Wind power plant layout configuration method, device, equipment, system and program product

Abstract

The embodiment of the invention provides a wind power plant layout configuration method. The method comprises the following steps: acquiring working parameter information corresponding to a plurality of to-be-configured fans and interference information among the fans; generating wind power plant graph structure data based on the working parameter information corresponding to the plurality of to-be-configured fans and the interference information among the fans; and inputting the wind power plant graph structure data into a pre-trained graph neural network to obtain wind power plant layout configuration information. As each wind power plant is used as a node in the graph neural network to carry out feature design adaptive to the environment, and the complex relationship among the nodes is captured by utilizing the edges in the graph neural network, the interference among the wind power plants is simulated, and the layout configuration process of the wind power plants is realized. Therefore, the method can be suitable for complex environmental conditions and large-scale wind power plant layout configuration processes, and the accuracy of wind power plant layout configuration is improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular, to a method, device, equipment, system and program product for wind farm layout configuration. Background Art

[0002] The layout optimization of a wind farm is a key issue in the design and operation of a wind farm. The layout of the wind farm directly affects the power generation capacity and overall efficiency of the wind turbines.

[0003] Generally, the layout optimization methods of wind farms mostly rely on physical models and heuristic algorithms, such as genetic algorithms, particle swarm optimization, etc.

[0004] However, with the complexity of the environment during the wind farm configuration process and the increase in the number of wind turbines, the above methods are difficult to handle complex environmental factors and non-linear interference between wind turbines, affecting the accuracy of wind farm layout configuration. Summary of the Invention

[0005] In view of this, this application provides a method for wind farm layout configuration to improve the accuracy of wind farm layout configuration.

[0006] According to the first aspect of the embodiments of this application, a method for wind farm layout configuration is provided, which can be applied to a system or program with a wind farm layout configuration function in a terminal device, and specifically includes:

[0007] Obtain the working parameter information corresponding to multiple wind turbines to be configured and the interference information between each wind turbine;

[0008] Based on the working parameter information corresponding to the multiple wind turbines to be configured and the interference information between each wind turbine, generate wind farm graph structure data, where the nodes of the wind farm graph structure data are used to represent the working parameter information corresponding to each of the multiple wind turbines to be configured, and the edges between the nodes of the wind farm graph structure data are used to represent the interference information between each wind turbine;

[0009] Input the wind farm graph structure data into a pre-trained graph neural network to obtain wind farm layout configuration information.

[0010] Optionally, in some possible implementation manners, the generating wind farm graph structure data based on the working parameter information corresponding to the multiple wind turbines to be configured and the interference information between each wind turbine includes:

[0011] Based on the working parameter information corresponding to the multiple wind turbines to be configured, determine the environmental parameters and configuration parameters corresponding to each of the wind turbines to be configured;

[0012] Configure node features through the environmental parameters and the configuration parameters;

[0013] Aggregate the interference information among the individual wind turbines according to the association relationship among the individual wind turbines to obtain edge features;

[0014] Associate the node features and the edge features to generate the wind farm graph structure data.

[0015] Optionally, in some possible implementation manners, the training process of the graph neural network includes:

[0016] Obtain simulation data and real data for training;

[0017] Determine the wind farm layout rule information indicated by the simulation data, and perform supervised training on the graph neural network based on the wind farm layout rule information;

[0018] Determine the actual layout information of the wind farm indicated by the real data, and perform reinforcement training on the graph neural network after supervised training based on the actual layout information of the wind farm.

[0019] Optionally, in some possible implementation manners, the determining the wind farm layout rule information indicated by the simulation data, and performing supervised training on the graph neural network based on the wind farm layout rule information includes:

[0020] Determine the wind farm layout rule information indicated by the simulation data and training nodes;

[0021] Estimate the expected power generation corresponding to the training nodes through the environmental parameters and operating parameters corresponding to the training nodes;

[0022] Configure training interference parameters through the wake effect between the training nodes indicated by the training nodes;

[0023] Configure a loss function with the goal of maximizing the expected power generation and minimizing the training interference parameters;

[0024] Calculate the loss function according to the wind farm layout rule information to perform supervised training on the graph neural network.

[0025] Optionally, in some possible implementation manners, the method further includes:

[0026] Obtain a test set corresponding to the simulation data and the real data;

[0027] Perform performance evaluation on the graph neural network according to evaluation metrics based on the test set to obtain evaluation information, where the evaluation metrics include power generation, interference intensity, and layout efficiency;

[0028] Adjust the parameters of the graph neural network according to a preset strategy under the guidance of the evaluation information.

[0029] Optionally, in some possible implementation manners, the inputting the wind farm graph structure data into the pre-trained graph neural network to obtain the wind farm layout configuration information includes:

[0030] Determining the configuration situation information corresponding to each of the to-be-configured wind turbines;

[0031] Marking the to-be-configured wind turbines based on the configuration situation information to update the wind farm graph structure data;

[0032] Inputting the updated wind farm graph structure data into the graph neural network to obtain the wind farm layout configuration information, adjusting the device parameters of the to-be-configured wind turbines whose marked nodes in the wind farm layout configuration information are indicated as established nodes, adjusting the device parameters of the to-be-configured wind turbines whose marked nodes in the wind farm layout configuration information are indicated as unestablished nodes, and performing position configuration.

[0033] According to a second aspect of the embodiments of the present application, there is provided a wind farm layout configuration device, including:

[0034] An obtaining unit, configured to obtain the working parameter information corresponding to a plurality of to-be-configured wind turbines and the interference information between each pair of wind turbines;

[0035] A configuration unit, configured to generate wind farm graph structure data based on the working parameter information corresponding to the plurality of to-be-configured wind turbines and the interference information between each pair of wind turbines, where the nodes of the wind farm graph structure data are used to represent the working parameter information corresponding to each of the plurality of to-be-configured wind turbines, and the edges between the nodes of the wind farm graph structure data are used to represent the interference information between each pair of wind turbines;

[0036] The configuration unit is further configured to input the wind farm graph structure data into the pre-trained graph neural network to obtain the wind farm layout configuration information.

[0037] Optionally, in some possible implementation manners, when generating the wind farm graph structure data based on the working parameter information corresponding to the plurality of to-be-configured wind turbines and the interference information between each pair of wind turbines, the configuration unit is configured to determine the environmental parameters and configuration parameters corresponding to each of the to-be-configured wind turbines based on the working parameter information corresponding to the plurality of to-be-configured wind turbines; configure the node features through the environmental parameters and the configuration parameters; aggregate the interference information between each pair of wind turbines according to the association relationship between each pair of wind turbines to obtain edge features; and associate the node features and the edge features to generate the wind farm graph structure data.

[0038] Optionally, in some possible implementation manners, the configuration unit is configured to obtain simulated data and real data for training; determine the wind farm layout rule information indicated by the simulated data, so as to perform supervised training on the graph neural network based on the wind farm layout rule information; determine the actual layout information of the wind farm indicated by the real data, and perform reinforcement training on the supervised-trained graph neural network based on the actual layout information of the wind farm.

[0039] Optionally, in some possible implementation manners, when the configuration unit determines the wind farm layout rule information indicated by the simulated data to perform supervised training on the graph neural network based on the wind farm layout rule information, the configuration unit is configured to determine the wind farm layout rule information and training nodes indicated by the simulated data; estimate the expected power generation amount corresponding to the training nodes through the environmental parameters and operating parameters corresponding to the training nodes; configure training interference parameters through the wake effect between the training nodes indicated by the training nodes; configure a loss function with the goal of maximizing the expected power generation amount and minimizing the training interference parameters; calculate the loss function according to the wind farm layout rule information to perform supervised training on the graph neural network.

[0040] Optionally, in some possible implementation manners, the configuration unit is configured to obtain a test set corresponding to the simulated data and the real data; perform performance evaluation on the graph neural network according to evaluation metrics based on the test set to obtain evaluation information, where the evaluation metrics include power generation amount, interference intensity, and layout efficiency; adjust the parameters of the graph neural network according to a preset strategy under the guidance of the evaluation information.

[0041] Optionally, in some possible implementation manners, when the configuration unit inputs the wind farm graph structure data into a pre-trained graph neural network to obtain wind farm layout configuration information, the configuration unit is configured to determine the configuration situation information corresponding to each to-be-configured fan; mark the to-be-configured fans based on the configuration situation information to update the wind farm graph structure data; input the updated wind farm graph structure data into the graph neural network to obtain the wind farm layout configuration information, and adjust the device parameters of the to-be-configured fans whose marked nodes in the wind farm layout configuration information are indicated as established nodes, and adjust the device parameters and perform position configuration on the to-be-configured fans whose marked nodes in the wind farm layout configuration information are indicated as unestablished nodes.

[0042] According to a third aspect of the embodiments of the present application, a wind farm layout configuration device is provided, including an input / output component and a processor;

[0043] The input / output component is configured to obtain the working parameter information corresponding to a plurality of to-be-configured fans and the interference information between each fan;

[0044] The processor is configured to perform layout configuration on multiple wind turbines to be configured determined by the input / output component by executing the wind farm layout configuration method described in the first aspect or any implementation manner of the first aspect as described above.

[0045] According to a fourth aspect of the embodiments of the present application, there is provided a wind farm layout configuration system, including an interaction client and a server;

[0046] The interaction client is configured to obtain working parameter information corresponding to multiple wind turbines to be configured and interference information between each pair of wind turbines, and send the working parameter information corresponding to the multiple wind turbines to be configured and the interference information between each pair of wind turbines to the server, and display the wind farm layout configuration result output by the server;

[0047] The server is configured to perform layout configuration on multiple wind turbines to be configured determined by the interaction client by executing the wind farm layout configuration method described in the first aspect or any implementation manner of the first aspect.

[0048] According to a fifth aspect of the embodiments of the present application, there is provided a computer program product, including: a computer program, which when executed by a processor, implements the wind farm layout configuration method described in the first aspect or any implementation manner of the first aspect.

[0049] It can be seen from the above technical embodiments that the embodiments of the present application have the following advantages:

[0050] By obtaining the working parameter information corresponding to multiple wind turbines to be configured and the interference information between each pair of wind turbines; then generating wind farm graph structure data based on the working parameter information corresponding to the multiple wind turbines to be configured and the interference information between each pair of wind turbines, where the nodes of the wind farm graph structure data are used to represent the working parameter information corresponding to each of the multiple wind turbines to be configured, and the edges between the nodes of the wind farm graph structure data are used to represent the interference information between each pair of wind turbines; and inputting the wind farm graph structure data into a pre-trained graph neural network to obtain wind farm layout configuration information. Thus, a reliable wind farm layout configuration process is realized. Since each wind farm is used as a node in the graph neural network for feature design adapted to the environment, and the edges in the graph neural network are used to capture the complex relationships between the nodes to simulate the interference between wind farms, it can be applied to complex environmental conditions and large-scale wind farm layout configuration processes, improving the accuracy of wind farm layout configuration. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] To more clearly illustrate the technical embodiments in the embodiments of the present application or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only the embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on the provided accompanying drawings.

[0052] Figure 1 It is a network architecture diagram for the operation of the wind farm layout configuration system;

[0053] Figure 2 It is a process architecture diagram for a wind farm layout configuration provided by an embodiment of the present application;

[0054] Figure 3 It is a flowchart of a method for wind farm layout configuration provided by an embodiment of the present application;

[0055] Figure 4 It is a scenario schematic diagram of a method for wind farm layout configuration provided by an embodiment of the present application;

[0056] Figure 5 It is a structural schematic diagram of a device for wind farm layout configuration provided by an embodiment of the present application;

[0057] Figure 6 It is a structural schematic diagram of a terminal device provided by an embodiment of the present application. Detailed implementation manners

[0058] The following will clearly and completely describe the technical embodiments in the embodiments of the present application in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0059] It should be understood that the wind farm layout configuration method provided by the present application can be applied to a system or program with a wind farm layout configuration function in a terminal device, such as an energy management application. Specifically, the wind farm layout configuration system can run in a network architecture as Figure 1 shown. As Figure 1 shown, it is a network architecture diagram for the operation of the wind farm layout configuration system. As can be seen from the figure, the wind farm layout configuration system can provide a wind farm layout configuration process with multiple information sources, that is, determine the wind turbines to be configured through the acquisition operation on the terminal side, and then send them to the server for layout configuration of the wind turbines to be configured. It can be understood that Figure 1A variety of terminal devices are shown. The terminal device can be a computer device. In actual scenarios, there can be more or fewer types of terminal devices participating in the process of wind farm layout configuration. The specific quantity and types depend on the actual scenario and are not limited here. Additionally, Figure 1 a server is shown, but in actual scenarios, multiple servers can also participate, especially in multi-disciplinary pair scenarios. The specific number of servers depends on the actual scenario.

[0060] In this embodiment, the server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal can be a smart phone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, etc., but is not limited thereto. The terminal and the server can be directly or indirectly connected through wired or wireless communication methods. The terminal and the server can be connected to form a blockchain network, which is not limited in this application.

[0061] It can be understood that the above wind farm layout configuration system can run on a personal mobile terminal, for example, as an application such as energy management, or can run on a server, or can also run on a third-party device to provide wind farm layout configuration to obtain the wind farm layout configuration processing result of the information source. The specific wind farm layout configuration system can run in the above devices in the form of a program, or can run as a system component in the above devices, or can also be a type of cloud service program. The specific operation mode depends on the actual scenario and is not limited here.

[0062] Wind farm layout optimization is a key issue in wind farm design and operation. The layout of the wind farm directly affects the power generation capacity and overall efficiency of the wind turbines.

[0063] Generally, the methods for wind farm layout optimization mostly rely on physical models and heuristic algorithms, such as genetic algorithms, particle swarm optimization, etc.

[0064] However, with the complexity of the environment during the wind farm configuration process and the increase in the number of wind turbines, the above methods are difficult to handle complex environmental factors and non-linear interference between wind turbines, affecting the accuracy of wind farm layout configuration.

[0065] To solve the above problems, this application proposes a wind farm layout configuration method, which is applied to Figure 2 the process framework of wind farm layout configuration shown, as Figure 2As shown in the figure, it is a flow architecture diagram of a wind farm layout configuration provided by an embodiment of the present application. Multiple wind turbines to be configured are determined through a terminal, and then the server performs feature configuration on the working parameter information and interference information corresponding to the wind turbines to obtain graph structure data, and then inputs it into a graph neural network for skill-based layout configuration.

[0066] It can be understood that the wind farm layout configuration method provided by the present application can be a program written as a processing logic in a hardware system, or can be a wind farm layout configuration device, which realizes the above processing logic in an integrated or external connection manner. As an implementation method, the wind farm layout configuration device obtains the working parameter information corresponding to multiple wind turbines to be configured and the interference information between each wind turbine; then, based on the working parameter information corresponding to the multiple wind turbines to be configured and the interference information between each wind turbine, wind farm graph structure data is generated. The nodes of the wind farm graph structure data are used to represent the working parameter information corresponding to each of the multiple wind turbines to be configured, and the edges between the nodes of the wind farm graph structure data are used to represent the interference information between each wind turbine; and the wind farm graph structure data is input into a pre-trained graph neural network to obtain wind farm layout configuration information. Thus, a reliable wind farm layout configuration process is realized. Since each wind farm is used as a node in the graph neural network for feature design adapted to the environment, and the edges in the graph neural network are used to capture the complex relationships between the nodes to simulate the interference between wind farms, it can be applied to complex environmental conditions and large-scale wind farm layout configuration processes, improving the accuracy of wind farm layout configuration.

[0067] Combined with the above process architecture, the wind farm layout configuration method in the present application will be introduced below. Please refer to Figure 3 , Figure 3 It is a flowchart of a wind farm layout configuration method provided by an embodiment of the present application. The embodiment of the present application at least includes the following steps:

[0068] 301. Obtain the working parameter information corresponding to multiple wind turbines to be configured and the interference information between each wind turbine.

[0069] In this embodiment, the wind turbines to be configured can be established wind turbines, and the layout configuration is to adjust the wind turbine parameters; or they can be wind turbines that have not been established, and the layout configuration process of comprehensively adjusting the position and wind turbine parameters is carried out to improve the power generation of the entire wind farm.

[0070] Specifically, the working parameter information corresponding to the wind turbines to be configured includes multiple dimensions. That is, the wind farm is abstracted as a graph structure, where each node represents a wind turbine, and the working parameter information covers environmental factors such as terrain, wind speed, and wind direction, as well as wind turbine type and operating parameters. Thus, it comprehensively reflects the complex environment and wind turbine characteristics inside the wind farm.

[0071] In addition, the interference information between each pair of wind turbines is used to represent the edges in the graph neural network. That is, the definition of an edge is that it represents the mutual interference between wind turbines, mainly manifested in the wake effect. By calculating the relative distance, wind direction angle, and wind speed attenuation between wind turbines, the interference intensity can be accurately reflected. This definition of the graph structure can more flexibly represent the complex relationships in the wind farm layout.

[0072] It can be understood that considering that traditional methods are difficult to accurately capture the complex non-linear interference and environmental factors between wind turbines. In this embodiment, the graph neural network (Graph Neural Networks, GNN) can effectively model these complex relationships and improve the accuracy of layout optimization. Improve computational efficiency: Existing methods have high computational resource requirements when dealing with large-scale wind farms. GNN optimizes the computational process through deep learning methods, improving the computational efficiency and making it possible to optimize the layout of large-scale wind farms. Enhance adaptability and flexibility: The purpose of this embodiment is to improve the model's adaptability to dynamic environments and diverse conditions. GNN can improve the performance of traditional methods under changing conditions through real-time data updates and adaptive learning. Global optimization ability, solving the problem that traditional heuristic algorithms are limited to local optima. GNN captures the global layout relationship through deep learning, realizes the global optimization of the wind farm layout, and improves the overall power generation efficiency and system stability.

[0073] Next, the generation of wind farm graph structure data will be described in combination with the above-obtained working parameter information and interference information, so as to be adapted to the input of the graph neural network.

[0074] 302. Generate wind farm graph structure data based on the working parameter information corresponding to multiple wind turbines to be configured and the interference information between each pair of wind turbines. The nodes of the wind farm graph structure data are used to represent the working parameter information corresponding to each of the multiple wind turbines to be configured, and the edges between the nodes of the wind farm graph structure data are used to represent the interference information between each pair of wind turbines.

[0075] In this embodiment, the wind farm graph structure data is a process of integrating the features of multiple wind turbines, that is, representing the overall feature dimension of multiple wind turbines to be configured, so as to be adapted to the configuration process of the graph neural network.

[0076] Specifically, for the generation process of the wind farm graph structure data, it is based on the node features and edge features in the graph neural network. That is, first, based on the working parameter information corresponding to multiple wind turbines to be configured, determine the environmental parameters and configuration parameters corresponding to each wind turbine to be configured; then configure the node features through the environmental parameters and configuration parameters; and aggregate the interference information between each pair of wind turbines according to the association relationship between each pair of wind turbines to obtain the edge features; and then associate the node features and edge features to generate the wind farm graph structure data.

[0077] It can be understood that for the configuration of the working parameter information, it is a multi-dimensional feature construction process. In the problem of wind farm layout optimization, traditional methods usually rely on physical models. However, these methods often have difficulty in effectively dealing with multi-dimensional environmental factors and complex interactions between wind turbines. To solve this problem, this embodiment introduces a graph structure, such as Figure 4 shown Figure 4 is a schematic diagram of the scenario of a wind farm layout configuration method provided by an embodiment of the present application. The figure shows that the entire wind farm is abstracted as an undirected graph to better represent the complex relationships inside the wind farm. In the graph structure, each wind turbine is regarded as a node in the graph. The feature vector of the node contains multiple key environmental factors, which directly affect the operating efficiency and power generation of the wind turbine. Specifically, the feature vector includes the following dimensions:

[0078] Terrain features: The terrain where the wind farm is located affects the distribution of wind speed and the change of wind direction. The terrain features of the node can be composed of data such as altitude and terrain undulation. These features help the model understand the impact of the terrain on the wind turbine layout and avoid arranging wind turbines in positions with low wind speed or variable wind direction.

[0079] Wind speed and wind direction: Wind speed and wind direction are the two most important factors affecting the power generation efficiency of wind turbines. The wind speed feature of the node can be obtained through historical wind speed data or a wind resource assessment model, while the wind direction feature represents the main distribution of the wind direction within a specific time period. Reasonable feature design enables the model to identify the optimal orientation for wind turbine arrangement, thereby maximizing power generation.

[0080] Wind turbine type and operating parameters: Different models of wind turbines have different power curves and operating characteristics. This part of the features of the node includes the power capacity, tower height, blade length, etc. of the wind turbine. These parameters will affect the performance of the wind turbine at different wind speeds and thus affect the overall layout effect.

[0081] In addition, for the configuration of edge features, that is, after completing the node feature design, it is necessary to construct the edges in the graph to represent the interference relationship between wind turbines. The interference between wind turbines is mainly reflected in the wake effect, that is, the wind turbine located upwind will obstruct the downwind wind turbine, reducing the wind speed and power generation efficiency of the downwind wind turbine. By calculating the relative distance, wind direction angle and wind speed attenuation between wind turbines, the weight of the edge in the graph is defined to accurately reflect the interference intensity. Through the construction of this graph structure, various complex relationships in the wind farm can be comprehensively and accurately mapped into a mathematical model, laying a foundation for subsequent layout optimization based on graph neural networks. This method is more flexible and extensible than traditional physical models. Especially when dealing with large-scale wind farm layout problems, it can better capture and utilize the complex associations between nodes, thereby improving the rationality and optimization effect of the overall layout.

[0082] The generation process of the above wind farm graph structure data is to regard the wind farm as an undirected graph, where the nodes represent wind turbines and the edges represent the interference between wind turbines. The GNN automatically learns the complex non-linear relationships and environmental factors in the wind farm through deep learning, and optimizes the layout configuration of the wind turbines. Compared with traditional technologies, the GNN can enhance the modeling ability, better capture the complex relationships between nodes, and provide more accurate optimization examples. It also has high computational performance. Through deep learning methods, it reduces the computational time and resource consumption. It achieves excellent generalization ability, can adapt to the environmental conditions of different wind farms, and improves the stability of layout optimization.

[0083] 303. Input the wind farm graph structure data into a pre-trained graph neural network to obtain the wind farm layout configuration information.

[0084] In this embodiment, adapted to the above edge feature configuration, the graph neural network is a multi-layer GNN model that updates the feature representation of nodes through a layer-by-layer information aggregation mechanism. This architecture can gradually capture the information of remote nodes and achieve global layout optimization. Based on the graph neural network, it can more flexibly process the complex graph structure data of the wind farm, capture the complex non-linear relationships between nodes, and has higher adaptability and accuracy compared with traditional methods. And it has high computational efficiency. Through deep learning methods, the graph neural network can perform effective optimization with less computational resources, while traditional heuristic algorithms usually require more computational time and resources. It also has the ability of global optimization. The graph neural network model can achieve the global optimization of the wind farm layout and avoid the local optimum problem in traditional methods.

[0085] Specifically, for the selection of the graph neural network as the core optimization tool. Compared with traditional optimization algorithms such as genetic algorithms and particle swarm algorithms, the GNN can better process graph structure data, especially having significant advantages in capturing the complex non-linear relationships between nodes. The GNN aggregates the information of neighboring nodes by applying a neural network layer to each node in the graph, thereby gradually updating the feature representation of the nodes. Finally, the representation of each node not only contains its own features but also the information passed from neighboring nodes, forming a global understanding of the entire wind farm layout. To achieve this process, a multi-layer GNN model is designed, where each layer performs an information aggregation operation. Specifically, each node receives information from its neighboring nodes (i.e., the wind turbines connected to it), performs a weighted sum through a set of trainable weights, and then processes it through a non-linear activation function. This recursive information aggregation mechanism enables the nodes to gradually capture the information of farther neighboring nodes, thereby achieving global layout optimization.

[0086] Furthermore, a multi-layer GNN refers to a hierarchical recursive information aggregation process. Specifically: The operations of each layer of GNN: At each layer, the features of a node are aggregated with the features of its neighbor nodes (usually a weighted sum or other aggregation operations), and then non-linearly transformed through a neural network (such as an MLP) to update the representation of the node. Among them, the role of the multi-layer structure: The multi-layer design allows nodes to gradually obtain information from more distant neighbors. For example:

[0087] The first layer: The node only aggregates information from its direct neighbors.

[0088] The second layer: The node starts to aggregate information from "neighbors of neighbors".

[0089] The third layer: The node can capture global information at a greater distance. Therefore, the multi-layer GNN will ultimately integrate local information layer by layer to form a global representation of each node, which is suitable for solving global optimization problems, such as the layout optimization of a wind farm.

[0090] Next, in combination with the above structural configuration of the graph neural network, the training process of the graph neural network will be described. This training process combines actual data and simulated data to improve the performance of the GNN model in order to achieve the best optimization of the wind farm layout.

[0091] Specifically, the model training is divided into two stages: supervised learning and reinforcement learning. That is, first obtain the simulated data and real data for training; then determine the information on the layout rules of the wind farm indicated by the simulated data to conduct supervised training on the graph neural network based on the information on the layout rules of the wind farm; and determine the actual layout information of the wind farm indicated by the real data, and conduct reinforcement training on the graph neural network after supervised training based on the actual layout information of the wind farm.

[0092] Optionally, to train the GNN model, data needs to be prepared and preprocessed. The data is mainly divided into two categories: simulated data and real data. The simulated data can be generated by a wind resource assessment model, covering the power generation and interference information of wind turbines under different wind speeds, wind directions, and terrain conditions. The real data comes from the operation records of actual wind farms, including wind speed, wind direction, power generation, interference between wind turbines, etc. Data preprocessing includes data cleaning, standardization, and feature selection to ensure the quality and consistency of the input data.

[0093] It can be understood that for the process of supervised learning, that is, in the initial stage, preprocessed simulation data is used for supervised learning. The goal of supervised learning is to enable the GNN model to quickly learn the basic laws of the wind farm layout. During the training process, the data is divided into a training set and a validation set, and the model parameters are optimized through the backpropagation algorithm. The loss function combines two objectives: maximizing power generation and minimizing interference, guiding the model to gradually improve the layout embodiment during training. Cross-validation and early stopping strategies are adopted to prevent overfitting and ensure the generalization ability of the model.

[0094] For the process of reinforcement learning, that is, after the initial training of the model is completed, it enters the reinforcement learning stage. In this stage, the model is further fine-tuned mainly through layout optimization experiments in the simulation environment. A method based on policy gradient is designed. By executing different layout strategies in the simulation environment, the model parameters are gradually adjusted to optimize the layout effect. During this process, the reward signal is fed back according to the actual layout result to continuously improve the decision-making strategy of the model.

[0095] In this embodiment, a phased training process is carried out through the configuration of simulation data and real data. Among them, the role of real data is to reflect the actual situation. Real data can reflect the actual wind speed, wind direction, power generation, and interference conditions in the wind farm, ensuring that the model can adapt to the real scenario. Model verification can also be carried out, using real data to evaluate the model and verify its generalization ability and actual optimization effect under real conditions. It can also supplement the deficiencies of simulation data. Simulation data may have biases or simplified assumptions (such as ignoring some environmental factors), and real data can supplement these deficiencies to improve the accuracy of the model. And as the reward signal feedback, the reward signal is generated through the effect of the real layout (such as the increase in actual power generation or the reduction of interference), which is used to optimize the strategy in reinforcement learning.

[0096] Regarding the differences between the above two data training processes, the simulation data training process is used for supervised learning in the initial stage to help the model quickly learn the basic laws of the wind farm layout. Its simulation data volume is large and controllable, and it is easy to generate data under different conditions (such as different terrains, wind speeds, and wind directions). However, there may be deviations from the actual situation, and the model may not be able to generalize well to the real scenario only relying on simulation data.

[0097] For the process of real data training, it is used to fine-tune the model parameters subsequently to improve the adaptability of the model in the real environment. The real data volume is usually small during training, but it can better reflect the complex characteristics of the actual wind farm. And combined with reinforcement learning, the reward signal fed back by real data is used to dynamically adjust the layout strategy.

[0098] Therefore, in this embodiment, a comprehensive training strategy is adopted. Simulated data is used for pre-training: through a large amount of simulated data, supervised learning is completed to enable the model to quickly master the basic laws of wind farm layout. Real data is used for fine-tuning. Through real data, reinforcement learning is carried out to enhance the model's adaptability to real scenarios and avoid overfitting of the model to simulated data. To realize the process of joint training, real data and simulated data are combined and used, and a hybrid training strategy is designed to improve the robustness and practical application effect of the model.

[0099] Optionally, to further improve the performance of the model, a training strategy combining supervised learning and reinforcement learning is introduced. In the initial stage, a simulated data set is used for supervised learning to enable the model to quickly master the basic laws of wind farm layout. Then, through the method of reinforcement learning, layout optimization experiments are carried out in a simulated environment to continuously fine-tune the model parameters to cope with more diverse and complex wind field conditions. In addition, to prevent the model from falling into a local optimal solution, a variety of regularization strategies are also designed, such as randomly discarding some nodes (Dropout) to improve the generalization ability of the model, or dynamically adjusting the learning rate during training to stabilize the convergence process.

[0100] Specifically, to guide the training of the GNN model, a specific loss function is designed. This function comprehensively considers two main objectives of the wind farm layout: maximizing power generation and minimizing the mutual interference between wind turbines. For the first objective, it is measured by the sum of the expected power generation of each wind turbine, and the expected power generation can be estimated through features such as wind speed, wind direction, and wind turbine characteristics. The second objective is achieved by calculating the interference effect between wind turbines. This interference usually manifests as a reduction in power generation efficiency caused by the wake effect. These two objectives are combined into a weighted loss function, enabling the GNN model to balance the two during the training process to obtain the best layout embodiment. Therefore, for the process of supervised training and reinforcement training, the wind farm layout rule information and training nodes indicated by the simulated data can be determined; then, the expected power generation corresponding to the training nodes can be estimated through the environmental parameters and operating parameters corresponding to the training nodes; and the training interference parameters can be configured according to the wake effect between the training nodes indicated by the training nodes; a loss function is configured with the goal of maximizing the expected power generation and minimizing the training interference parameters; and then the loss function is calculated according to the wind farm layout rule information to perform supervised training and reinforcement training on the graph neural network.

[0101] Optionally, after training, the performance of the GNN model can also be evaluated and adaptively adjusted. That is, first, the test sets corresponding to the simulated data and real data are obtained; then, the performance of the graph neural network is evaluated based on the test sets according to the evaluation metrics to obtain evaluation information. The evaluation metrics include power generation, interference intensity, and layout efficiency; and guided by the evaluation information, the parameters of the graph neural network are adjusted according to the preset strategy.

[0102] Among them, the evaluation metrics include power generation, interference intensity, and layout efficiency. By evaluating the performance of the model on the validation set and the test set, it is ensured that the model is not only effective on the training data but also performs well in practical applications. Common evaluation methods include accuracy, mean squared error (MSE), and power generation improvement ratio, etc.

[0103] And the preset strategies include:

[0104] Hyperparameter tuning: Adjust the hyperparameters of the GNN model, such as learning rate, number of layers, number of nodes in each layer, etc., through grid search and random search methods to find the best model configuration.

[0105] Regularization and enhancement: To improve the generalization ability of the model, regularization techniques such as Dropout and L2 regularization are used to prevent the model from overfitting. In addition, through data augmentation techniques, such as random data transformation and noise injection, the adaptability of the model to different wind farm conditions is improved.

[0106] Ensemble learning: The method of ensemble learning is also considered, and the outputs of multiple GNN models are combined to improve the stability and accuracy of the overall prediction. By integrating the prediction results of different models, the bias and error that may be brought by a single model are reduced.

[0107] Furthermore, in the final stage of model optimization, the GNN model is tested for stability and practicality. Under different wind farm configurations and environmental conditions, the stability and practical application effects of the model are evaluated to ensure that it can perform stably and reliably in actual deployment. Through the above steps, the efficiency and reliability of the GNN model in wind farm layout optimization are ensured, providing a solid foundation for practical applications.

[0108] Combined with the above training process of the graph neural network, compared with traditional optimization methods, the GNN-based layout optimization model shows significant advantages in dealing with large-scale and complex wind farm layout problems. Traditional methods often rely on heuristic rules and preset parameters and are difficult to flexibly handle complex environmental variables. While GNN automatically captures and learns the complex non-linear relationships inside the wind farm through deep learning, significantly improving the accuracy and efficiency of layout optimization. Through the design of this step, the GNN model can not only achieve global optimization of the wind farm layout but also effectively cope with different wind field conditions and diverse layout requirements, providing a solid foundation for subsequent model training and optimization.

[0109] Through the construction of the graph structure, the design and training of the GNN model, result analysis and practical applications can be carried out. The goal of this step is to evaluate the actual performance of the GNN model in wind farm layout optimization and explore its application potential in the real environment.

[0110] Specifically, for result analysis, that is, a detailed analysis of the optimization results of the GNN model. This includes comparing the output of the model with the results of traditional layout optimization methods and evaluating its performance in different aspects. The specific evaluation dimensions include:

[0111] Optimization efficiency: By comparing the efficiency of the GNN model with traditional optimization algorithms (such as genetic algorithms, particle swarm optimization, etc.) in the layout optimization task, evaluate the advantages of the GNN in reducing computational time and improving the optimization speed. The GNN model can usually converge to a better solution in a shorter time. In contrast, traditional methods may require more iterations and computations.

[0112] Layout rationality: Analyze whether the layout embodiments generated by the GNN model are superior in terms of actual power generation and interference minimization. Specifically, compare the power generation increase ratio in the wind farm, the reduction of interference between wind turbines, and the rationality of the overall layout. The layout embodiments of the GNN model can usually maximize power generation while effectively reducing interference and improving the overall operation efficiency of the wind farm.

[0113] Power generation increase: Through the data of simulated and real wind farms, compare the changes in power generation before and after the optimization of the GNN model. The experimental results usually show that the GNN model can significantly improve the power generation capacity of the wind farm, because it can more accurately capture the complex mutual relationships and environmental factors between wind turbines.

[0114] In addition, after completing the result analysis, extend the application of the GNN model to the actual wind farm environment to explore its practical application potential. Case studies can also be conducted, that is, select several actual wind farms for case studies and apply the GNN model to the layout optimization of these wind farms. By comparing with the actual operation data, verify the performance of the GNN model in the real environment, including the actual increase in power generation and the stability of wind turbine operation. Case studies help to further understand the applicability and actual effects of the GNN model under different wind farm conditions. And conduct deployment and integration, study how to integrate the GNN model into the existing wind farm layout optimization system. Explore the technical details of model deployment, such as how to process the data input of large-scale wind farms and how to cooperate with existing optimization tools. Ensure that the model can be seamlessly integrated in practical applications and improve the overall efficiency of the system. User feedback and improvement can also be carried out, collect user feedback in practical applications, and understand the satisfaction and usage experience of wind farm operators with the optimization results of the GNN model. According to user feedback, further optimize the model and improve its functions and performance to better meet the actual application requirements.

[0115] In another possible scenario, for the process of wind farm layout configuration in practical applications, there may be a process of optimizing the layout and adding new wind turbines on the basis of an established wind farm. At this time, the configuration information corresponding to each wind turbine to be configured can be determined; then, based on the configuration information, the wind turbines to be configured are marked to update the wind farm graph structure data; and the updated wind farm graph structure data is input into the graph neural network to obtain the wind farm layout configuration information, and the equipment parameters of the wind turbines to be configured whose marked nodes in the wind farm layout configuration information are indicated as established nodes are adjusted, and the equipment parameters of the wind turbines to be configured whose marked nodes in the wind farm layout configuration information are indicated as unestablished nodes are adjusted, as well as the position configuration.

[0116] Specifically, for a real wind farm that has been built, the optimization usually does not involve physically "moving the location", but rather optimizes and adjusts the layout in the following two ways:

[0117] 1. Optimize operating parameters (main method):

[0118] Blade angle adjustment (pitch angle adjustment): By optimizing the pitch angle of the wind turbine, its operating efficiency at the current wind speed is maximized.

[0119] Operating mode selection: According to the real-time wind speed and direction, adjust the power output mode of the wind turbine to reduce the impact of the wake effect.

[0120] Wind turbine dynamic shutdown: Under specific wind conditions, some wind turbines may be temporarily shut down to reduce the wake interference to downstream wind turbines and optimize the overall power generation.

[0121] Real-time power scheduling: Through a dynamic power distribution strategy, optimize the overall power generation performance under different wind conditions and load demands.

[0122] 2. Propose local renovation suggestions (major adjustments):

[0123] Add auxiliary equipment: Add flow guiding equipment or wind fences at appropriate positions to optimize the wind speed distribution of the wind farm.

[0124] Locally fine-tune the wind turbine configuration: Adjust the direction (yaw angle) of individual wind turbines within a small range to optimize their efficiency in receiving wind energy and reduce the wake effect.

[0125] Technical upgrade: Replace existing wind turbines with new components (such as blades or control systems) to improve the single-unit performance.

[0126] 3. Data-driven strategy optimization:

[0127] Utilize the GNN model, combined with the data of the real wind farm, to continuously optimize the real-time operation strategy:

[0128] Wake prediction and interference optimization: Predict the wake effect through a model and adjust the operation mode and angle of the wind turbines in advance.

[0129] Adaptation to environmental changes: When the wind speed and direction change dynamically, optimize the operation parameters in real time to adapt to the new wind field conditions.

[0130] It should be noted that for the design without moving the positions of the wind turbines, it is because the infrastructure of the wind turbines in the actual wind farm has been fixed, and the relocation cost is extremely high and unrealistic. The ultimate goal is to improve the overall power generation efficiency and economic benefits through intelligent optimization of operation parameters and local strategy adjustments without large-scale physical transformation.

[0131] Combined with the above embodiments, by obtaining the working parameter information corresponding to multiple wind turbines to be configured and the interference information between each wind turbine; then, based on the working parameter information corresponding to the multiple wind turbines to be configured and the interference information between each wind turbine, generate wind farm graph structure data, where the nodes of the wind farm graph structure data are used to represent the working parameter information corresponding to each of the multiple wind turbines to be configured, and the edges between the nodes of the wind farm graph structure data are used to represent the interference information between each wind turbine; and input the wind farm graph structure data into a pre-trained graph neural network to obtain wind farm layout configuration information. Thus, a reliable wind farm layout configuration process is realized. Since each wind farm is designed as a node in the graph neural network to adapt to environmental characteristics, and the edges in the graph neural network are used to capture the complex relationships between the nodes to simulate the interference between wind farms, it can be applied to complex environmental conditions and large-scale wind farm layout configuration processes, improving the accuracy of wind farm layout configuration.

[0132] To better implement the above embodiments of the present application, the following also provides related devices for implementing the above embodiments. Please refer to Figure 5 , Figure 5 which is a schematic structural diagram of a wind farm layout configuration device provided by an embodiment of the present application. The configuration device 500 includes:

[0133] An acquisition unit 501, configured to acquire the working parameter information corresponding to multiple wind turbines to be configured and the interference information between each wind turbine;

[0134] A configuration unit 502, configured to generate wind farm graph structure data based on the working parameter information corresponding to the multiple wind turbines to be configured and the interference information between each wind turbine, where the nodes of the wind farm graph structure data are used to represent the working parameter information corresponding to each of the multiple wind turbines to be configured, and the edges between the nodes of the wind farm graph structure data are used to represent the interference information between each wind turbine;

[0135] The configuration unit 502 is further configured to input the wind farm graph structure data into a pre-trained graph neural network to obtain wind farm layout configuration information.

[0136] Optionally, in some possible implementation manners, when generating the wind farm graph structure data based on the working parameter information of the multiple wind turbines to be configured and the interference information between the wind turbines, the configuration unit 502 is configured to determine the environmental parameters and configuration parameters corresponding to each of the wind turbines to be configured based on the working parameter information of the multiple wind turbines to be configured; configure the node features through the environmental parameters and the configuration parameters; aggregate the interference information between the wind turbines according to the association relationship between the wind turbines to obtain edge features; and associate the node features and the edge features to generate the wind farm graph structure data.

[0137] Optionally, in some possible implementation manners, the configuration unit 502 is configured to obtain simulation data and real data for training; determine the wind farm layout rule information indicated by the simulation data to perform supervised training on the graph neural network based on the wind farm layout rule information; determine the actual layout information of the wind farm indicated by the real data, and perform reinforcement training on the graph neural network after supervised training based on the actual layout information of the wind farm.

[0138] Optionally, in some possible implementation manners, when determining the wind farm layout rule information indicated by the simulation data to perform supervised training on the graph neural network based on the wind farm layout rule information, the configuration unit 502 is configured to determine the wind farm layout rule information indicated by the simulation data and training nodes; estimate the expected power generation amount corresponding to the training nodes through the environmental parameters and operating parameters corresponding to the training nodes; configure training interference parameters through the wake effect between the training nodes indicated by the training nodes; configure a loss function with the goal of maximizing the expected power generation amount and minimizing the training interference parameters; and calculate the loss function according to the wind farm layout rule information to perform supervised training on the graph neural network.

[0139] Optionally, in some possible implementation manners, the configuration unit 502 is configured to obtain a test set corresponding to the simulation data and the real data; perform performance evaluation on the graph neural network according to evaluation metrics based on the test set to obtain evaluation information, where the evaluation metrics include power generation amount, interference intensity, and layout efficiency; and adjust the parameters of the graph neural network according to a preset strategy under the guidance of the evaluation information.

[0140] Optionally, in some possible implementation manners, the configuration unit 502 is configured to determine the configuration situation information corresponding to each of the to-be-configured wind turbines when inputting the wind farm graph structure data into a pre-trained graph neural network to obtain the wind farm layout configuration information; mark the to-be-configured wind turbines based on the configuration situation information to update the wind farm graph structure data; input the updated wind farm graph structure data into the graph neural network to obtain the wind farm layout configuration information, adjust the device parameters of the to-be-configured wind turbines whose marked nodes in the wind farm layout configuration information are indicated as established nodes, and adjust the device parameters and perform position configuration on the to-be-configured wind turbines whose marked nodes in the wind farm layout configuration information are indicated as unestablished nodes.

[0141] By obtaining the working parameter information corresponding to a plurality of to-be-configured wind turbines and the interference information between each pair of wind turbines; then generating wind farm graph structure data based on the working parameter information corresponding to the plurality of to-be-configured wind turbines and the interference information between each pair of wind turbines, where the nodes of the wind farm graph structure data are used to represent the working parameter information corresponding to each of the plurality of to-be-configured wind turbines, and the edges between the nodes of the wind farm graph structure data are used to represent the interference information between each pair of wind turbines; and inputting the wind farm graph structure data into a pre-trained graph neural network to obtain the wind farm layout configuration information. Thus, a reliable wind farm layout configuration process is realized. Since each wind farm is used as a node in the graph neural network for feature design adapted to the environment, and the edges in the graph neural network are used to capture the complex relationships between the nodes to simulate the interference between wind farms, it can be applied to complex environmental conditions and large-scale wind farm layout configuration processes, improving the accuracy of wind farm layout configuration.

[0142] The wind farm layout configuration device provided in this embodiment belongs to the same inventive concept as the method provided in the foregoing embodiments of the present application, can execute the wind farm layout configuration method provided in any foregoing embodiment of the present application, and has the corresponding functional modules and beneficial effects for executing the wind farm layout configuration method. For the technical details not described in detail in this embodiment, reference may be made to the specific processing content of the wind farm layout configuration method provided in the foregoing embodiments of the present application, which will not be elaborated here.

[0143] The functions implemented by the above-mentioned acquisition unit 501 and configuration unit 502 can be respectively implemented by the same or different processors, which is not limited in the embodiments of the present application.

[0144] It should be understood that each unit in the above device can be implemented in the form of a processor invoking software. For example, the device includes a processor, the processor is connected to a memory, instructions are stored in the memory, and the processor invokes the instructions stored in the memory to implement any of the above methods or the functions of each unit of the device. The processor can be a general-purpose processor, such as a CPU or a microprocessor, etc., and the memory can be a memory inside the device or a memory outside the device. Alternatively, the units in the device can be implemented in the form of hardware circuits, and the functions of some or all of the units can be implemented through the design of the hardware circuits. The hardware circuits can be understood as one or more processors. For example, in one implementation, the hardware circuit is an ASIC, and the functions of some or all of the above units are implemented through the design of the logical relationships of the components in the circuit. Again, for example, in another implementation, the hardware circuit can be implemented through a PLD. Taking an FPGA as an example, it can include a large number of logic gate circuits, and the connection relationships between the logic gate circuits are configured through a configuration file, so as to implement the functions of some or all of the above units. All units of the above device can be all implemented in the form of a processor invoking software, or all implemented in the form of hardware circuits, or some implemented in the form of a processor invoking software, and the remaining part implemented in the form of hardware circuits.

[0145] In the embodiments of the present application, a processor is a circuit with the ability to process signals. In one implementation, the processor can be a circuit with the ability to read and execute instructions, such as a CPU, a microprocessor, a GPU, or a DSP, etc. In another implementation, the processor can implement certain functions through the logical relationships of hardware circuits, and the logical relationships of the hardware circuits are fixed or can be reconstructed. For example, the processor is a hardware circuit implemented by an ASIC or a PLD, such as an FPGA, etc. In a reconfigurable hardware circuit, the process of the processor loading a configuration document to implement the configuration of the hardware circuit can be understood as the process of the processor loading instructions to implement the functions of some or all of the above units. In addition, it can also be a hardware circuit designed for artificial intelligence, which can be understood as a type of ASIC, such as an NPU, a TPU, a DPU, etc.

[0146] It can be seen that each unit in the above device can be one or more processors (or processing circuits) configured to implement the above method, such as: a CPU, a GPU, an NPU, a TPU, a DPU, a microprocessor, a DSP, an ASIC, an FPGA, or a combination of at least two of these processor forms.

[0147] In addition, each unit in the above device can be fully or partially integrated together, or can be independently implemented. In one implementation, these units are integrated together and implemented in the form of an SOC. The SOC can include at least one processor for implementing any of the above methods or implementing the functions of each unit of the device. The types of the at least one processor can be different, for example, including CPU and FPGA, CPU and artificial intelligence processor, CPU and GPU, etc.

[0148] An embodiment of the present application also proposes a wind farm layout configuration device; the wind farm layout configuration device includes a processor and an input / output component. The input / output component is used to obtain the working parameter information corresponding to a plurality of to-be-configured wind turbines and the interference information between each wind turbine;

[0149] The processor is used to perform layout configuration on the plurality of to-be-configured wind turbines determined by the input / output component by executing any one of the wind farm layout configuration methods in the above embodiments.

[0150] The above interface circuit can be any interface circuit capable of implementing data communication functions, for example, it can be a USB interface circuit, a Type-C interface circuit, a serial port circuit, a PCIE circuit, etc.

[0151] Optionally, an embodiment of the present application also provides a system. The interaction client is used to obtain the working parameter information corresponding to a plurality of to-be-configured wind turbines and the interference information between each wind turbine, send the working parameter information corresponding to the plurality of to-be-configured wind turbines and the interference information between each wind turbine to the server, and display the wind farm layout configuration result output by the server;

[0152] The server is used to perform layout configuration on the plurality of to-be-configured wind turbines determined by the interaction client by executing any one of the wind farm layout configuration methods in the above embodiments.

[0153] Another embodiment of the present application also proposes a wind farm layout configuration device, see Figure 6 as shown, the device includes:

[0154] a memory 600 and a processor 610;

[0155] Wherein, the memory 600 is connected to the processor 610 and is used to store programs;

[0156] The processor 610 is used to implement the wind farm layout configuration method disclosed in any of the above embodiments by running the programs stored in the memory 600.

[0157] Specifically, the above wind farm layout configuration device may further include: a bus, a communication interface 620, an input device 630, and an output device 640.

[0158] The processor 610, the memory 600, the communication interface 620, the input device 630, and the output device 640 are interconnected via a bus. Among them:

[0159] The bus may include a passage for transmitting information between various components of the computer system.

[0160] The processor 610 may be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or may be an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the embodiments of the present invention. It may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0161] The processor 610 may include a main processor, and may also include a baseband chip, a modem, etc.

[0162] The memory 600 stores the program for implementing the technical embodiments of the present invention, and may also store an operating system and other critical services. Specifically, the program may include program code, and the program code includes computer operation instructions. More specifically, the memory 600 may include a read-only memory (ROM), other types of static storage devices that can store static information and instructions, a random access memory (RAM), other types of dynamic storage devices that can store information and instructions, a disk memory, a flash memory, etc.

[0163] The input device 630 may include a device for receiving user input data and information, such as a keyboard, a mouse, a camera, a scanner, a light pen, a voice input device, a touch screen, a pedometer, or a gravity sensor, etc.

[0164] The output device 640 may include a device for allowing information to be output to the user, such as a display screen, a printer, a speaker, etc.

[0165] The communication interface 620 may include a device of any transceiver type for communicating with other devices or communication networks, such as Ethernet, a radio access network (RAN), a wireless local area network (WLAN), etc.

[0166] The processor 610 executes the program stored in the memory 600 and calls other devices, and can be used to implement each step of any one of the wind farm layout configuration methods provided in the above embodiments of the present application.

[0167] In addition to the above methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions. When the computer program instructions are run by a processor, the processor is caused to execute the steps in the wind farm layout configuration method according to various embodiments of the present application described in any of the above embodiments of this specification.

[0168] The computer program product may be written in any combination of one or more programming languages to write program code for performing the operations of the embodiments of the present application. The programming languages include object-oriented programming languages such as Java, C++, etc., and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, executed as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0169] In addition, an embodiment of the present application may also be a storage medium, on which a computer program is stored. The computer program is executed by a processor to perform the steps in the wind farm layout configuration method according to various embodiments of the present application described in any of the above embodiments of this specification. Specifically, the following steps may be implemented:

[0170] Obtain the working parameter information corresponding to a plurality of wind turbines to be configured and the interference information between each wind turbine;

[0171] Based on the working parameter information corresponding to a plurality of wind turbines to be configured and the interference information between each wind turbine, generate wind farm graph structure data. The nodes of the wind farm graph structure data are used to represent the working parameter information corresponding to each of the plurality of wind turbines to be configured, and the edges between the nodes of the wind farm graph structure data are used to represent the interference information between each wind turbine;

[0172] Input the wind farm graph structure data into a pre-trained graph neural network to obtain wind farm layout configuration information.

[0173] For the foregoing method embodiments, for simplicity of description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present application is not limited by the described action sequence, because according to the present application, certain steps may be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the present application.

[0174] It should be noted that the various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the similarities between the various embodiments, reference can be made to each other. For device embodiments, since they are basically similar to method embodiments, they are described relatively simply. For the relevant parts, reference can be made to the corresponding descriptions in the method embodiments.

[0175] The steps in the methods of the various embodiments of the present application can be adjusted, combined, and deleted according to actual needs. The technical features recorded in the various embodiments can be replaced or combined.

[0176] The modules and sub-modules in the devices and terminals in the various embodiments of the present application can be combined, divided, and deleted according to actual needs.

[0177] In the several embodiments provided by the present application, it should be understood that the disclosed terminals, devices, and methods can be implemented in other ways. For example, the terminal embodiments described above are merely illustrative. For example, the division of modules or sub-modules is only a logical function division. In actual implementation, there may be other division methods. For example, multiple sub-modules or modules can be combined or integrated into another module, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of devices or modules can be in electrical, mechanical, or other forms.

[0178] The modules or sub-modules described as separate components may or may not be physically separated. The components as modules or sub-modules may or may not be physical modules or sub-modules, that is, they can be located in one place, or they can be distributed to multiple network modules or sub-modules. Some or all of the modules or sub-modules can be selected according to actual needs to achieve the purpose of the embodiments of this embodiment.

[0179] In addition, the various functional modules or sub-modules in the various embodiments of the present application can be integrated in a processing module, or each module or sub-module can exist physically alone, or two or more modules or sub-modules can be integrated in one module. The above-mentioned integrated modules or sub-modules can be implemented in the form of hardware or in the form of software functional modules or sub-modules.

[0180] Those skilled in the art may further realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical embodiments. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0181] The steps of the methods or algorithms described in combination with the embodiments disclosed herein can be directly implemented by hardware, software units executed by a processor, or a combination of the two. The software units can be placed in a random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium well-known in the technical field.

[0182] Finally, it should also be noted that in this text, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including", or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article, or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article, or device. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article, or device including the said element.

[0183] The above description of the disclosed embodiments enables those skilled in the art to implement or use this application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application will not be limited to the embodiments shown herein, but rather will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for layout configuration of a wind farm, characterized in that, Including: Obtaining the working parameter information corresponding to multiple wind turbines to be configured and the interference information between each wind turbine; Generating wind farm graph structure data based on the working parameter information corresponding to the multiple wind turbines to be configured and the interference information between each wind turbine, where the nodes of the wind farm graph structure data are used to represent the working parameter information corresponding to each of the multiple wind turbines to be configured, and the edges between the nodes of the wind farm graph structure data are used to represent the interference information between each wind turbine; Inputting the wind farm graph structure data into a pre-trained graph neural network to obtain wind farm layout configuration information.

2. The method according to claim 1, characterized in that, The generating wind farm graph structure data based on the working parameter information corresponding to the multiple wind turbines to be configured and the interference information between each wind turbine includes: Determining the environmental parameters and configuration parameters corresponding to each of the wind turbines to be configured based on the working parameter information corresponding to the multiple wind turbines to be configured; Configuring node features through the environmental parameters and the configuration parameters; Aggregating the interference information between each wind turbine according to the association relationship between each wind turbine to obtain edge features; Associating the node features and the edge features to generate the wind farm graph structure data.

3. The method according to claim 1, wherein The training process of the graph neural network includes: Obtaining simulation data and real data for training; Determining the wind farm layout rule information indicated by the simulation data, and performing supervised training on the graph neural network based on the wind farm layout rule information; Determining the actual wind farm layout information indicated by the real data, and performing reinforcement training on the graph neural network after supervised training based on the actual wind farm layout information.

4. The method according to claim 3, wherein The determining the wind farm layout rule information indicated by the simulation data, and performing supervised training on the graph neural network based on the wind farm layout rule information includes: Determining the wind farm layout rule information indicated by the simulation data and training nodes; Estimating the expected power generation corresponding to the training nodes through the environmental parameters and operating parameters corresponding to the training nodes; Configuring training interference parameters through the wake effect between the training nodes indicated by the training nodes; Taking maximizing the expected power generation and minimizing the training interference parameters as the objective to configure a loss function; Calculating the loss function according to the wind farm layout rule information to perform supervised training on the graph neural network.

5. The method according to claim 3, characterized in that The method further includes: Obtaining a test set corresponding to the simulation data and the real data; Performing performance evaluation on the graph neural network according to evaluation metrics based on the test set to obtain evaluation information, where the evaluation metrics include power generation, interference intensity, and layout efficiency; Adjusting the parameters of the graph neural network according to a preset strategy under the guidance of the evaluation information.

6. The method according to claim 1, characterized in that, The inputting the wind farm graph structure data into a pre-trained graph neural network to obtain wind farm layout configuration information includes: Determining the configuration information corresponding to each of the wind turbines to be configured; Marking the wind turbines to be configured based on the configuration information to update the wind farm graph structure data. Input the updated wind farm graph structure data into the graph neural network to obtain the wind farm layout configuration information, adjust the device parameters of the to-be-configured wind turbines whose marked nodes in the wind farm layout configuration information are indicated as established nodes, adjust the device parameters of the to-be-configured wind turbines whose marked nodes in the wind farm layout configuration information are indicated as unestablished nodes, and perform position configuration.

7. A wind farm layout configuration device, characterized in that, It includes: An acquisition unit for acquiring the working parameter information corresponding to a plurality of to-be-configured wind turbines and the interference information between each pair of wind turbines; A configuration unit for generating wind farm graph structure data based on the working parameter information corresponding to the plurality of to-be-configured wind turbines and the interference information between each pair of wind turbines, where the nodes of the wind farm graph structure data are used to represent the working parameter information corresponding to each of the plurality of to-be-configured wind turbines, and the edges between the nodes of the wind farm graph structure data are used to represent the interference information between each pair of wind turbines; The configuration unit is further configured to input the wind farm graph structure data into a pre-trained graph neural network to obtain the wind farm layout configuration information.

8. A wind farm layout configuration device, comprising an input-output component and a processor; The input-output component is used to acquire the working parameter information corresponding to a plurality of to-be-configured wind turbines and the interference information between each pair of wind turbines; The processor is used to perform layout configuration on the plurality of to-be-configured wind turbines determined by the input-output component by executing the wind farm layout configuration method according to any one of claims 1 to 6.

9. A wind farm layout configuration system, comprising an interactive client and a server; The interactive client is used to acquire the working parameter information corresponding to a plurality of to-be-configured wind turbines and the interference information between each pair of wind turbines, send the working parameter information corresponding to the plurality of to-be-configured wind turbines and the interference information between each pair of wind turbines to the server, and display the wind farm layout configuration result output by the server; The server is used to perform layout configuration on the plurality of to-be-configured wind turbines determined by the interactive client by executing the wind farm layout configuration method according to any one of claims 1 to 6.

10. A computer program product, characterized in that, It includes: A computer program, which when executed by a processor, implements the wind farm layout configuration method according to any one of claims 1 to 6.

Images