Training method and device of wind speed prediction model, equipment and medium
By obtaining the historical wind speed data and correlation of multiple measurement points, and training the wind speed prediction model, the problem of low accuracy in wind speed prediction in the existing technology is solved, and the operating efficiency and reliability of the wind farm are improved.
Patent Information
- Application Number
- CN202411454824.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-17
- Publication Date
- 2025-03-21
AI Technical Summary
In the prior art, the historical wind speed data of a single data source or measurement point are relied on to predict wind speed, which has low accuracy, affecting the operating efficiency and reliability of the wind farm.
By obtaining the historical measured wind speed data and historical predicted wind speed data of multiple measurement points of the wind farm, the target correlation between different measurement points is determined, and the preset network is iteratively trained based on these data until the preset cutoff conditions are met, and the wind speed prediction model is obtained.
The prediction accuracy of the wind speed prediction model of the wind farm is improved, and the operating efficiency and reliability of the wind farm are improved.
Smart Images
Figure CN119669749A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the technical field of wind speed prediction, and in particular to a training method, device, equipment and medium for a wind speed prediction model. Background Art
[0002] In the wind power industry, high temporal resolution wind speed forecasting is of great significance for improving the operational efficiency and safety of wind farms. On the one hand, real-time and accurate forecasting of wind speed data can optimize the power output of wind turbines and reduce equipment wear, thereby improving the economic benefits and long-term sustainability of wind farms. On the other hand, high temporal resolution forecasting of wind speed data is crucial for wind farm wake control and is a key input for wind farm wake optimization and operational control.
[0003] In the related art, the historical wind speed data of a single data source or measuring point is mainly relied on to determine the predicted wind speed data. Obviously, the accuracy of this wind speed data prediction method is low, which in turn affects the operating efficiency and reliability of the wind farm. Summary of the invention
[0004] In order to solve the above technical problems, the present disclosure provides a training method, device, equipment and medium for a wind speed prediction model.
[0005] In a first aspect, the present disclosure provides a method for training a wind speed prediction model, comprising:
[0006] Obtain historical measured wind speed data and historical predicted wind speed data of the wind farm at historical moments, wherein the historical measured wind speed data includes at least one of historical measured wind speeds of wind turbines at multiple measuring points, historical measured wind speeds by laser radars, and historical measured wind speeds by wind towers;
[0007] determining a target correlation between a first measuring point and a second measuring point among a plurality of measuring points;
[0008] Based on the historical measured wind speed data, the historical predicted wind speed data and the target correlation, the preset network is iteratively trained until the preset network under the current training times meets the preset cutoff condition, thereby obtaining a wind speed prediction model for the wind farm.
[0009] In a second aspect, the present disclosure provides a training device for a wind speed prediction model, comprising:
[0010] An acquisition module is used to acquire historical measured wind speed data and historical predicted wind speed data of a wind farm at historical moments, wherein the historical measured wind speed data includes at least one of historical measured wind speeds of wind turbines at multiple measuring points, historical measured wind speeds of laser radars, and historical measured wind speeds of wind towers;
[0011] A determination module, used to determine a target correlation between a first measuring point and a second measuring point among a plurality of measuring points;
[0012] The training model is used to iteratively train the preset network based on the historical measured wind speed data, the historical predicted wind speed data and the target correlation until the preset network under the current training times meets the preset cutoff condition, thereby obtaining a wind speed prediction model for the wind farm.
[0013] In a third aspect, an embodiment of the present disclosure further provides an electronic device, the device comprising:
[0014] one or more processors;
[0015] a storage device for storing one or more programs,
[0016] When one or more programs are executed by one or more processors, the one or more processors implement the method provided in the first aspect.
[0017] In a fourth aspect, an embodiment of the present disclosure further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method provided in the first aspect is implemented.
[0018] Compared with the prior art, the technical solution provided by the embodiments of the present disclosure has the following advantages:
[0019] A training method, device, equipment and medium for a wind speed prediction model of an embodiment of the present disclosure includes: obtaining historical measured wind speed data and historical predicted wind speed data of a wind farm at historical moments, wherein the historical measured wind speed data includes at least one of the historical measured wind speed of wind turbines at multiple measuring points, the historical measured wind speed of a laser radar and the historical measured wind speed of a wind tower; determining the target correlation between a first measuring point and a second measuring point among multiple measuring points; iteratively training a preset network based on the historical measured wind speed data, the historical predicted wind speed data and the target correlation until the preset network under the current number of training times meets the preset cutoff condition, thereby obtaining a wind speed prediction model for the wind farm. Since the model training process takes into account the multiple measured wind speeds of multiple measuring points in the wind farm, and also takes into account the correlation between different measuring points, the prediction accuracy of the wind speed prediction model of the wind farm is improved, and the operation efficiency and reliability of the wind farm are further improved. 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 disclosure and, together with the description, serve to explain the principles of the present disclosure.
[0021] In order to more clearly illustrate the embodiments of the present invention 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 A schematic diagram of a flow chart of a training method for a wind speed prediction model provided in an embodiment of the present disclosure;
[0023] Figure 2 A schematic diagram of the structure of a wind speed prediction model training device provided in an embodiment of the present disclosure;
[0024] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION
[0025] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.
[0026] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.
[0027] Most wind speed prediction methods in related technologies rely on data at a time scale of 15 minutes or even longer, which cannot meet the prediction requirements at the second / minute level, resulting in wind farms being unable to effectively respond to rapidly changing wind speed conditions.
[0028] In order to solve the above problems, the following Figure 1 The training method of the wind speed prediction model provided in the embodiment of the present disclosure is described. In the embodiment of the present disclosure, the training method of the wind speed prediction model can be executed by an electronic device or a server. Among them, the electronic device may include a device with a communication function such as a tablet computer, a desktop computer, a laptop computer, and may also include a device simulated by a virtual machine or a simulator. The server may include a single server or a server cluster. This embodiment is specifically explained with an electronic device as the execution subject.
[0029] Figure 1 A flow chart of a training method for a wind speed prediction model provided by an embodiment of the present disclosure is shown.
[0030] like Figure 1 As shown, the training method of the wind speed prediction model may include the following steps.
[0031] S110. Obtain historical measured wind speed data and historical predicted wind speed data of the wind farm at historical moments, wherein the historical measured wind speed data includes at least one of historical measured wind speeds of wind turbines at multiple measuring points, historical measured wind speeds by lidar, and historical measured wind speeds by wind towers.
[0032] In this embodiment, when performing model training, the electronic device obtains real-time wind speed data of multiple measuring points in the wind farm, and the real-time wind speed data of the multiple measuring points include historical measured wind speed data and historical predicted wind speed data.
[0033] The historical measured wind speed data is the measured wind speed data before the historical moment, and can be obtained from at least one of wind turbines, laser radars and wind towers at multiple measuring points.
[0034] The historical predicted wind speed data are the measured wind speed data after the historical moment and before the current moment.
[0035] Among them, the measuring point refers to the wind speed measurement point, which is uniquely determined according to the longitude and latitude.
[0036] In order to improve the accuracy of model training, the historical measured wind speed data and the historical predicted wind speed data are processed by data situation, normalization and time synchronization to improve the quality and consistency of the historical measured wind speed data and the historical predicted wind speed data.
[0037] S120: Determine a target correlation between a first measuring point and a second measuring point among a plurality of measuring points.
[0038] In this embodiment, before model training, it is also necessary to analyze the correlation between different measuring points so that the model in the training process considers the information of other related measuring points when processing the data of one measuring point, thereby improving the wind speed prediction accuracy of the wind speed prediction model.
[0039] The first measuring point and the second measuring point are points corresponding to any two positions of the multiple measuring points.
[0040] Among them, the specific implementation method of S120 includes but is not limited to the following methods: determining the time correlation coefficient between the first measuring point and the second measuring point based on the historical measured wind speed data of the first measuring point and the historical measured wind speed data of the second measuring point; determining the spatial correlation coefficient between the first measuring point and the second measuring point based on the latitude and longitude data of the first measuring point and the latitude and longitude data of the second measuring point; performing weighted sum calculation on the time correlation coefficient and the spatial correlation coefficient according to the first weight corresponding to the time correlation coefficient and the second weight corresponding to the spatial correlation coefficient to obtain the target correlation degree.
[0041] Among them, the time correlation coefficient is the time correlation between different measuring points. Specifically, the method for determining the time correlation coefficient includes but is not limited to the following method: according to the preset time window, the historical measured wind speed data of the first measuring point is divided into multiple first historical measured wind speed windows, and the historical measured wind speed data of the second measuring point is divided into multiple second historical measured wind speed windows, wherein the time length of the preset time window is less than or equal to the preset time length threshold; the correlation calculation is performed on any first historical measured wind speed window of the first measuring point and any second historical measured wind speed window of the second measuring point to obtain the time correlation coefficient between the first measuring point and the second measuring point.
[0042] Specifically, the electronic device first divides the historical measured wind speed data of the first measuring point and the historical measured wind speed data of the second measuring point into data windows with time characteristics according to a preset time window, and obtains multiple first historical measured wind speed windows and multiple second historical measured wind speed windows. Then, the electronic device uses a time-lagged cross-correlation algorithm, a time similarity measurement algorithm, etc. to perform correlation calculation on any first historical measured wind speed window of the first measuring point and any second historical measured wind speed window of the second measuring point to obtain a correlation coefficient for each time window, and then adds the correlation coefficients of all time windows and averages them to obtain a time correlation coefficient.
[0043] The preset time length threshold may be a pre-set short time window. For example, the preset time length threshold is 30 seconds, 5 minutes, or 10 minutes. In this way, by dividing the historical measured wind speed data of the first measuring point and the historical measured wind speed data of the second measuring point into a plurality of data windows with time characteristics, the time resolution of the wind speed data is improved, which is conducive to capturing the short-term fluctuation characteristics of the wind speed data.
[0044] The spatial correlation coefficient is the spatial correlation between different measuring points. Specifically, the electronic device processes the longitude and latitude data of the first measuring point and the longitude and latitude data of the second measuring point by using a Euclidean distance calculation method or the like to obtain the spatial correlation coefficient between the first measuring point and the second measuring point.
[0045] The target correlation is a correlation coefficient that combines the time correlation coefficient and the space correlation coefficient between the first measuring point and the second measuring point. Specifically, after obtaining the time correlation coefficient and the space correlation coefficient, the electronic device normalizes the time correlation coefficient and the space correlation coefficient so that the time correlation coefficient and the space correlation coefficient are uniformly converted to a range of 0-1 to calculate the target correlation.
[0046] Optionally, the method for determining the target relevance can be expressed as follows:
[0047] E=ω1×C n +ω2×(1-D n )
[0048] Among them, E is the target relevance, ω1 is the first weight, C n is the time correlation coefficient determined based on n measurement points, ω2 is the second weight, D n It is the spatial correlation coefficient determined based on n measuring points.
[0049] Therefore, before model training, the correlation between different measurement points is also mined, which is conducive to improving the generalization ability of the model training process, so that the trained model can make more accurate predictions in real-time and changing environments.
[0050] S130, based on the historical measured wind speed data, the historical predicted wind speed data and the target correlation, iteratively train the preset network until the preset network under the current training times meets the preset cutoff condition, so as to obtain a wind speed prediction model for the wind farm.
[0051] In this embodiment, the electronic device needs to build a preset network before model training. The built prediction network needs to have the ability to capture the shared features and common features of each measurement point, and also needs to have the ability to learn the correlation between different measurement points.
[0052] Among them, the specific implementation method of S130 includes but is not limited to the following methods: based on the input layer in the preset network, the historical measured wind speed data is processed to obtain shared features between multiple measuring points; based on the attention layer in the preset network, the shared features are processed according to the target correlation to obtain attention features; based on the output layer in the preset network, the attention features are processed to obtain historical estimated wind speed data corresponding to multiple measuring points; based on the historical estimated wind speed data, the historical predicted wind speed data and the target correlation, the preset network is iteratively trained until the preset network under the current training times meets the preset cutoff conditions, thereby obtaining a wind speed prediction model for the wind farm.
[0053] The input layer can be understood as a shared layer, which is used to extract and share common features of multiple measurement points. The common features are shared features, so as to learn universal patterns and structures. Optionally, the input layer includes but is not limited to any one of a convolutional neural network (CNN), a recurrent neural network (RNN) and a graph convolutional network (GCN).
[0054] The output layer can be understood as a dedicated layer, allowing a specific output strategy to be customized for each task to predict specific features of each measuring point, and the specific features are used as historical estimated wind speed data. Optionally, the output layer includes but is not limited to a deep neural network (DNN).
[0055] Among them, the attention layer can be understood as an attention mechanism that guides the extraction of correlation, integrating the calculated correlations between multiple measuring points into a matrix, so that the model can also consider the wind speed data of other related measuring points when processing one measuring point.
[0056] Among them, based on the historical estimated wind speed data, the historical predicted wind speed data and the target correlation, the preset network is iteratively trained until the preset network under the current training times meets the preset cutoff condition, and the wind speed prediction model of the wind farm is obtained, including: using a preset loss function to process the historical estimated wind speed data, the historical predicted wind speed data and the target correlation to obtain the loss data of the preset network; according to the loss data, iteratively adjusting the model parameters of the preset network until the preset network under the current training times meets the preset cutoff condition, and the wind speed prediction model of the wind farm is obtained.
[0057] Among them, the preset loss function includes a first loss function and a second loss function; using the preset loss function, the historical estimated wind speed data, the historical predicted wind speed data and the target correlation are processed to obtain the loss data of the preset network, including: inputting the historical estimated wind speed data and the historical predicted wind speed data into the first loss function to obtain the first loss data; inputting the historical estimated wind speed data and the target correlation corresponding to different measuring points at the same time into the second loss function to obtain the second loss data; according to the first loss data, the second loss data and the preset hyperparameters, the loss data of the preset network is calculated.
[0058] Optionally, the first loss function can be expressed as follows:
[0059]
[0060] Among them, L i is the first loss data determined based on the historical estimated wind speed data and historical predicted wind speed data of the i-th measuring point, y i,m is the historical forecast wind speed data at the i-th measuring point at the m-th moment, is the historical estimated wind speed data at the i-th measuring point at the m-th moment, M i It is the time length of the historical estimated wind speed data and the historical predicted wind speed data of the i-th measuring point, and the time length includes multiple preset time windows.
[0061] Optionally, the second loss function can be expressed as follows:
[0062]
[0063] Among them, R is the second loss data, N is the number of measurement points, a ij is the target correlation between the i-th measuring point and the j-th measuring point, The historical estimated wind speed data at the jth measuring point at the mth moment.
[0064] Optionally, the preset loss function can be expressed as follows:
[0065]
[0066] Wherein, λ is a preset hyperparameter used to control the weight of the relevance regularization term to balance the learning of prediction accuracy and relevance. Specifically, the electronic device can apply a hyperparameter optimization method, such as a grid search method and a random search method, to determine the model hyperparameters that can achieve the best performance.
[0067] The preset cutoff condition may be that the loss data of the preset network is less than a preset loss threshold. The preset loss threshold may be a predetermined minimum value.
[0068] Through the above method, the wind speed data of multiple measuring points in the wind farm are used for model training, which improves the real-time and accuracy of wind speed prediction. At the same time, the wind speed data of each measuring point in the wind farm are divided into multiple data windows with time characteristics, which improves the time resolution of the wind speed data and is conducive to capturing the short-term fluctuation characteristics of the wind speed data. In addition, the method of joint analysis of multi-measurement point data is used to deeply explore the correlation between the measuring points in the wind farm, and the attention mechanism and loss function are customized according to this correlation, thereby effectively utilizing the mutual information between the wind speed data at different locations in the wind farm and improving the accuracy of wind speed prediction.
[0069] A training method for a wind speed prediction model of an embodiment of the present disclosure includes: obtaining historical measured wind speed data and historical predicted wind speed data of a wind farm at historical moments, wherein the historical measured wind speed data includes at least one of the historical measured wind speed of wind turbines at multiple measuring points, the historical measured wind speed of a laser radar, and the historical measured wind speed of a wind tower; determining the target correlation between a first measuring point and a second measuring point among multiple measuring points; iteratively training a preset network based on the historical measured wind speed data, the historical predicted wind speed data, and the target correlation until the preset network under the current number of training times meets the preset cutoff condition, thereby obtaining a wind speed prediction model for the wind farm. Since the model training process takes into account the multiple measured wind speeds of multiple measuring points in the wind farm, and also takes into account the correlation between different measuring points, the prediction accuracy of the wind speed prediction model of the wind farm is improved, and the operation efficiency and reliability of the wind farm are further improved.
[0070] In some embodiments of the present disclosure, after the electronic device has trained the wind speed prediction model of the wind farm, the method also includes the following steps: obtaining current measured wind speed data of the wind farm at the current moment, wherein the current measured wind speed data includes at least one of the historical measured wind speeds of wind turbines at multiple measuring points, the historical measured wind speeds of lidars, and the historical measured wind speeds of wind towers; using the wind speed prediction model of the wind farm, processing the current measured wind speed data to obtain predicted wind speed data of the wind farm at future moments.
[0071] Specifically, after the wind speed prediction model of the wind farm is trained, the model can be put into use. First, the current measured wind speed data collected at the wind farm at the current moment is input into the wind speed prediction model of the wind farm. Then, the wind speed prediction model of the wind farm is sequentially passed through the input layer, attention layer and output layer to predict the predicted wind speed data of the wind farm at the future moment.
[0072] It can be understood that since the current measured wind speed data at the current moment includes wind speed data from multiple measuring points, and the model training process is based on wind speed data with higher time resolution and the training process takes into account the correlation between different measuring points, the wind speed prediction model of the wind farm can accurately achieve wind speed prediction.
[0073] The disclosed embodiment also provides a training device for a wind speed prediction model for implementing the above-mentioned training method for a wind speed prediction model, and the device is configured in an electronic device or a server. The electronic device may include a tablet computer, a desktop computer, a laptop computer, and other devices with communication functions, and may also include a virtual machine or a device simulated by a simulator. The server may include a single server or a server cluster. This embodiment is specifically explained using an electronic device as a training device for a wind speed prediction model. Figure 2 Provide explanation.
[0074] Figure 2 A schematic structural diagram of a wind speed prediction model training device provided by an embodiment of the present disclosure is shown.
[0075] like Figure 2 As shown, the training device 200 of the wind speed prediction model may include:
[0076] An acquisition module 210 is used to acquire historical measured wind speed data and historical predicted wind speed data of a wind farm at historical moments, wherein the historical measured wind speed data includes at least one of historical measured wind speeds of wind turbines at multiple measuring points, historical measured wind speeds of laser radars, and historical measured wind speeds of wind towers;
[0077] A determination module 220, configured to determine a target correlation between a first measuring point and a second measuring point among a plurality of measuring points;
[0078] The training model 230 is used to iteratively train the preset network based on the historical measured wind speed data, the historical predicted wind speed data and the target correlation until the preset network under the current training times meets the preset cutoff condition, thereby obtaining a wind speed prediction model for the wind farm.
[0079] A training device for a wind speed prediction model of an embodiment of the present disclosure includes: obtaining historical measured wind speed data and historical predicted wind speed data of a wind farm at historical moments, wherein the historical measured wind speed data includes at least one of the historical measured wind speed of wind turbines at multiple measuring points, the historical measured wind speed of a laser radar, and the historical measured wind speed of a wind tower; determining the target correlation between a first measuring point and a second measuring point among multiple measuring points; iteratively training a preset network based on the historical measured wind speed data, the historical predicted wind speed data, and the target correlation until the preset network under the current number of training times meets the preset cutoff condition, thereby obtaining a wind speed prediction model for the wind farm. Since the model training process takes into account the multiple measured wind speeds of multiple measuring points in the wind farm, and also takes into account the correlation between different measuring points, the prediction accuracy of the wind speed prediction model of the wind farm is improved, and the operation efficiency and reliability of the wind farm are further improved.
[0080] In some embodiments of the present disclosure, the determination module 220 includes:
[0081] A first determining unit, configured to determine a time correlation coefficient between the first measuring point and the second measuring point based on historical measured wind speed data of the first measuring point and historical measured wind speed data of the second measuring point;
[0082] A second determining unit, configured to determine a spatial correlation coefficient between the first measuring point and the second measuring point based on the latitude and longitude data of the first measuring point and the latitude and longitude data of the second measuring point;
[0083] The third determining unit is used to perform weighted sum calculation on the time correlation coefficient and the spatial correlation coefficient according to the first weight corresponding to the time correlation coefficient and the second weight corresponding to the spatial correlation coefficient to obtain the target correlation degree.
[0084] In some embodiments of the present disclosure, the first determining unit is specifically configured to:
[0085] According to a preset time window, the historical measured wind speed data of the first measuring point is divided into a plurality of first historical measured wind speed windows, and the historical measured wind speed data of the second measuring point is divided into a plurality of second historical measured wind speed windows, wherein the time length of the preset time window is less than or equal to a preset time length threshold;
[0086] A correlation calculation is performed on any first historical measured wind speed window of the first measuring point and any second historical measured wind speed window of the second measuring point to obtain a time correlation coefficient between the first measuring point and the second measuring point.
[0087] In some embodiments of the present disclosure, the training model 230 includes:
[0088] A first processing unit, configured to process the historical measured wind speed data based on an input layer in the preset network to obtain shared features among a plurality of measuring points;
[0089] A second processing unit is used to process the shared features according to the target relevance based on the attention layer in the preset network to obtain attention features;
[0090] A third processing unit is used to process the attention feature based on the output layer in the preset network to obtain historical estimated wind speed data corresponding to multiple measuring points respectively;
[0091] The iterative training unit is used to iteratively train the preset network based on the historical estimated wind speed data, the historical predicted wind speed data and the target correlation until the preset network under the current training times meets the preset cutoff condition, thereby obtaining a wind speed prediction model for the wind farm.
[0092] In some embodiments of the present disclosure, the iterative training unit is specifically used to:
[0093] Using a preset loss function, the historical estimated wind speed data, the historical predicted wind speed data and the target correlation are processed to obtain loss data of the preset network;
[0094] According to the loss data, the model parameters of the preset network are iteratively adjusted until the preset network under the current training times meets the preset cutoff condition, so as to obtain the wind speed prediction model of the wind farm.
[0095] In some embodiments of the present disclosure, the preset loss function includes a first loss function and a second loss function;
[0096] The iterative training unit is specifically used to: input the historical estimated wind speed data and the historical predicted wind speed data into the first loss function to obtain first loss data;
[0097] Inputting the historical estimated wind speed data corresponding to different measuring points at the same time and the target correlation into the second loss function to obtain second loss data;
[0098] The loss data of the preset network is calculated according to the first loss data, the second loss data and the preset hyperparameters.
[0099] In some embodiments of the present disclosure, the device further includes:
[0100] A current measured wind speed acquisition module is used to obtain the current measured wind speed data of the wind farm at the current moment, wherein the current measured wind speed data includes at least one of the historical measured wind speeds of wind turbines at multiple measuring points, the historical measured wind speeds of laser radars, and the historical measured wind speeds of wind towers;
[0101] The wind speed prediction module is used to process the current measured wind speed data using the wind speed prediction model of the wind farm to obtain the predicted wind speed data of the wind farm at a future time.
[0102] It should be noted that Figure 2 The wind speed prediction model training device 200 shown can be executed Figure 1 The various steps in the method embodiment shown in the figure are implemented Figure 1 The various processes and effects in the method embodiment shown are not described in detail here.
[0103] Figure 3 A schematic structural diagram of an electronic device provided by an embodiment of the present disclosure is shown.
[0104] like Figure 3 As shown, the electronic device may include a processor 301 and a memory 302 storing computer program instructions.
[0105] Specifically, the processor 301 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.
[0106] The memory 302 may include a large capacity memory for information or instructions. By way of example and not limitation, the memory 302 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a tape, or a universal serial bus (USB) drive or a combination of two or more of these. Where appropriate, the memory 302 may include a removable or non-removable (or fixed) medium. Where appropriate, the memory 302 may be inside or outside the integrated gateway device. In a particular embodiment, the memory 302 is a non-volatile solid-state memory. In a particular embodiment, the memory 302 includes a read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (Electrically Erasable Programmable ROM, EEPROM), an electrically rewritable ROM (EAROM) or a flash memory, or a combination of two or more of these.
[0107] The processor 301 reads and executes the computer program instructions stored in the memory 302 to perform the steps of the training method of the wind speed prediction model provided in the embodiment of the present disclosure.
[0108] In one example, the electronic device may further include a transceiver 303 and a bus 304. Figure 3 As shown, the processor 301, the memory 302 and the transceiver 303 are connected via a bus 304 and communicate with each other.
[0109] The bus 304 includes hardware, software, or both. For example, but not limitation, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industrial Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a Memory bus, a MicroChannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses or a combination of two or more of these. Where appropriate, the bus 304 may include one or more buses. Although embodiments of the present application describe and illustrate a particular bus, the present application contemplates any suitable bus or interconnect.
[0110] The following is an embodiment of a computer-readable storage medium provided in an embodiment of the present disclosure. The computer-readable storage medium and the training method of the wind speed prediction model in the above-mentioned embodiments belong to the same inventive concept. For details not described in detail in the embodiment of the computer-readable storage medium, reference can be made to the embodiment of the training method of the wind speed prediction model in the above-mentioned embodiments.
[0111] This embodiment provides a storage medium containing computer executable instructions. When the computer executable instructions are executed by a computer processor, they are used to perform a training method for a wind speed prediction model. The method includes:
[0112] Obtain historical measured wind speed data and historical predicted wind speed data of the wind farm at historical moments, wherein the historical measured wind speed data includes at least one of historical measured wind speeds of wind turbines at multiple measuring points, historical measured wind speeds by laser radars, and historical measured wind speeds by wind towers;
[0113] determining a target correlation between a first measuring point and a second measuring point among a plurality of measuring points;
[0114] Based on the historical measured wind speed data, the historical predicted wind speed data and the target correlation, the preset network is iteratively trained until the preset network under the current training times meets the preset cutoff condition, thereby obtaining a wind speed prediction model for the wind farm.
[0115] Of course, the storage medium containing computer executable instructions provided in the embodiment of the present disclosure is not limited to the above method operations, and the computer executable instructions can also execute related operations in the training method of the wind speed prediction model provided in any embodiment of the present disclosure.
[0116] Through the above description of the implementation methods, technicians in the relevant field can clearly understand that the present disclosure can be implemented with the help of software and necessary general hardware, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present disclosure can be essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer cloud platform (which can be a personal computer, a server, or a network cloud platform, etc.) to execute the training method of the wind speed prediction model provided in each embodiment of the present disclosure.
[0117] Note that the above are only preferred embodiments of the present disclosure and the technical principles used. Those skilled in the art will understand that the present disclosure is not limited to the specific embodiments herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present disclosure. Therefore, although the present disclosure is described in more detail through the above embodiments, the present disclosure is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present disclosure, and the scope of the present disclosure is determined by the scope of the appended claims.
Claims
1. A training method for a wind speed prediction model, characterized in that: include: Obtain historical measured wind speed data and historical predicted wind speed data of the wind farm at historical moments, wherein the historical measured wind speed data includes at least one of historical measured wind speeds of wind turbines at multiple measuring points, historical measured wind speeds by laser radars, and historical measured wind speeds by wind towers; determining a target correlation between a first measuring point and a second measuring point among a plurality of measuring points; Based on the historical measured wind speed data, the historical predicted wind speed data and the target correlation, the preset network is iteratively trained until the preset network under the current training times meets the preset cutoff condition, thereby obtaining a wind speed prediction model for the wind farm.
2. The method according to claim 1, characterized in that The determining of the target correlation between a first measuring point and a second measuring point among the plurality of measuring points comprises: Determine a time correlation coefficient between the first measuring point and the second measuring point based on historical measured wind speed data of the first measuring point and historical measured wind speed data of the second measuring point; Determining a spatial correlation coefficient between the first measuring point and the second measuring point based on the latitude and longitude data of the first measuring point and the latitude and longitude data of the second measuring point; The target correlation degree is obtained by performing a weighted sum calculation on the time correlation coefficient and the space correlation coefficient according to a first weight corresponding to the time correlation coefficient and a second weight corresponding to the space correlation coefficient.
3. The method according to claim 2, characterized in that The determining of the time correlation coefficient between the first measuring point and the second measuring point based on the historical measured wind speed data of the first measuring point and the historical measured wind speed data of the second measuring point includes: According to a preset time window, the historical measured wind speed data of the first measuring point is divided into a plurality of first historical measured wind speed windows, and the historical measured wind speed data of the second measuring point is divided into a plurality of second historical measured wind speed windows, wherein the time length of the preset time window is less than or equal to a preset time length threshold; A correlation calculation is performed on any first historical measured wind speed window of the first measuring point and any second historical measured wind speed window of the second measuring point to obtain a time correlation coefficient between the first measuring point and the second measuring point.
4. The method according to claim 1, characterized in that: The method of iteratively training a preset network based on the historical measured wind speed data, the historical predicted wind speed data and the target correlation degree until the preset network under the current number of training times meets the preset cutoff condition, thereby obtaining a wind speed prediction model for the wind farm, comprises: Based on the input layer in the preset network, the historical measured wind speed data is processed to obtain shared features among multiple measuring points; Based on the attention layer in the preset network, the shared features are processed according to the target relevance to obtain attention features; Based on the output layer in the preset network, the attention feature is processed to obtain historical estimated wind speed data corresponding to multiple measuring points respectively; Based on the historical estimated wind speed data, the historical predicted wind speed data and the target correlation, the preset network is iteratively trained until the preset network under the current training times meets the preset cutoff condition, thereby obtaining a wind speed prediction model for the wind farm.
5. The method according to claim 4, characterized in that The method of iteratively training a preset network based on the historical estimated wind speed data, the historical predicted wind speed data and the target correlation degree until the preset network under the current number of training times meets the preset cutoff condition, and obtaining a wind speed prediction model for the wind farm includes: Using a preset loss function, the historical estimated wind speed data, the historical predicted wind speed data and the target correlation are processed to obtain loss data of the preset network; According to the loss data, the model parameters of the preset network are iteratively adjusted until the preset network under the current training times meets the preset cutoff condition, so as to obtain the wind speed prediction model of the wind farm.
6. The method according to claim 5, characterized in that The preset loss function includes a first loss function and a second loss function; The method of using a preset loss function to process the historical estimated wind speed data, the historical predicted wind speed data, and the target correlation to obtain the loss data of the preset network includes: Inputting the historical estimated wind speed data and the historical predicted wind speed data into the first loss function to obtain first loss data; Inputting the historical estimated wind speed data corresponding to different measuring points at the same time and the target correlation into the second loss function to obtain second loss data; The loss data of the preset network is calculated according to the first loss data, the second loss data and the preset hyperparameters.
7. The method according to claim 1, characterized in that Also includes: Acquire current measured wind speed data of the wind farm at the current moment, wherein the current measured wind speed data includes at least one of the historical measured wind speeds of wind turbines at multiple measuring points, the historical measured wind speeds of laser radars, and the historical measured wind speeds of wind towers; The wind speed prediction model of the wind farm is used to process the currently measured wind speed data to obtain predicted wind speed data of the wind farm at a future time.
8. A training device for a wind speed prediction model, characterized in that: include: An acquisition module is used to acquire historical measured wind speed data and historical predicted wind speed data of a wind farm at historical moments, wherein the historical measured wind speed data includes at least one of historical measured wind speeds of wind turbines at multiple measuring points, historical measured wind speeds of laser radars, and historical measured wind speeds of wind towers; A determination module, used to determine a target correlation between a first measuring point and a second measuring point among a plurality of measuring points; The training model is used to iteratively train the preset network based on the historical measured wind speed data, the historical predicted wind speed data and the target correlation until the preset network under the current training times meets the preset cutoff condition, thereby obtaining a wind speed prediction model for the wind farm.
9. An electronic device, characterized in that: include: processor; A memory for storing executable instructions; The processor is used to read the executable instructions from the memory and execute the executable instructions to implement the method described in any one of claims 1 to 7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the processor implements the method according to any one of claims 1 to 7.