Wind power generation prediction method based on graph neural model and related equipment
Through the combination of graph neural model and bidirectional long and short-term memory model, the fan space characteristics are extracted, and the problem of insufficient prediction accuracy of wind power generation in the existing technology is solved, achieving higher prediction accuracy and grid stability.
Patent Information
- Application Number
- CN202111015724.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-08-31
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2041-08-31
AI Technical Summary
The existing wind power generation prediction methods are not accurate, and it is difficult to effectively consider the spatial topological characteristics and time series changes between fans, which affects the stable operation of the power grid.
Using the graph neural model method, the fan space feature extraction is performed by obtaining the fan feature matrix and the fan adjacency matrix, and combining the bidirectional long and short-term memory model and the full connection layer to predict the wind power generation, fully considering the space between the fans Topological and time series features.
It improves the accuracy and reliability of wind power generation forecasting, enhances the short-term prediction capability of wind farm power generation, and is suitable for power scheduling and system stability planning.
Smart Images

Figure CN113779875B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of wind power generation technology, and in particular to a wind power generation prediction method based on a graph neural model and related equipment. Background Art
[0002] With the development of computer technology and the application of advanced control systems, modern industrial system equipment and structures are becoming increasingly complex, and industrial processes have accumulated a large amount of historical data, which contains rich information such as the operating rules of the process, operator experience, product quality and problems in the process. Wind power generation is a renewable energy source that can effectively save standard coal and reduce the emission of gases such as sulfur dioxide, carbon dioxide and nitrogen oxides during the power generation process. Carrying out short-term forecasts of wind power generation is of great significance for power system dispatchers to formulate power generation plans, arrange spare capacity, increase the reliability of system operation, reduce the impact on system stability, and plan the development of power systems and plan maintenance arrangements. The power generation of wind power depends on the constraints of meteorological factors and environmental conditions, and there is a lot of randomness and uncertainty, which has a certain impact on the stable operation of the power grid.
[0003] Common wind power generation prediction methods include grey prediction, neural network, support vector machine and autoregressive moving average model. Wind power generation in wind farms is not only greatly affected by weather factors, but also related to factors such as wind farm equipment operation and dispatch management. The time series of wind power generation has a large fluctuation range, making modeling and prediction relatively difficult, and the accuracy of existing prediction methods is not high enough. Summary of the invention
[0004] The purpose of the embodiments of the present application is to propose a wind power generation prediction method based on a graph neural model and related equipment to improve the accuracy of power generation prediction.
[0005] In order to solve the above technical problems, the embodiment of the present application provides a method for predicting wind power generation based on a graph neural model, which adopts the following technical solution:
[0006] A wind power generation prediction method based on a graph neural model comprises the following steps:
[0007] Obtain the characteristics of the fans and the relationships between the fans in the preset space, and generate a fan characteristic matrix and a fan adjacency matrix respectively;
[0008] Inputting the fan feature matrix and the fan adjacency matrix into a pre-trained graph neural model to extract fan spatial features, and obtaining output fan spatial features;
[0009] The wind turbine spatial characteristics are input into a pre-trained power generation prediction model to obtain an output wind turbine power generation prediction value.
[0010] Furthermore, the power generation prediction model includes a bidirectional long short-term memory model and a fully connected layer, and the step of inputting the wind turbine spatial features into the pre-trained power generation prediction model to obtain the output wind turbine power generation prediction value includes:
[0011] Inputting the fan spatial features into the bidirectional long short-term memory model in sequence based on the time sequence, memorizing the fan spatial features at the historical moment based on the input gate, forget gate and output gate in the bidirectional long short-term memory model, and fusing them with the fan spatial features at the current moment to obtain the output fan memory features;
[0012] The wind turbine memory feature is input into the fully connected layer to perform probability prediction on the wind turbine memory feature to obtain a predicted value of the wind turbine power generation at the next moment.
[0013] Furthermore, the graph neural model includes a Fourier domain filter and a plurality of hidden layers, and the step of inputting the fan feature matrix and the fan adjacency matrix into the pre-trained graph neural model to extract the fan spatial features and obtaining the output fan spatial features includes:
[0014] The fan characteristic matrix and the fan adjacency matrix are passed through the Fourier domain filter, and the spatial characteristics between all fans are captured through the first-order neighborhood of the Fourier domain recorder to obtain initial fan spatial characteristics;
[0015] The initial fan space feature is passed through the multiple hidden layers to perform multi-level abstraction on the initial fan space feature to obtain the fan space feature.
[0016] Furthermore, the steps of obtaining the characteristics of the fans and the relationships between the fans in the preset space and respectively generating a fan characteristic matrix and a fan adjacency matrix include:
[0017] Acquire the characteristics of each fan, and combine the characteristics of the fans to generate the fan characteristic matrix;
[0018] The fan adjacency matrix is generated based on the number of remaining fans and the fan codes within the preset influence range of each fan.
[0019] Furthermore, the step of inputting the fan feature matrix and the fan adjacency matrix into a pre-trained graph neural model to obtain output fan spatial features includes:
[0020] The fan feature matrix and the fan adjacency matrix are sequentially input into the graph neural model based on time sequence to obtain the fan spatial features.
[0021] Furthermore, before the step of inputting the fan feature matrix and the fan adjacency matrix into a pre-trained graph neural model to obtain output fan spatial features:
[0022] Receiving a historical wind turbine feature matrix and a historical wind turbine adjacency matrix, and using the corresponding historical power generation as labels of the historical wind turbine feature matrix and the historical wind turbine adjacency matrix to generate training data;
[0023] Based on the training data, a preset initial graph neural model and an initial power generation prediction model are trained until the initial graph neural model and the initial power generation prediction model converge, thereby obtaining the graph neural model and the power generation prediction model.
[0024] In order to solve the above technical problems, the embodiment of the present application also provides a wind power generation prediction device based on a graph neural model, which adopts the following technical solution:
[0025] A wind power generation prediction device based on a graph neural model, comprising:
[0026] An acquisition module is used to acquire the characteristics of the fans and the relationship between the fans in a preset space, and generate a fan characteristic matrix and a fan adjacency matrix respectively;
[0027] An input module, used for inputting the fan feature matrix and the fan adjacency matrix into a pre-trained graph neural model to obtain output fan spatial features;
[0028] The prediction module is used to input the wind turbine spatial characteristics into a pre-trained power generation prediction model to obtain an output wind turbine power generation prediction value.
[0029] Further, the acquisition module includes a first fan characteristic generation submodule and a second fan characteristic generation submodule;
[0030] The first fan feature generation submodule is used to obtain the features of each fan, and combine the features of the fans to generate the fan feature matrix;
[0031] The second wind turbine feature generation submodule is used to generate the wind turbine adjacency matrix based on the number of remaining wind turbines and wind turbine codes within a preset influence range of each wind turbine.
[0032] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:
[0033] A computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the above-mentioned wind power generation prediction method based on a graph neural model are implemented.
[0034] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:
[0035] A computer-readable storage medium having computer-readable instructions stored thereon, wherein the computer-readable instructions, when executed by a processor, implement the steps of the above-mentioned method for predicting wind power generation based on a graph neural model.
[0036] Compared with the prior art, the embodiments of the present application have the following beneficial effects:
[0037] This application generates a wind turbine feature matrix and a wind turbine adjacency matrix based on the characteristics of the wind turbines and the relationship between the wind turbines, and predicts wind power generation based on the wind turbine feature matrix and the wind turbine adjacency matrix. It can fully consider the spatial topological characteristics of the distribution between wind turbines and improve the accuracy of wind power generation prediction. At the same time, the spatial characteristics of the wind turbine are extracted through the graph neural model, and then the spatial characteristics of the wind turbine are input into the power generation prediction model to predict wind power generation, that is, through the cooperation between the graph neural model and the power generation prediction model, accurate prediction of wind power generation is achieved. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0039] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;
[0040] Figure 2 is a flow chart of an embodiment of a method for predicting wind power generation based on a graph neural model according to the present application;
[0041] Figure 3 It is a structural schematic diagram of an embodiment of a wind power generation prediction device based on a graph neural model according to the present application;
[0042] Figure 4 It is a structural diagram of an embodiment of a computer device according to the present application.
[0043] Figure numerals: 200, computer device; 201, memory; 202, processor; 203, network interface; 300, wind power generation prediction device based on graph neural model; 301, acquisition module; 302, input module; 303, prediction module. DETAILED DESCRIPTION
[0044] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.
[0045] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.
[0046] 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 accompanying drawings.
[0047] like Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.
[0048] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.
[0049] Terminal devices 101, 102, 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 (Moving Picture Experts Group Audio Layer IV) players, laptop computers, desktop computers, etc.
[0050] The server 105 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal devices 101 , 102 , and 103 .
[0051] It should be noted that the wind power generation prediction method based on the graph neural model provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the wind power generation prediction device based on the graph neural model is generally set in the server / terminal device.
[0052] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.
[0053] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for predicting wind power generation based on a graph neural model according to the present application. The method for predicting wind power generation based on a graph neural model comprises the following steps:
[0054] S1: Obtain the characteristics of the fans and the relationships between the fans in the preset space, and generate a fan feature matrix and a fan adjacency matrix respectively.
[0055] In this embodiment, the preset space refers to a wind farm, and there are N wind turbines in the wind farm. The features of all wind turbines form a wind turbine feature matrix. Each wind turbine is a node, and the relationship between nodes forms an N×N wind turbine adjacency matrix A.
[0056] Specifically, the steps of obtaining the characteristics of the fans and the relationships between the fans in the preset space and respectively generating a fan characteristic matrix and a fan adjacency matrix include:
[0057] Acquire the characteristics of each fan, and combine the characteristics of each fan to obtain the fan characteristic matrix;
[0058] The fan adjacency matrix is generated based on the number of remaining fans and the fan codes within the preset influence range of each fan.
[0059] In this embodiment, each wind turbine has its own characteristics, and these characteristics constitute an N×D wind turbine characteristic matrix X. Specifically: each wind turbine has D-dimensional characteristics, and there are N wind turbines in total, forming an N×D wind turbine characteristic matrix. Among them, the characteristics of the wind turbine include at least wind turbine equipment parameters, location, wind force and wind direction. Among them, information such as wind turbine equipment parameters and location are received from the user device, and the location can also be obtained by positioning according to GPS, and the wind force and wind direction are received from the wind force and wind direction measurement device. The wind turbine adjacency matrix is generated based on the number of wind turbines and wind turbine codes of the remaining wind turbines within the preset influence range of each wind turbine. For example, if wind turbine No. 1 is surrounded by wind turbines No. 2 and No. 3, the second column and the third column of the first row of the wind turbine adjacency matrix are 1, and the rest of the positions in the first row are 0. If wind turbine No. 2 is surrounded by wind turbines No. 3 and No. 5, the third column and the fifth column of the second row of the wind turbine adjacency matrix are 1, and the rest of the positions in the second row are 0. And so on, the wind turbine adjacency matrix is generated.
[0060] In this embodiment, the D-dimensional features of the wind turbine include characteristic variables such as equipment parameters, position, and state feature codes (eg, wind force, wind direction).
[0061] S2: Input the fan feature matrix and the fan adjacency matrix into a pre-trained graph neural model to obtain output fan spatial features.
[0062] In this embodiment, the graph neural network model is a GCN model, and the fan feature matrix X and the fan adjacency matrix A (also called the relationship matrix) are input into the GCN model together to extract the fan spatial features.
[0063] Specifically, the step of inputting the fan feature matrix and the fan adjacency matrix into a pre-trained graph neural model to obtain the output fan spatial features includes:
[0064] The fan feature matrix and the fan adjacency matrix are sequentially input into the graph neural model based on time sequence to obtain the fan spatial features.
[0065] In this embodiment, the data of the historical time series with a length of n is used as input, wherein each moment of the historical time series has a corresponding fan feature matrix X, and the historical time series refers to a continuous period of time. For example, n fan feature matrices X from time tn to time t. Since the fan feature matrix and the fan adjacency matrix are input based on the time sequence, the fan spatial features are also output based on the time sequence, and the fan spatial features are subsequently input into the bidirectional long short-term memory model, also according to the time sequence.
[0066] The graph neural model includes a Fourier domain filter and a plurality of hidden layers, and the step of inputting the fan feature matrix and the fan adjacency matrix into the pre-trained graph neural model to extract the fan spatial features and obtain the output fan spatial features includes:
[0067] The fan characteristic matrix and the fan adjacency matrix are passed through the Fourier domain filter, and the spatial characteristics between all fans are captured through the first-order neighborhood of the Fourier domain recorder to obtain initial fan spatial characteristics;
[0068] The initial fan spatial features are passed through the multiple hidden layers to obtain the fan spatial features.
[0069] In this embodiment, the present application extracts spatial dependency features (i.e., fan spatial features): for the fan adjacency matrix A and the fan feature matrix X. The GCN model constructs a filter in the Fourier domain. The filter acts on all fans, and the filter captures the spatial features between all fans at time t through its first-order neighborhood, and then constructs the GCN model by superimposing multiple hidden layers. If a certain node is a central fan, the GCN model can obtain the topological relationship between the central fan and the surrounding fans (i.e., the fan adjacency matrix), encode the fan adjacency matrix A and the fan feature matrix X, and obtain spatial dependency features (i.e., fan spatial features). Multiple hidden layers are actually multi-level abstractions of input features, and the ultimate goal is to better linearly divide different types of data (the role of hidden layers). In other words, in order to solve nonlinear problems, hidden layers are introduced. The role of the hidden layer is like mapping the input to the output when softmax is running. If the labels in this mapping can still correspond one to one, this mapping is valid, so the mapping process must be linear. This process feels similar to the kernel function of the support vector machine, which projects the data onto other planes to better distinguish features.
[0070] Fourier's idea is that any function can be expressed as the sum of sines / cosines of different frequencies. The Fourier domain transforms many functions into the frequency domain, and the filter performs filtering in the frequency domain, which is generally called the Fourier domain filter. The Fourier idea is to consider the seemingly chaotic signal as a combination of basic sine (cosine) signals with certain amplitude, phase, and frequency. The purpose of Fourier's idea is to find the frequency corresponding to the signal with larger amplitude (higher energy) among these basic sine (cosine) signals, so as to find the main vibration frequency characteristics in the chaotic signal.
[0071] S3: Inputting the wind turbine spatial features into a pre-trained power generation prediction model to obtain an output wind turbine power generation prediction value.
[0072] In this embodiment, power generation is predicted based on the spatial characteristics of wind turbines, which fully considers the positional relationship between wind turbines and improves the accuracy of prediction.
[0073] Specifically, the power generation prediction model includes a bidirectional long short-term memory model and a fully connected layer, and the step of inputting the wind turbine spatial features into the pre-trained power generation prediction model to obtain the output wind turbine power generation prediction value includes:
[0074] Inputting the fan spatial features into the bidirectional long short-term memory model in sequence based on the time sequence, memorizing the fan spatial features at the historical moment based on the input gate, forget gate and output gate in the bidirectional long short-term memory model, and fusing them with the fan spatial features at the current moment to obtain the output fan memory features;
[0075] The wind turbine memory feature is input into the fully connected layer to perform probability prediction on the wind turbine memory feature to obtain a predicted value of the wind turbine power generation at the next moment.
[0076] In this embodiment, the spatial characteristics of the wind turbine are input into the bidirectional long short-term memory model (BiLSTM) according to the time sequence to obtain the dynamic information changes between each node, that is, the temporal characteristics are obtained, and the power generation prediction model is implemented to consider the temporal characteristics of the wind turbine and the spatial characteristics of the wind turbine, thereby improving the accuracy of wind power generation prediction. Finally, through a fully connected layer, the predicted value of wind turbine power generation is obtained, which is the power generation of the wind farm at the next time T. The fully connected layers (FC) play the role of "classifier" in the entire network.
[0077] Recurrent neural network (RNN) is not suitable for long-term prediction due to gradient explosion. LSTM, as a variant of recurrent neural network, overcomes the above problems. Its principle is to use gate-level mechanism to store as long period information as possible. On the basis of ordinary RNN, memory units are added to each neural unit in the hidden layer, so that the memory information on the time series can be controlled. Each time it is transmitted between the units in the hidden layer, several controllable gates (forget gate, input gate, candidate gate, output gate) can be used to control the memory and forgetting degree of previous information and current information, so that the RNN network has long-term memory function, which has a huge effect on the practical application of RNN. However, the unidirectional LSTM can only predict the future through historical states. In order to extract the time-dependent information with complete time, this application adopts a bidirectional long short-term memory model (BiLSTM). In the bidirectional long short-term memory model (BiLSTM), each training sequence is two LSTMs forward and backward respectively, and is connected to the same output layer. This structure provides the output layer with complete past and future context information for each point in the input sequence. This application inputs the spatial features of the fan into a bidirectional long short-term memory model (BiLSTM) to extract and predict forward and backward time-dependent features.
[0078] In some optional implementations of this embodiment, before step S2, that is, before the step of inputting the fan feature matrix and the fan adjacency matrix into the pre-trained graph neural model to obtain the output fan spatial features, the electronic device may further perform the following steps:
[0079] Receiving a historical wind turbine feature matrix and a historical wind turbine adjacency matrix, and using the corresponding historical power generation as labels of the historical wind turbine feature matrix and the historical wind turbine adjacency matrix to generate training data;
[0080] Based on the training data, a preset initial graph neural model and an initial power generation prediction model are trained until the initial graph neural model and the initial power generation prediction model converge, thereby obtaining the graph neural model and the power generation prediction model.
[0081] In this embodiment, the initial graph neural model and the initial power generation prediction model are trained by using the historical power generation as the label of the historical wind turbine feature matrix and the historical wind turbine adjacency matrix, so that the graph neural model and the power generation prediction model can be used for power generation prediction.
[0082] This application takes into account the influencing factors of both space and time, especially the influence between wind turbines in wind farms. Therefore, the model shows good performance in prediction accuracy, spatiotemporal prediction ability and long-term prediction ability, and has strong applicability. Effective and accurate short-term prediction of wind power generation can provide a reference for power dispatching and promote the promotion and application of wind power generation.
[0083] This application generates a wind turbine feature matrix and a wind turbine adjacency matrix based on the characteristics of the wind turbines and the relationship between the wind turbines, and predicts wind power generation based on the wind turbine feature matrix and the wind turbine adjacency matrix. It can fully consider the spatial topological characteristics of the distribution between wind turbines and improve the accuracy of wind power generation prediction. At the same time, the spatial characteristics of the wind turbine are extracted through the graph neural model, and then the spatial characteristics of the wind turbine are input into the power generation prediction model to predict wind power generation, that is, through the cooperation between the graph neural model and the power generation prediction model, accurate prediction of wind power generation is achieved.
[0084] In this embodiment, the wind power generation prediction method based on the graph neural model runs on an electronic device (e.g. Figure 1 The server / terminal device shown in the figure can receive the historical wind turbine feature matrix and the historical wind turbine adjacency matrix through a wired connection or a wireless connection. It should be noted that the above wireless connection method can include but is not limited to 3G / 4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other wireless connection methods currently known or to be developed in the future.
[0085] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the computer-readable instructions are executed, they can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).
[0086] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.
[0087] Further references Figure 3 , as a response to the above Figure 2The present application provides an embodiment of a wind power generation prediction device based on a graph neural model. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.
[0088] like Figure 3 As shown, the wind power generation prediction device 300 based on the graph neural model described in this embodiment includes: an acquisition module 301, an input module 302 and a prediction module 303. Among them: the acquisition module 301 is used to obtain the characteristics of the wind turbines and the relationship between the wind turbines in the preset space, and generate a wind turbine feature matrix and a wind turbine adjacency matrix respectively; the input module 302 is used to input the wind turbine feature matrix and the wind turbine adjacency matrix into the pre-trained graph neural model to obtain the output wind turbine space characteristics; the prediction module 303 is used to input the wind turbine space characteristics into the pre-trained power generation prediction model to obtain the output wind turbine power generation prediction value.
[0089] In this embodiment, the present application generates a wind turbine feature matrix and a wind turbine adjacency matrix based on the characteristics of the wind turbines and the relationship between the wind turbines, and predicts wind power generation based on the wind turbine feature matrix and the wind turbine adjacency matrix. It can fully consider the spatial topological characteristics of the distribution between wind turbines and improve the accuracy of wind power generation prediction. At the same time, the spatial characteristics of the wind turbine are extracted through the graph neural model, and then the spatial characteristics of the wind turbine are input into the power generation prediction model to predict wind power generation, that is, through the cooperation between the graph neural model and the power generation prediction model, accurate prediction of wind power generation is achieved.
[0090] In some optional implementations of this embodiment, the acquisition module 301 includes a first fan feature generation submodule and a second fan feature generation submodule;
[0091] The first fan feature generation submodule is used to obtain the features of each fan, and combine the features of the fans to generate the fan feature matrix;
[0092] The second wind turbine feature generation submodule is used to generate the wind turbine adjacency matrix based on the number of remaining wind turbines and wind turbine codes within a preset influence range of each wind turbine.
[0093] In some optional implementations of this embodiment, the input module 302 is further used to: input the fan feature matrix and the fan adjacency matrix into the graph neural model in sequence based on time order to obtain the fan spatial features.
[0094] In some optional implementations of this embodiment, the graph neural model includes a Fourier domain filter and a hidden layer, and the above-mentioned input module 302 includes a filtering submodule and an acquisition submodule, wherein the filtering submodule is used to pass the fan feature matrix and the fan adjacency matrix through the Fourier domain filter, capture the spatial characteristics between all fans through the first-order neighborhood of the Fourier domain recorder, and obtain the initial fan spatial characteristics; the acquisition submodule is used to pass the initial fan spatial characteristics through the multiple hidden layers to obtain the fan spatial characteristics.
[0095] The power generation prediction model includes a bidirectional long short-term memory model and a fully connected layer, and the prediction module 303 includes an input submodule and a prediction submodule, wherein the input submodule is used to input the wind turbine spatial features into the bidirectional long short-term memory model in sequence based on time sequence, and based on the input gate, forget gate and output gate in the bidirectional long short-term memory model, remember the wind turbine spatial features at historical moments, and fuse them with the wind turbine spatial features at the current moment to obtain the output wind turbine memory features; the prediction submodule is used to input the wind turbine memory features into the fully connected layer to perform probability prediction on the wind turbine memory features to obtain the predicted value of the wind turbine power generation at the next moment.
[0096] In some optional implementations of the present embodiment, the above-mentioned device 300 also includes: a receiving module and a training module, wherein the receiving module is used to receive a historical wind turbine feature matrix and a historical wind turbine adjacency matrix, and use the corresponding historical power generation as labels of the historical wind turbine feature matrix and the historical wind turbine adjacency matrix to generate training data; the training module is used to train a preset initial graph neural model and an initial power generation prediction model based on the training data until the initial graph neural model and the initial power generation prediction model converge to obtain the graph neural model and the power generation prediction model.
[0097] This application generates a wind turbine feature matrix and a wind turbine adjacency matrix based on the characteristics of the wind turbines and the relationship between the wind turbines, and predicts wind power generation based on the wind turbine feature matrix and the wind turbine adjacency matrix. It can fully consider the spatial topological characteristics of the distribution between wind turbines and improve the accuracy of wind power generation prediction. At the same time, the spatial characteristics of the wind turbine are extracted through the graph neural model, and then the spatial characteristics of the wind turbine are input into the power generation prediction model to predict wind power generation, that is, through the cooperation between the graph neural model and the power generation prediction model, accurate prediction of wind power generation is achieved.
[0098] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.
[0099] The computer device 200 includes a memory 201, a processor 202, and a network interface 203 that are interconnected and communicated through a system bus. It should be noted that the figure only shows a computer device 200 having components 201-203, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), field-programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.
[0100] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.
[0101] The memory 201 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 201 can be an internal storage unit of the computer device 200, such as a hard disk or memory of the computer device 200. In other embodiments, the memory 201 can also be an external storage device of the computer device 200, such as a plug-in hard disk equipped on the computer device 200, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Of course, the memory 201 can also include both the internal storage unit of the computer device 200 and its external storage device. In this embodiment, the memory 201 is generally used to store an operating system and various application software installed on the computer device 200, such as computer-readable instructions of a wind power generation prediction method based on a graph neural model. In addition, the memory 201 can also be used to temporarily store various data that have been output or are to be output.
[0102] The processor 202 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 202 is generally used to control the overall operation of the computer device 200. In this embodiment, the processor 202 is used to run computer-readable instructions or process data stored in the memory 201, such as computer-readable instructions for running the wind power generation prediction method based on the graph neural model.
[0103] The network interface 203 may include a wireless network interface or a wired network interface. The network interface 203 is generally used to establish a communication connection between the computer device 200 and other electronic devices.
[0104] In this embodiment, the wind power generation is predicted based on the wind turbine feature matrix and the wind turbine adjacency matrix. The spatial topological characteristics of the distribution between wind turbines can be fully considered to improve the accuracy of the wind power generation prediction. Through the cooperation between the graph neural model and the power generation prediction model, the accurate prediction of wind power generation is achieved.
[0105] The present application also provides another embodiment, namely, providing a computer-readable storage medium, which stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the wind power generation prediction method based on the graph neural model as described above.
[0106] In this embodiment, the wind power generation is predicted based on the wind turbine feature matrix and the wind turbine adjacency matrix. The spatial topological characteristics of the distribution between wind turbines can be fully considered to improve the accuracy of the wind power generation prediction. Through the cooperation between the graph neural model and the power generation prediction model, the accurate prediction of wind power generation is achieved.
[0107] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods 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, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.
[0108] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.
Claims
1. A wind power generation prediction method based on graph neural model, It is characterized in that The steps include: Obtain the characteristics of the fans and the relationships between the fans in the preset space, and generate a fan characteristic matrix and a fan adjacency matrix respectively; Inputting the fan feature matrix and the fan adjacency matrix into a pre-trained graph neural model to extract fan spatial features, and obtaining output fan spatial features; Inputting the wind turbine spatial features into a pre-trained power generation prediction model to obtain an output wind turbine power generation prediction value; The power generation prediction model includes a bidirectional long short-term memory model and a fully connected layer. The step of inputting the wind turbine spatial features into the pre-trained power generation prediction model to obtain the output wind turbine power generation prediction value includes: Input the fan spatial features into the bidirectional long short-term memory model in sequence based on the time sequence, obtain the dynamic information changes between the nodes, and obtain the fan time features. Based on the input gate, forget gate and output gate in the bidirectional long short-term memory model, the fan spatial features at the historical moment are memorized, and are merged with the fan spatial features at the current moment to obtain the output fan memory features, wherein each fan is regarded as a node; Inputting the wind turbine memory feature into the fully connected layer to perform probability prediction on the wind turbine memory feature to obtain a predicted value of the wind turbine power generation at the next moment; The step of obtaining the characteristics of the fans and the relationship between the fans in the preset space and generating a fan characteristic matrix and a fan adjacency matrix respectively comprises: Acquire the characteristics of each fan, and combine the characteristics of the fans to generate the fan characteristic matrix; Generating the wind turbine adjacency matrix based on the number of wind turbines and wind turbine codes of the remaining wind turbines within a preset influence range of each wind turbine; The characteristics of the wind turbine include wind turbine equipment parameters, location, wind force and wind direction, wherein the wind turbine equipment parameters and location information are received from user equipment, and the wind force and wind direction are received from wind force and direction measuring equipment.
2. According to claim 1, the wind power generation prediction method based on the graph neural model, It is characterized in that The graph neural model includes a Fourier domain filter and a plurality of hidden layers. The step of inputting the fan feature matrix and the fan adjacency matrix into the pre-trained graph neural model to extract the fan spatial features and obtaining the output fan spatial features includes: The fan feature matrix and the fan adjacency matrix are passed through the Fourier domain filter, and the spatial characteristics between all fans are captured through the first-order neighborhood of the Fourier domain filter to obtain initial fan spatial characteristics; The initial fan space feature is passed through the multiple hidden layers to perform multi-level abstraction on the initial fan space feature to obtain the fan space feature.
3. The wind power generation prediction method based on the graph neural model according to claim 1, It is characterized in that The step of inputting the fan feature matrix and the fan adjacency matrix into a pre-trained graph neural model to obtain output fan spatial features includes: The fan feature matrix and the fan adjacency matrix are sequentially input into the graph neural model based on time sequence to obtain the fan spatial features.
4. The wind power generation prediction method based on the graph neural model according to claim 1, It is characterized in that Before the step of inputting the fan feature matrix and the fan adjacency matrix into a pre-trained graph neural model to obtain output fan spatial features: Receiving a historical wind turbine feature matrix and a historical wind turbine adjacency matrix, and using the corresponding historical power generation as labels of the historical wind turbine feature matrix and the historical wind turbine adjacency matrix to generate training data; Based on the training data, a preset initial graph neural model and an initial power generation prediction model are trained until the initial graph neural model and the initial power generation prediction model converge, thereby obtaining the graph neural model and the power generation prediction model.
5. A wind power generation prediction device based on graph neural model, It is characterized in that The wind power generation prediction device based on the graph neural model implements the steps of the wind power generation prediction method based on the graph neural model according to any one of claims 1 to 4 when executed, and the wind power generation prediction device based on the graph neural model includes: An acquisition module is used to acquire the characteristics of the fans and the relationship between the fans in a preset space, and generate a fan characteristic matrix and a fan adjacency matrix respectively; An input module, used for inputting the fan feature matrix and the fan adjacency matrix into a pre-trained graph neural model to obtain output fan spatial features; The prediction module is used to input the wind turbine spatial characteristics into a pre-trained power generation prediction model to obtain an output wind turbine power generation prediction value.
6. The wind power generation prediction device based on the graph neural model according to claim 5, It is characterized in that The acquisition module includes a first fan characteristic generation submodule and a second fan characteristic generation submodule; The first fan feature generation submodule is used to obtain the features of each fan, and combine the features of the fans to generate the fan feature matrix; The second wind turbine feature generation submodule is used to generate the wind turbine adjacency matrix based on the number of remaining wind turbines and wind turbine codes within a preset influence range of each wind turbine.
7. A computer device, It is characterized in that It includes a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the wind power generation prediction method based on the graph neural model as described in any one of claims 1 to 4 are implemented.
8. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the method for predicting wind power generation based on a graph neural model as described in any one of claims 1 to 4.
