also includes: inputting the historical data of the wind farm and the historical data of the target meteorological elements into the embedding layer of the target neural network, outputting a first feature matrix of the historical data of the wind farm with a unified dimension and a second feature matrix of the historical data of the target meteorological elements with a unified dimension; converting the first feature matrix and the second feature matrix into a first one-dimensional feature vector and a second one-dimensional feature vector through a fully connected layer of the target neural network.

[0013] In an exemplary embodiment, after fusing the wind farm output of the first time period within the historical time period and the target meteorological element output of the first time period to determine a first fusion result of the first time period, the method further includes: obtaining the first fusion result, the wind farm output of the second time period, and the target meteorological element output of the second time period; when j=2, determining the second fusion result of the second time period according to the following formula

[0014] and Among them, σ, tanh and Softmax represent three different activation functions, CINF Tj is the fusion result of the jth time period, CINF Tj-1 is the fusion result of the j-1th time period, is the wind farm output in the Tj time period, It is the target meteorological element output in the Tj time period.

[0015] According to another aspect of an embodiment of the present application, a wind field forecasting device is also provided, including: an acquisition module, used to input the historical data of the wind field within a historical time period and the historical data of the target meteorological elements within the historical time period into a timing network to obtain the wind field output of each time period within the historical time period and the target meteorological element output of each time period, wherein the target meteorological element includes at least one of the following: temperature, humidity, air pressure, precipitation; a fusion module, used to fuse the wind field output of a first time period within the historical time period and the target meteorological element output of the first time period to determine a first fusion result of the first time period; a forecasting module, used to perform a wind field forecast for a third time period based on the wind field output of a second time period within the historical time period and the target meteorological element output of the second time period, and the first fusion result, wherein each time period includes: the first time period, the second time period, and the third time period; the first time period is earlier than the second time period, and the second time period is earlier than the third time period, and the first time period, the second time period and the third time period are adjacent in sequence.

[0016] According to another aspect of the embodiments of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned wind field forecasting method when running.

[0017] According to another aspect of an embodiment of the present application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the wind field forecasting method through the computer program.

[0018] According to another aspect of the embodiments of the present application, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps of the method described in each embodiment of the present application are implemented.

[0019] Through this application, the historical data of the wind farm in the historical time period and the historical data of the target meteorological elements in the historical time period are processed through a time series network to obtain the wind farm output and the target meteorological element output of each time period in the historical time period; the first fusion result is determined according to the wind farm output of the first time period and the target meteorological element output of the first time period; the wind farm forecast of the third time period is performed according to the wind farm output of the second time period and the target meteorological element output of the second time period and the first fusion result. This solves the problem in the related art that the wind farm forecast cannot be accurately performed due to the relatively limited useful information extracted from the historical data of the wind farm. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0021] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0022] Figure 1 It is a hardware structure block diagram of a computer terminal of a wind field forecasting method according to an embodiment of the present application;

[0023] Figure 2 is a flow chart of a wind farm forecasting method according to an embodiment of the present application;

[0024] Figure 3 is a schematic diagram of a wind field forecasting method according to an embodiment of the present application;

[0025] Figure 4 is a schematic diagram of calculation of fusion results of a wind field forecasting method according to an embodiment of the present application;

[0026] Figure 5 It is a structural block diagram of a wind field forecasting device according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] The method embodiments provided in the embodiments of the present application can be executed in a computer terminal or a similar computing device. Taking running on a computer terminal as an example, Figure 1 FIG. 1 is a hardware structure block diagram of a computer terminal of a wind field forecasting method according to an embodiment of the present application. Figure 1 As shown, the computer terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor (Central Processing Unit, MCU) or a programmable logic device (Field Programmable Gate Array, FPGA)) and a memory 104 for storing data, wherein the above-mentioned computer terminal may also include a transmission device 106 and an input and output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above-mentioned computer terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0030] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the wind farm forecasting method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories can be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0031] A wireless network provided by a communication provider of a computer terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0032] In this embodiment, a wind field forecasting method is provided. Figure 2is a flow chart of a wind field forecasting method according to an embodiment of the present application, such as Figure 2 As shown, the process includes the following steps:

[0033] Step S202, inputting the historical data of the wind farm in the historical time period and the historical data of the target meteorological element in the historical time period into the time series network to obtain the wind farm output of each time period in the historical time period and the target meteorological element output of each time period, wherein the target meteorological element includes at least one of the following: temperature, humidity, air pressure, precipitation;

[0034] Step S204, fusing the wind field output of the first time period within the historical time period and the target meteorological element output of the first time period to determine a first fusion result of the first time period;

[0035] Step S206, performing wind field forecast for a third time period based on the wind field output of a second time period within the historical time period, the target meteorological element output of the second time period, and the first fusion result, wherein each time period includes: the first time period, the second time period, and the third time period; the first time period is earlier than the second time period, and the second time period is earlier than the third time period, and the first time period, the second time period and the third time period are adjacent in sequence.

[0036] Through the above steps, the historical data of the wind field in the historical time period and the historical data of the target meteorological elements in the historical time period are processed through the time series network to obtain the wind field output and the target meteorological element output of each time period in the historical time period; the first fusion result is determined according to the wind field output of the first time period and the target meteorological element output of the first time period; the wind field forecast of the third time period is performed according to the wind field output of the second time period and the target meteorological element output of the second time period and the first fusion result. This solves the problem in the related art that the wind field forecast cannot be accurately performed due to the relatively limited useful information extracted from the historical data of the wind field.

[0037] It should be clarified that the above-mentioned target meteorological elements include at least one of the following: temperature, humidity, air pressure, and precipitation. When using multiple target meteorological elements, it is necessary to splice the above-mentioned multiple target meteorological elements in the time dimension, that is, for each time period, vectors or matrices such as temperature, humidity, air pressure, and precipitation are arranged in order to form a composite matrix containing multiple target meteorological elements. It should be clarified that the splicing order of the above-mentioned multiple target meteorological elements is not specifically specified.

[0038] In an exemplary embodiment, before obtaining the wind field output of each time period within the historical time period and the target meteorological element output of each time period, the method further includes: obtaining a one-dimensional feature vector of the wind field historical data in the first time period, and a one-dimensional feature vector of the historical data of the target meteorological element in the first time period; determining the wind field output of the first time period, the target meteorological element output of the first time period, the conditional feature matrix of the wind field historical data output through the timing network in the first time period, and the conditional feature matrix of the historical data of the target meteorological element output through the timing network in the first time period based on the one-dimensional feature vector of the wind field historical data in the first time period and the one-dimensional feature vector of the historical data of the target meteorological element in the first time period.

[0039] Optionally, assume that the one-dimensional feature vector of the wind farm historical data in the first time period is WD T1 The one-dimensional feature vector of the historical data of the target meteorological element in the first time period is WV T1 , according to the formula and Determine the wind farm output for the first time period Output of target meteorological elements for the first time period The conditional characteristic matrix of wind farm historical data output by the timing network in the first time period The conditional feature matrix of the historical data of the target meteorological element output by the time series network in the first time period in, is a zero matrix or a normally distributed random matrix.

[0040] In an exemplary embodiment, the wind farm output of each time period within the historical time period and the target meteorological element output of each time period are obtained, including: obtaining a one-dimensional feature vector of the wind farm historical data in each time period, a one-dimensional feature vector of the target meteorological element historical data in each time period, the wind farm output of the first time period, the target meteorological element output of the first time period, a conditional feature matrix of the wind farm historical data output through the timing network in the first time period, and a conditional feature matrix of the target meteorological element historical data output through the timing network in the first time period; when j is greater than 1, the wind farm output of each time period is determined according to the following formula And determine the target meteorological element output for each time period according to the following formula: Where Tj is the jth time period, Tj-1 is the j-1th time period, j is a positive integer, and WD Tjis the one-dimensional feature vector of the wind farm historical data in the Tj time period, WV Tj is the one-dimensional feature vector of the historical data of the target meteorological element in the Tj time period, is the wind farm output in the Tj time period, is the target meteorological element output in the Tj time period, is the wind farm output in the Tj-1 time period, is the target meteorological element output in the Tj-1 time period, is the conditional characteristic matrix of the wind farm historical data output through the timing network in the Tj time period, is the conditional feature matrix of the historical data of the target meteorological element output by the time series network in the Tj time period, is the conditional characteristic matrix of the wind farm historical data output by the timing network in the Tj-1 time period, It is the conditional feature matrix output by the time series network of the historical data of the target meteorological element in the Tj-1 time period.

[0041] Optionally, the first time period to the third time period is represented as T1-T3, and the one-dimensional feature vector of the wind farm historical data in T1-T3 is WD T1 , WD T2 , WD T3 The one-dimensional feature vector of the historical data of the target meteorological element in T1-T3 is WV T1 , WV T2 , WV T3 , where WD T1 , W.D. T2 , W.D. T3 , W.V. T1 , W.V. T2 , W.V. T3 ∈R Dimension When j = 2, the wind farm output of T2 is determined according to the following formula: And determine the target meteorological element output of T2 according to the following formula in, is the wind farm output of T1, is the target meteorological element output of T1, is the conditional feature matrix of the wind farm historical data output through the time series network at T1, It is the conditional feature matrix of the historical data of the target meteorological element output through the time series network at T1. By analogy, the wind field output and the target meteorological element output of each subsequent time period can be determined.

[0042] In an exemplary embodiment, the wind farm output of the first time period within the historical time period and the target meteorological element output of the first time period are fused to determine a first fusion result of the first time period, including: obtaining the wind farm output of the first time period and the target meteorological element output of the first time period; determining the first fusion result according to the following formula

[0043] and Among them, σ, tanh and Softmax represent three different activation functions, CINF T0 is a normally distributed random matrix or a zero matrix, is the wind farm output in the first time period, is the target meteorological element output for the first time period. Optionally, the wind field output for the first time period is The target meteorological element output for the first time period is: Determine the first fusion result CINF according to the formula of this embodiment T1 .in, and It is the hidden state matrix obtained by processing the historical data of wind farm and historical data of target meteorological elements through the time series network. and Contains the characteristic information of the historical data of the wind farm and the historical data of the target meteorological elements in T1. In , σ is the sigmoid activation function, which is used to perform nonlinear transformation on the wind farm output to determine which information is important in the current time period, i.e., the first time period, and which should be filtered out. In , the tanh activation function processes the output of the target meteorological element to construct an auxiliary matrix, which will provide additional information for a better understanding of the changing trend of the wind field. In , the Softmax function is used to calculate the relative importance of the target meteorological element output to the wind field forecast. T1 is a weight matrix, δ T1 Determine the impact of the target meteorological element output in the T1 time period on the wind field forecast.

[0044] In an exemplary embodiment, the wind farm forecast for a third time period is performed based on the wind farm output of a second time period within the historical time period and the target meteorological element output of the second time period, as well as the first fusion result, including: obtaining a conditional feature matrix of the wind farm historical data output through the time series network in the second time period and a conditional feature matrix of the target meteorological element historical data output through the time series network in the second time period; when j=2, determining the result of the wind farm forecast according to the following formula Where W and b represent training parameters, ReLU represents activation function, α, β, γ are parameters, 0≤α≤1, 0≤β≤1, 0≤γ≤1, is the wind farm output in the Tj time period, is the target meteorological element output in the Tj time period, CINF Tj-1 is the j-1th fusion result in time period Tj-1, is the conditional characteristic matrix of the wind farm historical data output through the timing network in the Tj time period, It is the conditional feature matrix output by the time series network of the historical data of the target meteorological element in the Tj time period, where j is a positive integer.

[0045] Optionally, when j=2, according to the wind farm output in the second time period Output of target meteorological elements for the second time period The conditional feature matrix of wind farm historical data output by the time series network in the T2 period The conditional feature matrix of the historical data of the target meteorological element output by the time series network in the T2 period And the first fusion result CINF T1 , the wind farm forecast result for the second time period is determined by the formula of this embodiment. Similarly, to determine the wind farm forecast result for the jth time period, it is necessary to obtain the wind farm output for the jth time period. Output of target meteorological elements for the jth time period The j-1th fusion result CINF of the j-1th time period Tj-1 It should be noted that when j = 1, it means that there is no fusion result CINF for the previous time period. T0 , then CINF T0 Determine a zero matrix or a random matrix that conforms to the normal distribution, and output the wind field in the first time period Output of target meteorological elements for the first time period The conditional feature matrix of wind farm historical data output through the time series network in the T1 period The conditional feature matrix of the historical data of the target meteorological element output by the time series network in the T1 period And the fusion result CINF T0 The result of the wind field forecast for the first time period is determined.

[0046] In an exemplary embodiment, before inputting the historical data of the wind farm within a historical time period and the historical data of the target meteorological elements within the historical time period into the time series network, the method also includes: inputting the historical data of the wind farm and the historical data of the target meteorological elements into the embedding layer of the target neural network, outputting a first feature matrix of the historical data of the wind farm with a unified dimension and a second feature matrix of the historical data of the target meteorological elements with a unified dimension; converting the first feature matrix and the second feature matrix into a first one-dimensional feature vector and a second one-dimensional feature vector through a fully connected layer of the target neural network.

[0047] Optionally, assuming that the historical data of the wind farm is a three-dimensional matrix and the historical data of the target meteorological element is a two-dimensional matrix, after being processed by the embedding layer of the target neural network, the historical data of the wind farm is converted into a first feature matrix, and the historical data of the target meteorological element is converted into a second feature matrix, wherein the first feature matrix and the second feature matrix have a uniform dimension. The data in the first feature matrix and the second feature matrix are converted into a one-dimensional feature vector, i.e., a first one-dimensional feature vector and a second one-dimensional feature vector, through the fully connected layer of the target neural network. It should be clear that the embedding layer of the target neural network in this embodiment includes a convolutional neural network, batch normalization, and an activation function.

[0048] In an exemplary embodiment, after fusing the wind farm output of the first time period within the historical time period and the target meteorological element output of the first time period to determine a first fusion result of the first time period, the method further includes: obtaining the first fusion result, the wind farm output of the second time period, and the target meteorological element output of the second time period; when j=2, determining the second fusion result of the second time period according to the following formula

[0049] and Among them, σ, tanh and Softmax represent three different activation functions, CINF Tj is the fusion result of the jth time period, CINF Tj-1 is the fusion result of the j-1th time period, is the wind farm output in the Tj time period, It is the target meteorological element output in the Tj time period.

[0050] When the first fusion result is determined and j=2, the second fusion result of the second time period can be determined by the formula of this embodiment according to the first fusion result, the wind field output of the second time period, and the target meteorological element output of the second time period. Similarly, the third fusion result of the third time period is determined according to the second fusion result of the second time period, the wind field output of the third time period, and the target meteorological element output of the third time period. Similarly, the fusion result of the current time period is determined according to the fusion result of the previous time period, the wind field output of the current time period, and the target meteorological element output of the current time period.

[0051] In order to better understand the process of the above-mentioned wind farm forecasting method, the implementation method of the above-mentioned wind farm forecasting is described below in combination with an optional embodiment, but it is not used to limit the technical solution of the embodiment of the present application.

[0052] Figure 3 is a schematic diagram of a wind field forecasting method according to an embodiment of the present application, such as Figure 3 As shown, specifically including the following:

[0053] This embodiment uses the historical data of wind fields and other historical data of meteorological elements from the first time period to the fourth time period as input, and outputs the wind field forecast data for the fifth time period, wherein other meteorological elements include but are not limited to temperature, humidity, air pressure, etc. When multiple target meteorological elements are used, it is necessary to first splice these multiple target meteorological elements in the time dimension and synthesize them into a matrix as input.

[0054] The historical data of the wind farm within the historical time period and the historical data of the target meteorological elements within the historical time period are input into the embedding layer of the target neural network, and the first feature matrix of the historical data of the wind farm with unified dimensions and the second feature matrix of the historical data of the target meteorological elements with unified dimensions are output, wherein the embedding layer includes a convolutional neural network, batch normalization and an activation function. The first feature matrix and the second feature matrix are converted into the first one-dimensional feature vector and the second one-dimensional feature vector through the fully connected layer of the target neural network, and the first one-dimensional feature vector from the first time period to the fourth time period is represented as WD T1 ,WD T2 ,WD T3 ,WD T4 , the second one-dimensional feature vector from the first time period to the fourth time period is represented by WV T1 ,WV T2 ,WV T3 ,WV T4 Among them, WD T1 ,WD T2 ,WD T3 ,WD T4 ,WV T1 ,WV T2 ,WVT3 ,WV T4 ∈R Dimension .

[0055] The first one-dimensional feature vector from the first time period to the fourth time period and the second one-dimensional feature vector from the first time period to the fourth time period are input into the time series network running in parallel to obtain the wind field output from the first time period to the fourth time period and the target meteorological element output from the first time period to the fourth time period. The wind field output of the first time period and the target meteorological element output of the first time period are fused to determine the first fusion result of the first time period. The wind field output of the second time period, the target meteorological element output of the second time period and the first fusion result are fused to obtain the fusion result of the second time period. The wind field output of the third time period, the target meteorological element output of the third time period and the second fusion result are fused to obtain the fusion result of the third time period. According to the wind field output of the fourth time period, the target meteorological element output of the fourth time period and the third fusion result, the wind field forecast of the fifth time period is performed.

[0056] The above embodiment describes the entire process of wind field forecasting. In order to better understand the calculation process of the fusion result in the above wind field forecasting method, the calculation process of the fusion result is explained in combination with the embodiment below, but it is not used to limit the technical solution of the embodiment of the present application.

[0057] Figure 4 FIG. 1 is a schematic diagram of calculating the fusion results of a wind field forecasting method according to an embodiment of the present application. Figure 4 As shown, specifically including the following:

[0058] The first one-dimensional eigenvectors of the wind farm historical data from the first time period to the fourth time period are WD T1 ,WD T2 ,WD T3 ,WD T4 The second one-dimensional feature vector of the historical data of the target meteorological element from the first time period to the fourth time period is represented by WV T1 ,WV T2 ,WV T3 ,WV T4 Among them, WD T1 ,WD T2 ,WD T3 ,WD T4 ,WV T1 ,WV T2 ,WV T3 ,WV T4 ∈R Dimension According to the formula Determine the wind farm output from the first time period to the fourth time period, the target meteorological element output from the first time period to the fourth time period, the conditional feature matrix of the wind farm historical data output through the timing network from the first time period to the fourth time period, and the conditional feature matrix of the target meteorological element historical data output through the timing network from the first time period to the fourth time period, where j is a positive integer, and in this embodiment 1≤j<4. It should be clear that when j=1, is a zero matrix or a normally distributed random matrix.

[0059] The fusion result of the current time period is determined based on the fusion result of the previous time period, the wind field output of the current time period, the target meteorological element output of the current time period, and the following formula

[0060]

[0061] Among them, W and b represent training parameters, σ, tanh and Softmax represent three different activation functions. and It is the hidden state matrix obtained by processing the historical data of wind farm and historical data of target meteorological elements through the time series network. and Contains the characteristic information of the historical data of the wind field and the historical data of the target meteorological elements in the Tj time period. In , σ is the sigmoid activation function, which is used to perform nonlinear transformation on the wind farm output to determine which information is important in the current time period, i.e., the first time period, and which should be filtered out. In , the tanh activation function processes the output of the target meteorological element to construct an auxiliary matrix, which will provide additional information for a better understanding of the changing trend of the wind field. In , the Softmax function is used to calculate the relative importance of the target meteorological element output to the wind field forecast. Tj is a weight matrix, δ Tj Determine the impact of the target meteorological element output in the Tj time period on the wind field forecast. The correlation result δ Tj and and Multiply to obtain weighted information and fuse it with the previous moment result CINF Tj-1 The fusion fully considers the dynamic relationship between the historical data of the wind field and the historical data of the target meteorological elements, thereby improving the accuracy of the forecast. It should be clear that when j = 1, CINF T0 is a random matrix or a zero matrix that conforms to the normal distribution.

[0062] According to the wind farm output in the T4 period Output of target meteorological elements in time period T4 The conditional feature matrix of wind farm historical data output through the time series network in the T4 period The conditional feature matrix of the historical data of the target meteorological element output by the time series network in the T4 period And the third fusion result CINF of the previous time period, i.e., time period T3 T3 , the wind field forecast result for the fifth time period is determined by the following formula:

[0063] Among them, ReLU is an activation function, α, β, γ are parameters, 0≤α≤1, 0≤β≤1, 0≤γ≤1, in this embodiment, α, β, γ are set to 0.8, 0.2, 0.5 respectively, but the sizes of α, β, γ can also be adjusted according to actual experimental results, and the setting usually satisfies α>β.

[0064] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0065] In this embodiment, a wind farm forecasting device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made are not repeated here. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0066] Figure 5 : is a structural block diagram of a wind field forecasting device according to an embodiment of the present application, the device comprising:

[0067] The acquisition module 52 is used to input the historical data of the wind farm in the historical time period and the historical data of the target meteorological element in the historical time period into the time series network to obtain the wind farm output of each time period in the historical time period and the target meteorological element output of each time period, wherein the target meteorological element includes at least one of the following: temperature, humidity, air pressure, precipitation;

[0068] A fusion module 54, configured to fuse the wind field output of a first time period within the historical time period and the target meteorological element output of the first time period to determine a first fusion result of the first time period;

[0069] The forecast module 56 is used to perform wind field forecast for a third time period based on the wind field output of a second time period within the historical time period and the target meteorological element output of the second time period, as well as the first fusion result, wherein each time period includes: the first time period, the second time period, and the third time period; the first time period is earlier than the second time period, and the second time period is earlier than the third time period, and the first time period, the second time period and the third time period are adjacent in sequence.

[0070] Through the above device, the historical data of the wind field in the historical time period and the historical data of the target meteorological elements in the historical time period are input into the time series network to obtain the wind field output of each time period in the historical time period and the target meteorological element output of each time period, wherein the target meteorological element includes at least one of the following: temperature, humidity, air pressure, precipitation; the wind field output of the first time period in the historical time period and the target meteorological element output of the first time period are fused to determine the first fusion result of the first time period; the wind field forecast of the third time period is performed according to the wind field output of the second time period in the historical time period and the target meteorological element output of the second time period, and the first fusion result, wherein each time period includes: the first time period, the second time period, and the third time period; the first time period is earlier than the second time period, and the second time period is earlier than the third time period, and the first time period, the second time period and the third time period are adjacent in sequence. Thus, the problem that the wind field forecast cannot be accurately performed due to the relatively limited useful information extracted from the historical data of the wind field in the related art is solved.

[0071] In an exemplary embodiment, the acquisition module 52 is also used to obtain the one-dimensional feature vector of the wind farm historical data in the first time period, and the one-dimensional feature vector of the historical data of the target meteorological element in the first time period; based on the one-dimensional feature vector of the wind farm historical data in the first time period, and the one-dimensional feature vector of the historical data of the target meteorological element in the first time period, determine the wind farm output in the first time period, the target meteorological element output in the first time period, the conditional feature matrix of the wind farm historical data output through the timing network in the first time period, and the conditional feature matrix of the historical data of the target meteorological element output through the timing network in the first time period.

[0072] In an exemplary embodiment, the acquisition module 52 is also used to obtain the one-dimensional feature vector of the wind farm historical data in each time period, the one-dimensional feature vector of the historical data of the target meteorological element in each time period, the wind farm output in the first time period, the target meteorological element output in the first time period, the conditional feature matrix of the wind farm historical data output through the timing network in the first time period, and the conditional feature matrix of the historical data of the target meteorological element output through the timing network in the first time period; when j is greater than 1, the wind farm output in each time period is determined according to the following formula And determine the target meteorological element output for each time period according to the following formula: Where Tj is the jth time period, Tj-1 is the j-1th time period, j is a positive integer, and WD Tj is the one-dimensional feature vector of the wind farm historical data in the Tj time period, WV Tj is the one-dimensional feature vector of the historical data of the target meteorological element in the Tj time period, is the wind farm output in the Tj time period, is the target meteorological element output in the Tj time period, is the wind farm output in the Tj-1 time period, is the target meteorological element output in the Tj-1 time period, is the conditional characteristic matrix of the wind farm historical data output through the timing network in the Tj time period, is the conditional feature matrix of the historical data of the target meteorological element output by the time series network in the Tj time period, is the conditional characteristic matrix of the wind farm historical data output by the timing network in the Tj-1 time period, It is the conditional feature matrix output by the time series network of the historical data of the target meteorological element in the Tj-1 time period.

[0073] In an exemplary embodiment, the fusion module 54 is further used to obtain the wind field output in the first time period and the target meteorological element output in the first time period; the first fusion result is determined according to the following formula: and Among them, σ, tanh and Softmax represent three different activation functions, CINF T0 is a normally distributed random matrix or a zero matrix, is the wind farm output in the first time period, Output of target meteorological elements for the first time period.

[0074] In an exemplary embodiment, the forecast module 56 is further used to obtain the conditional feature matrix of the wind farm historical data output by the time series network in the second time period and the conditional feature matrix of the target meteorological element historical data output by the time series network in the second time period; when j=2, the result of the wind farm forecast is determined according to the following formula Where W and b represent training parameters, ReLU represents activation function, α, β, γ are parameters, 0≤α≤1, 0≤β≤1, 0≤γ≤1, is the wind farm output in the Tj time period, is the target meteorological element output in the Tj time period, CINF Tj-1 is the j-1th fusion result in time period Tj-1, is the conditional characteristic matrix of the wind farm historical data output through the timing network in the Tj time period, It is the conditional feature matrix output by the time series network of the historical data of the target meteorological element in the Tj time period, where j is a positive integer.

[0076] In an exemplary embodiment, the acquisition module 52 is also used to input the wind farm historical data and the historical data of the target meteorological elements into the embedding layer of the target neural network, output a first feature matrix of the wind farm historical data of a unified dimension and a second feature matrix of the historical data of the target meteorological elements of a unified dimension; and convert the first feature matrix and the second feature matrix into a first one-dimensional feature vector and a second one-dimensional feature vector through the fully connected layer of the target neural network.

[0077] In an exemplary embodiment, the fusion module 54 is further used to obtain the first fusion result, the wind field output of the second time period, and the target meteorological element output of the second time period; when j=2, the second fusion result of the second time period is determined according to the following formula:

[0078] and Among them, σ, tanh and Softmax represent three different activation functions, CINF Tj is the fusion result of the jth time period, CINF Tj-1 is the fusion result of the j-1th time period, is the wind farm output in the Tj time period, It is the target meteorological element output in the Tj time period.

[0079] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when running.

[0080] Optionally, in this embodiment, the storage medium may be configured to store a computer program for performing the following steps:

[0081] S1, inputting historical data of the wind farm in a historical time period and historical data of the target meteorological element in the historical time period into a time series network to obtain the wind farm output of each time period in the historical time period and the target meteorological element output of each time period, wherein the target meteorological element includes at least one of the following: temperature, humidity, air pressure, and precipitation;

[0082] S2, fusing the wind field output of a first time period within the historical time period and the target meteorological element output of the first time period to determine a first fusion result of the first time period;

[0083] S3, performing wind field forecast for a third time period based on the wind field output of a second time period within the historical time period, the target meteorological element output of the second time period, and the first fusion result, wherein each time period includes: the first time period, the second time period, and the third time period; the first time period is earlier than the second time period, and the second time period is earlier than the third time period, and the first time period, the second time period and the third time period are adjacent in sequence.

[0084] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0085] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail herein.

[0086] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0087] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0088] S1, inputting historical data of the wind farm in a historical time period and historical data of the target meteorological element in the historical time period into a time series network to obtain the wind farm output of each time period in the historical time period and the target meteorological element output of each time period, wherein the target meteorological element includes at least one of the following: temperature, humidity, air pressure, and precipitation;

[0089] S2, fusing the wind field output of a first time period within the historical time period and the target meteorological element output of the first time period to determine a first fusion result of the first time period;

[0090] S3, performing wind field forecast for a third time period based on the wind field output of a second time period within the historical time period, the target meteorological element output of the second time period, and the first fusion result, wherein each time period includes: the first time period, the second time period, and the third time period; the first time period is earlier than the second time period, and the second time period is earlier than the third time period, and the first time period, the second time period and the third time period are adjacent in sequence.

[0091] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0092] An embodiment of the present application further provides a computer program product, comprising a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program product, and when the computer program is executed by a processor, the steps of the method described in each embodiment of the present application are implemented.

[0093] Optionally, in this embodiment, the above computer program may be configured to implement the following steps when executed by a processor:

[0094] S1, inputting historical data of the wind farm in a historical time period and historical data of the target meteorological element in the historical time period into a time series network to obtain the wind farm output of each time period in the historical time period and the target meteorological element output of each time period, wherein the target meteorological element includes at least one of the following: temperature, humidity, air pressure, and precipitation;

[0095] S2, fusing the wind field output of a first time period within the historical time period and the target meteorological element output of the first time period to determine a first fusion result of the first time period;

[0096] S3, performing wind field forecast for a third time period based on the wind field output of a second time period within the historical time period, the target meteorological element output of the second time period, and the first fusion result, wherein each time period includes: the first time period, the second time period, and the third time period; the first time period is earlier than the second time period, and the second time period is earlier than the third time period, and the first time period, the second time period and the third time period are adjacent in sequence.

[0097] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail herein.

[0098] Obviously, those skilled in the art should understand that the above modules or steps of the present application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0099] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A wind field forecasting method, characterized in that: include: Inputting historical data of the wind farm in a historical time period and historical data of the target meteorological element in the historical time period into a time series network to obtain the wind farm output of each time period in the historical time period and the target meteorological element output of each time period, wherein the target meteorological element includes at least one of the following: temperature, humidity, air pressure, and precipitation; Fusing the wind field output of a first time period within the historical time period with the target meteorological element output of the first time period to determine a first fusion result of the first time period; The wind field forecast for the third time period is performed based on the wind field output of the second time period within the historical time period, the target meteorological element output of the second time period, and the first fusion result, wherein each time period includes: the first time period, the second time period, and the third time period; the first time period is earlier than the second time period, and the second time period is earlier than the third time period, and the first time period, the second time period and the third time period are adjacent in sequence.

2. The method according to claim 1, characterized in that Before obtaining the wind field output of each time period within the historical time period and the target meteorological element output of each time period, the method further includes: Acquire a one-dimensional feature vector of the historical data of the wind farm in the first time period, and a one-dimensional feature vector of the historical data of the target meteorological element in the first time period; The wind farm output in the first time period, the target meteorological element output in the first time period, the conditional characteristic matrix of the wind farm historical data output through the timing network in the first time period, and the conditional characteristic matrix of the target meteorological element historical data output through the timing network in the first time period are determined based on the one-dimensional characteristic vector of the wind farm historical data in the first time period and the one-dimensional characteristic vector of the target meteorological element historical data in the first time period.

3. The method according to claim 2, characterized in that Obtaining the wind field output for each time period within the historical time period and the target meteorological element output for each time period, including: Obtain a one-dimensional feature vector of the wind farm historical data in each time period, a one-dimensional feature vector of the historical data of the target meteorological element in each time period, the wind farm output in the first time period, the target meteorological element output in the first time period, a conditional feature matrix of the wind farm historical data output through the timing network in the first time period, and a conditional feature matrix of the historical data of the target meteorological element output through the timing network in the first time period; When j is greater than 1, the wind farm output for each time period is determined according to the following formula: as well as The target meteorological element output for each time period is determined according to the following formula Where Tj is the jth time period, Tj-1 is the j-1th time period, j is a positive integer, and WD Tj is the one-dimensional feature vector of the wind farm historical data in the Tj time period, WV Tj is the one-dimensional feature vector of the historical data of the target meteorological element in the Tj time period, is the wind farm output in the Tj time period, is the target meteorological element output in the Tj time period, is the wind farm output in the Tj-1 period, is the target meteorological element output in the Tj-1 time period, is the conditional characteristic matrix of the wind farm historical data output through the timing network in the Tj time period, is the conditional feature matrix of the historical data of the target meteorological element output by the time series network in the Tj time period, is the conditional characteristic matrix of the wind farm historical data output by the timing network in the Tj-1 time period, It is the conditional feature matrix output by the time series network of the historical data of the target meteorological element in the Tj-1 time period.

4. The method according to claim 1, characterized in that The wind field output of a first time period within the historical time period and the target meteorological element output of the first time period are fused to determine a first fusion result of the first time period, including: Obtaining the wind field output in the first time period and the target meteorological element output in the first time period; The first fusion result CINF is determined according to the following formula T1 : and Among them, σ, tanh and Softmax represent three different activation functions, CINF T0 is a normally distributed random matrix or a zero matrix, is the wind farm output in the first time period, Output of target meteorological elements for the first time period.

5. The method according to claim 1, characterized in that The method further comprises: performing a wind field forecast for a third time period according to the wind field output for a second time period within the historical time period, the target meteorological element output for the second time period, and the first fusion result, including: Acquire a conditional characteristic matrix of the wind farm historical data output by the time series network in the second time period and a conditional characteristic matrix of the target meteorological element historical data output by the time series network in the second time period; When j = 2, the wind field forecast result y is determined according to the following formula: Where W and b represent training parameters, ReLU represents activation function, α, β, γ are parameters, 0≤α≤1, 0≤β≤1, 0≤γ≤1, is the wind farm output in the Tj time period, is the target meteorological element output in the Tj time period, CINF Tj-1 is the j-1th fusion result in time period Tj-1, is the conditional characteristic matrix of the wind farm historical data output through the timing network in the Tj time period, It is the conditional feature matrix output by the time series network of the historical data of the target meteorological element in the Tj time period, where j is a positive integer.

6. The method according to claim 1, characterized in that Before inputting the historical data of the wind farm within the historical time period and the historical data of the target meteorological element within the historical time period into the time series network, the method further comprises: Inputting the wind farm historical data and the target meteorological element historical data into an embedding layer of a target neural network, and outputting a first feature matrix of the wind farm historical data of a unified dimension and a second feature matrix of the target meteorological element historical data of a unified dimension; The first feature matrix and the second feature matrix are converted into a first one-dimensional feature vector and a second one-dimensional feature vector through a fully connected layer of the target neural network.

7. The method according to claim 1, characterized in that After fusing the wind field output of a first time period within the historical time period with the target meteorological element output of the first time period to determine a first fusion result of the first time period, the method further includes: Obtaining the first fusion result, the wind field output of the second time period, and the target meteorological element output of the second time period; In the case of j=2, the second fusion result CINF of the second time period is determined according to the following formula: T2 : and Among them, σ, tanh and Softmax represent three different activation functions, CINF Tj is the fusion result of the jth time period, CINF Tj-1 is the fusion result of the j-1th time period, is the wind farm output in the Tj time period, It is the target meteorological element output in the Tj time period.

8. A wind field forecasting device, characterized in that: include: An acquisition module is used to input the historical data of the wind farm in the historical time period and the historical data of the target meteorological element in the historical time period into the time series network to obtain the wind farm output of each time period in the historical time period and the target meteorological element output of each time period, wherein the target meteorological element includes at least one of the following: temperature, humidity, air pressure, precipitation; A fusion module, used for fusing the wind field output of a first time period within the historical time period and the target meteorological element output of the first time period to determine a first fusion result of the first time period; A forecast module is used to perform wind field forecast for a third time period based on the wind field output of a second time period within the historical time period and the target meteorological element output of the second time period, as well as the first fusion result, wherein each time period includes: the first time period, the second time period, and the third time period; the first time period is earlier than the second time period, and the second time period is earlier than the third time period, and the first time period, the second time period and the third time period are adjacent in sequence.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program executes the method according to any one of claims 1 to 7 when executed.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images