A Method and System for Capacity Configuration of New Energy Equipment Based on Graph Neural Network

Abstract

The present invention relates to the technical field of smart grids, and discloses a method and system for configuring the capacity of new energy equipment based on a graph neural network. The method includes: constructing a corresponding directed weighted graph structure based on the topological data of a target new energy equipment group; updating the directed weighted graph structure based on the real-time operation data of the target new energy equipment group to obtain a corresponding time-varying heterogeneous graph structure; adjusting the parameters of a first multi-modal graph neural network model based on the time-varying heterogeneous graph structure to obtain a second multi-modal graph neural network model; performing multi-modal feature fusion on the time-varying heterogeneous graph structure based on the second multi-modal graph neural network model to obtain a corresponding set of embedding vectors; constructing a capacity planning model with the goal of matching the power generation of the target new energy equipment group with the power consumption demand based on the set of embedding vectors, and solving the capacity planning model to obtain a distributed capacity configuration strategy for the target new energy equipment group. The present invention can effectively improve the reliability of capacity planning.

Description

Technical Field

[0001] The present invention relates to the technical field of smart grids, and particularly to a method and system for configuring the capacity of new energy devices based on graph neural networks. Background Art

[0002] As the proportion of new energy power generation devices in the power grid continues to rise, the reasonable planning of their capacity is of great significance for ensuring the stable and efficient operation of the power grid. At present, the technology of using graph neural network models to plan the capacity of new energy power generation devices has been applied to a certain extent. The construction of its initial graph structure mostly depends on static data such as the geographical location, type, and power generation capacity of new energy power generation devices, as well as the connection information of power grid nodes. However, the power grid environment is not fixed. The real-time collection of dynamic information such as device operation data and meteorological data by sensors makes the attributes of nodes and edges in the graph constantly change frequently.

[0003] Existing technologies have some drawbacks when dealing with high-frequency dynamic changes. On the one hand, it is difficult to balance real-time performance and accuracy at the same time. When dealing with newly added or changed nodes and edges, information lag or error accumulation is likely to occur. For example, when the meteorological conditions suddenly change, causing a large fluctuation in the power generation capacity of a certain new energy power generation device, the model cannot quickly and accurately adjust the attributes of this node and re-evaluate its interaction effects on surrounding nodes. On the other hand, new energy power generation devices are diverse in type and highly heterogeneous in layout, and the interaction relationships between different types of devices are intricate, making it difficult to establish a unified model, which also exacerbates the difficulty of the model in dealing with dynamic changes, resulting in a deviation between the capacity planning result and the actual demand, and affecting the efficient utilization and stable operation of new energy power generation devices in the power grid. Summary of the Invention

[0004] In view of the above-mentioned drawbacks of the existing technology, the present invention provides a method and system for configuring the capacity of new energy devices based on graph neural networks.

[0005] In a first aspect, an embodiment of the present invention provides a method for configuring the capacity of new energy devices based on graph neural networks, including:

[0006] Constructing a corresponding directed weighted graph structure based on the topological data of the target new energy device group;

[0007] Updating the directed weighted graph structure based on the real-time operation data of the target new energy device group to obtain a corresponding time-varying heterogeneous graph structure;

[0008] Modeling the directed weighted graph structure to obtain a first multi-modal graph neural network model, and adjusting the parameters of the first multi-modal graph neural network model based on the time-varying heterogeneous graph structure to obtain a second multi-modal graph neural network model;

[0009] Perform multimodal feature fusion on the time-varying graph structure based on the second multimodal graph neural network model to obtain a corresponding set of embedding vectors;

[0010] Based on the set of embedding vectors, construct a capacity planning model with the goal of matching the power generation of the target new energy equipment group to the electricity demand, and solve the capacity planning model to obtain the distributed capacity configuration strategy of the target new energy equipment group.

[0011] Preferably, constructing the corresponding directed weighted graph structure based on the topological data of the target new energy equipment group includes:

[0012] Obtain the topological data of the target new energy equipment group, and determine the corresponding physical connection topology graph based on the topological data, where the topological data includes the connection relationship and energy flow direction between nodes in the target new energy equipment group;

[0013] Obtain the node static attribute data of the target new energy equipment group, and assign weights to the physical connection topology graph based on the node static attribute data to obtain the corresponding directed weighted graph structure, where the node static attribute data includes the geographical location and power generation performance data of each node in the target new energy equipment group.

[0014] Preferably, obtaining the node static attribute data of the target new energy equipment group, and assigning weights to the physical connection topology graph based on the node static attribute data to obtain the corresponding directed weighted graph structure includes:

[0015] Calculate the distance between nodes in the target new energy equipment group based on the geographical location of each node;

[0016] Calculate the energy transmission intensity between nodes in the target new energy equipment group based on the distance between nodes and the power generation performance data of each node;

[0017] Assign weights to the corresponding directed edges in the physical connection topology graph based on the energy transmission intensity between nodes to obtain a directed weighted graph structure.

[0018] Preferably, updating the directed weighted graph structure based on the real-time operation data of the target new energy equipment group to obtain the corresponding time-varying graph structure includes:

[0019] Obtain the real-time operation data of each node in the target new energy equipment group and the real-time meteorological data of the area where the target new energy equipment group is located;

[0020] Align the real-time operation data and the real-time meteorological data of each node in terms of time to obtain the dynamic attribute data of each node;

[0021] Perform real-time updates on the directed weighted graph structure based on the dynamic attribute data of each node to obtain a corresponding time-varying heterogeneous graph structure.

[0022] Preferably, modeling the directed weighted graph structure to obtain a first multi-modal graph neural network model, and adjusting the parameters of the first multi-modal graph neural network model based on the time-varying heterogeneous graph structure to obtain a second multi-modal graph neural network model, including:

[0023] Extract features from the directed weighted graph structure to obtain node features of each node and edge features of each edge. Among them, the node features include power generation performance data, and the edge features include energy transmission intensity;

[0024] Train a preset graph convolutional network based on the node features of each node and the edge features of each edge to obtain a first multi-modal graph neural network model;

[0025] Extract features from the time-varying heterogeneous graph structure to obtain real-time node features of each node and real-time edge features of each edge. Among them, the real-time node features include real-time operation data and real-time meteorological data, and the real-time edge features include real-time energy transmission intensity;

[0026] Adjust the parameters of the first multi-modal graph neural network model based on the real-time node features of each node and the real-time edge features of each edge to obtain a second multi-modal graph neural network model.

[0027] Preferably, adjusting the parameters of the first multi-modal graph neural network model based on the real-time node features of each node and the real-time edge features of each edge to obtain a second multi-modal graph neural network model, including:

[0028] Adjust the parameters of the first multi-modal graph neural network model based on the real-time node features of each node and the real-time edge features of each edge by using an incremental learning mechanism to obtain a second multi-modal graph neural network model.

[0029] Preferably, performing multi-modal feature fusion on the time-varying heterogeneous graph structure based on the second multi-modal graph neural network model to obtain a corresponding set of embedding vectors, including:

[0030] Extract data from the time-varying heterogeneous graph structure to obtain multi-modal data of each node. Among them, the multi-modal data includes real-time operation data and real-time meteorological data;

[0031] Input the real-time operation data and the real-time meteorological data of each node into the second multi-modal graph neural network model for modal alignment and semantic mapping to obtain a corresponding set of embedding vectors.

[0032] Preferably, based on the embedding vector set, a capacity planning model is constructed with the goal of matching the power generation of the target new energy equipment group with the electricity demand, and the distributed capacity allocation strategy of the target new energy equipment group is obtained by solving the capacity planning model, including:

[0033] Extract data from the embedding vector set to obtain the power generation efficiency of each node in the target new energy equipment group;

[0034] Based on the power generation efficiency of each node, a target function is constructed with the goal of minimizing the deviation between the power generation of the target new energy equipment group and the electricity demand, and the target function is characterized as the capacity planning model;

[0035] Based on the preset constraint conditions, a solver is used to solve the capacity planning model to obtain the capacity allocation strategy of each node in the target new energy equipment group, where the preset constraint conditions include a capacity non-negativity constraint condition and a maximum capacity constraint condition.

[0036] Preferably, the capacity planning model is characterized by the following formula:

[0037]

[0038] Wherein, represents the power generation efficiency of the i-th node, represents the planned capacity of the i-th node, represents the electricity demand, represents the number of nodes.

[0039] In a second aspect, an embodiment of the present invention provides a new energy equipment capacity allocation system based on a graph neural network, including:

[0040] A static structure determination module for constructing a corresponding directed weighted graph structure based on the topological data of the target new energy equipment group;

[0041] A dynamic structure determination module for updating the directed weighted graph structure based on the real-time operation data of the target new energy equipment group to obtain a corresponding time-varying heterogeneous graph structure;

[0042] A parameter adjustment module for modeling the directed weighted graph structure to obtain a first multi-modal graph neural network model, and adjusting the parameters of the first multi-modal graph neural network model based on the time-varying heterogeneous graph structure to obtain a second multi-modal graph neural network model;

[0043] A multi-modal fusion module for performing multi-modal feature fusion on the time-varying heterogeneous graph structure based on the second multi-modal graph neural network model to obtain a corresponding embedding vector set;

[0044] A capacity configuration strategy determination module, configured to construct a capacity planning model based on the embedded vector set with the goal of matching the power generation of the target new energy equipment group with the power consumption demand, and solve the capacity planning model to obtain the distributed capacity configuration strategy of the target new energy equipment group.

[0045] Compared with the prior art, the new energy equipment capacity configuration method and system based on a graph neural network in an embodiment of the present invention have the following beneficial effects: constructing a directed weighted graph structure according to the topological data of the target new energy equipment group, laying a structured foundation for subsequent analysis, and being able to intuitively present the connection and weight relationship between devices; updating the directed weighted graph structure with real-time operation data to generate a time-varying graph structure, which can effectively capture the dynamic changes of the operation state of the equipment group over time; adjusting parameters based on the time-varying graph structure to obtain a second multi-modal graph neural network model, improving the model's fitting and prediction capabilities for dynamic data, and the embedded vector set obtained by fusing multi-modal features of the time-varying graph structure by the model, integrating the key information of the equipment group, providing strong data support for capacity planning; constructing a capacity planning model with the goal of matching power generation with power consumption demand, and obtaining a distributed capacity configuration strategy through solution, which can effectively improve the matching degree between the power generation of the new energy equipment group and the power consumption demand, thereby enhancing the stability and reliability of the entire energy system. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 is a schematic flowchart of a new energy equipment capacity configuration method based on a graph neural network in an embodiment of the present invention;

[0047] Figure 2 is a schematic structural diagram of a new energy equipment capacity configuration system based on a graph neural network in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0049] In the description of the present invention, the terms "first", "second", "third", etc. are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second", "third", etc. may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, the meaning of "a plurality" is two or more.

[0050] In the description of the present invention, it should be noted that, unless otherwise clearly defined and limited, the terms "mounted", "connected", "coupled" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection, an electrical connection; it may be directly connected, or indirectly connected through an intermediate medium, and it may be the communication inside two elements. The terms "vertical", "horizontal", "left", "right", "up", "down" and similar expressions used herein are for illustrative purposes only and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation to the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0051] In the description of the present invention, it should be noted that, unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as commonly understood by those skilled in the technical field to which this technology belongs. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0052] As Figure 1 shown, an embodiment of the present invention provides a new energy device capacity configuration method based on a graph neural network, including the steps of:

[0053] S1. Construct a corresponding directed weighted graph structure based on the topological data of the target new energy device group;

[0054] Specifically, step S1 includes:

[0055] 1) Obtain the topological data of the target new energy device group and determine the corresponding physical connection topology graph based on the topological data;

[0056] Specifically, the topological data includes the connection relationships and energy flow directions between nodes in the target new energy device group. The topological data can be obtained from the topological graph, connection documents, or monitoring system of the target new energy device group. It should be noted that each device in the target new energy device group can be regarded as a node, and the "nodes" described in the present invention can all be understood as new energy devices. Further, based on the topological data, the physical connection topology graph corresponding to the target new energy device group is determined.

[0057] 2) Obtain the node static attribute data of the target new energy device group, and assign weights to the physical connection topology graph based on the node static attribute data to obtain the corresponding directed weighted graph structure.

[0058] Specifically, the node static attribute data includes the geographical location and power generation performance data of each node in the target new energy device group.

[0059] Further, step 2) includes:

[0060] i) Calculate the distances between nodes in the target new energy device group based on the geographical location of each node;

[0061] ii) Calculate the energy transfer intensity between nodes in the target new energy device group based on the distances between nodes and the power generation performance data of each node;

[0062] The energy transfer intensity is related to the power generation performance data and is inversely proportional to the distance between nodes. Specifically, the energy transfer intensity is directly proportional to the power generation power of the node and inversely proportional to the square of the distance between nodes. Based on this correspondence, the energy transfer intensity between nodes in the target new energy device group can be determined.

[0063] iii) Assign weights to the corresponding directed edges in the physical connection topology graph based on the energy transfer intensity between nodes to obtain the directed weighted graph structure.

[0064] Characterize the energy transfer intensity between nodes as the weight of the corresponding directed edge. If the energy transfer intensity between two nodes is large, the weight of the edge is large; conversely, if the energy transfer intensity between two nodes is small, the weight of the edge is small.

[0065] S2. Update the directed weighted graph structure based on the real-time operation data of the target new energy device group to obtain the corresponding time-varying heterogeneous graph structure;

[0066] Specifically, step S2 includes:

[0067] 1) Obtain the real-time operation data of each node in the target new energy device group and the real-time meteorological data of the region where the target new energy device group is located;

[0068] The real-time operation data includes the output power and operating speed of the nodes, and the real-time meteorological data includes the light intensity, wind speed, temperature, and humidity.

[0069] 2) Align the real-time operation data and real-time meteorological data of each node in time to obtain the dynamic attribute data of each node;

[0070] It can be understood that the dynamic attribute data of each node includes the operation data and meteorological data at different time points.

[0071] 3) Based on the dynamic attribute data of each node, perform real-time updates on the directed weighted graph structure to obtain the corresponding time-varying heterogeneous graph structure.

[0072] Based on the dynamic attribute data of each node, update each node and each edge in the directed weighted graph respectively to obtain the corresponding time-varying heterogeneous graph structure. It can be understood that when the attributes of the nodes change, the energy transfer intensity between the nodes, that is, the weight of the edges, will also change accordingly.

[0073] S3. Model the directed weighted graph structure to obtain the first multi-modal graph neural network model, and adjust the parameters of the first multi-modal graph neural network model based on the time-varying heterogeneous graph structure to obtain the second multi-modal graph neural network model;

[0074] Specifically, step S3 includes:

[0075] 1) Extract features from the directed weighted graph structure to obtain the node features of each node and the edge features of each edge;

[0076] The node features include power generation performance data, and the edge features include energy transfer intensity.

[0077] 2) Based on the node features of each node and the edge features of each edge, train the preset graph convolutional network to obtain the first multi-modal graph neural network model;

[0078] Specifically, the training process includes:

[0079] i) Data partitioning

[0080] Partition the node features of each node and the edge features of each edge extracted into a training set, a validation set, and a test set. Specifically, in this embodiment, the partitioning is performed according to the ratio of 70%, 15%, and 15%.

[0081] ii) Select the graph neural network

[0082] The preset graph convolutional network updates the feature representation of the nodes by aggregating the adjacent information of the nodes, and is more suitable for processing the directed weighted graph structure.

[0083] iii) Model training

[0084] The preset graph convolutional network is iteratively trained multiple times on the training set. In each iteration, the value of the loss function is calculated, and the optimizer is used to update the parameters of the preset graph convolutional network. At the same time, validation is performed on the validation set to prevent overfitting. Finally, the performance of the preset graph convolutional network is evaluated on the test set to obtain the first multi-modal graph neural network.

[0085] 3) Feature extraction is performed on the time-varying heterogeneous graph structure to obtain the real-time node features of each node and the real-time edge features of each edge;

[0086] The real-time node features include real-time operation data and real-time meteorological data, and the real-time edge features include real-time energy transmission intensity.

[0087] 4) Based on the real-time node features of each node and the real-time edge features of each edge, the parameters of the first multi-modal graph neural network model are adjusted to obtain the second multi-modal graph neural network model.

[0088] Specifically, based on the real-time node features of each node and the real-time edge features of each edge, an incremental learning mechanism is used to adjust the parameters of the first multi-modal graph neural network model to obtain the second multi-modal graph neural network model. It should be noted that the adjusted parameters include the connection weights between the network layers in the first multi-modal graph neural network model and the bias parameters in each network layer. The incremental learning mechanism gradually integrates the real-time node features of each node and the real-time edge features of each edge into the model, avoiding the computational burden brought by one-time retraining. This approach improves the adaptability of the model to the dynamic environment and reduces resource waste.

[0089] S4. Based on the second multi-modal graph neural network model, multi-modal feature fusion is performed on the time-varying heterogeneous graph structure to obtain the corresponding embedding vector set;

[0090] Specifically, step S4 includes:

[0091] 1) Data extraction is performed on the time-varying heterogeneous graph structure to obtain the multi-modal data of each node;

[0092] The multi-modal data includes real-time operation data and real-time meteorological data.

[0093] 2) The real-time operation data and real-time meteorological data of each node are input into the second multi-modal graph neural network model for modal alignment and semantic mapping to obtain the corresponding embedding vector set.

[0094] The second multi-modal graph neural network model is used to perform cross-fusion on the multi-modal data of each node. By performing modal alignment and semantic mapping on the multi-modal data of each node, the node features of different modalities are mapped to a unified semantic embedding space, that is, the embedding vector set, so as to effectively capture the interaction patterns and cooperation laws of heterogeneous new energy devices at different scales.

[0095] S5. Based on the embedded vector set, construct a capacity planning model with the goal of matching the power generation of the target new energy equipment group to the electricity demand, and solve the capacity planning model to obtain the distributed capacity configuration strategy of the target new energy equipment group.

[0096] Specifically, step S5 includes:

[0097] 1) Extract data from the embedded vector set to obtain the power generation efficiency of each node in the target new energy equipment group;

[0098] The embedded vector set includes the multimodal feature information of the target new energy equipment group. By extracting the power generation efficiency among them, it can be used to assist in capacity planning.

[0099] 2) Based on the power generation efficiency of each node, construct an objective function with the goal of minimizing the deviation between the power generation of the target new energy equipment group and the electricity demand, and represent the objective function as a capacity planning model;

[0100] Specifically, the model construction process includes:

[0101] i) Define variables

[0102] The decision variable is the planned capacity of each node in the target new energy equipment group, and the parameters include the power generation efficiency, electricity demand, and maximum allowable construction capacity of each node.

[0103] ii) Construct the objective function

[0104] The goal is to minimize the deviation between the power generation of the target new energy equipment group and the electricity demand. Specifically, the capacity planning model is represented by the following formula:

[0105]

[0106] Among them, represents the power generation efficiency of the i-th node, represents the planned capacity of the i-th node, represents the electricity demand, represents the number of nodes.

[0107] iii) Determine the constraint conditions

[0108] The preset constraint conditions include the non-negativity constraint condition of capacity and the maximum capacity constraint condition. Specifically, the non-negativity constraint condition of capacity is that the planned capacity of each node cannot be negative, and the maximum capacity constraint condition is that the planned capacity of each node cannot exceed its maximum allowable construction capacity.

[0109] 3) Solve the capacity planning model using a solver based on preset constraint conditions to obtain the capacity configuration strategy for each node in the target new energy equipment group.

[0110] Since solving the above capacity planning model is a linear programming problem, in this embodiment, the Gurobi solver is used to solve the capacity planning model to obtain the capacity configuration strategy for each node in the target new energy equipment group.

[0111] In an embodiment of the present invention, a method for configuring the capacity of new energy equipment based on a graph neural network constructs a directed weighted graph structure according to the topological data of the target new energy equipment group, laying a structured foundation for subsequent analysis and being able to intuitively present the connection and weight relationship between devices; the real-time operation data updates the directed weighted graph structure to generate a time-varying graph structure, which can effectively capture the dynamic changes of the operation state of the equipment group over time; adjusting the parameters based on the time-varying graph structure to obtain a second multi-modal graph neural network model improves the model's fitting and prediction ability for dynamic data, and the embedding vector set obtained by performing multi-modal feature fusion on the time-varying graph structure by the model integrates the key information of the equipment group and provides strong data support for capacity planning; constructing a capacity planning model with the goal of matching power generation with power consumption demand and obtaining a distributed capacity configuration strategy through solution can effectively improve the matching degree of power generation and power consumption demand of the new energy equipment group, thereby enhancing the stability and reliability of the entire energy system.

[0112] Based on the above method for configuring the capacity of new energy equipment based on a graph neural network, as Figure 2 shown, an embodiment of the present invention provides a system for configuring the capacity of new energy equipment based on a graph neural network, including:

[0113] A static structure determination module 1, configured to construct a corresponding directed weighted graph structure based on the topological data of the target new energy equipment group;

[0114] A dynamic structure determination module 2, configured to update the directed weighted graph structure based on the real-time operation data of the target new energy equipment group to obtain a corresponding time-varying graph structure;

[0115] A parameter adjustment module 3, configured to model the directed weighted graph structure to obtain a first multi-modal graph neural network model, and adjust the parameters of the first multi-modal graph neural network model based on the time-varying graph structure to obtain a second multi-modal graph neural network model;

[0116] A multi-modal fusion module 4, configured to perform multi-modal feature fusion on the time-varying graph structure based on the second multi-modal graph neural network model to obtain a corresponding embedding vector set;

[0117] A capacity configuration strategy determination module 5, which is used to construct a capacity planning model based on the embedded vector set with the goal of matching the power generation of the target new energy device group with the power consumption demand, and solve the capacity planning model to obtain the distributed capacity configuration strategy of the target new energy device group.

[0118] It should be noted that each module in the above new energy device capacity configuration system based on a graph neural network can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor of the computer device in hardware form or be independent of it, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each of the above modules. For the specific limitations of a new energy device capacity configuration system based on a graph neural network, refer to the limitations of a new energy device capacity configuration method based on a graph neural network in the above text. The two have the same functions and effects, and will not be elaborated here.

[0119] In summary, for the new energy device capacity configuration method and system based on a graph neural network in the embodiments of the present invention, a directed weighted graph structure is constructed according to the topological data of the target new energy device group, laying a structured foundation for subsequent analysis and being able to intuitively present the connection and weight relationship between devices; the real-time operation data updates the directed weighted graph structure to generate a time-varying graph structure, which can effectively capture the dynamic changes of the device group operation state over time; adjusting parameters based on the time-varying graph structure to obtain a second multi-modal graph neural network model improves the model's fitting and prediction ability for dynamic data, and the embedded vector set obtained by the model through multi-modal feature fusion of the time-varying graph structure integrates the key information of the device group, providing strong data support for capacity planning; constructing a capacity planning model with the goal of matching power generation with power consumption demand and obtaining a distributed capacity configuration strategy through solution can effectively improve the matching degree between the power generation of the new energy device group and the power consumption demand, thereby enhancing the stability and reliability of the entire energy system.

[0120] Each embodiment in this specification is described in a progressive manner. For parts that are the same or similar in each embodiment, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and for related parts, reference can be made to the partial description of the method embodiment. It should be noted that the above technical features of the embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the above technical features in the embodiments are described. However, as long as the combinations of these technical features do not conflict, they should be considered as the scope described in this specification.

[0121] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and substitutions can be made, and these improvements and substitutions should also be regarded as the protection scope of the present invention.

Claims

1. A new energy equipment capacity configuration method based on graph neural network, characterized in that: include: Construct a corresponding directed weighted graph structure based on the topological data of the target new energy equipment group; The directed weighted graph structure is updated based on the real-time operation data of the target new energy equipment group to obtain a corresponding time-varying composition graph structure; Modeling the directed weighted graph structure to obtain a first multimodal graph neural network model, and adjusting parameters of the first multimodal graph neural network model based on the time-varying graph structure to obtain a second multimodal graph neural network model; Based on the second multimodal graph neural network model, multimodal feature fusion is performed on the time-varying graph structure to obtain a corresponding embedding vector set; Based on the embedded vector set, a capacity planning model is constructed with the goal of matching the power generation of the target new energy equipment group with the power demand, and the capacity planning model is solved to obtain a distributed capacity configuration strategy for the target new energy equipment group.

2. The new energy equipment capacity configuration method according to claim 1, characterized in that: The corresponding directed weighted graph structure is constructed based on the topological data of the target new energy equipment group, including: Acquire topological data of the target new energy device group, and determine a corresponding physical connection topological graph based on the topological data, wherein the topological data includes connection relationships and energy flow directions between nodes in the target new energy device group; Obtain node static attribute data of the target new energy device group, and weight the physical connection topology graph based on the node static attribute data to obtain a corresponding directed weighted graph structure, wherein the node static attribute data includes the geographical location and power generation performance data of each node in the target new energy device group.

3. The new energy equipment capacity configuration method according to claim 2, characterized in that: The acquiring of the node static attribute data of the target new energy equipment group, and weighting the physical connection topology graph based on the node static attribute data to obtain a corresponding directed weighted graph structure, includes: Calculating the distance between nodes in the target new energy equipment group based on the geographical location of each node; Calculating the energy transmission intensity between nodes in the target new energy equipment group based on the distance between the nodes and the power generation performance data of each node; The corresponding directed edges in the physical connection topology graph are weighted based on the energy transmission intensity between the nodes to obtain a directed weighted graph structure.

4. The new energy equipment capacity configuration method according to claim 1, characterized in that: The updating of the directed weighted graph structure based on the real-time operation data of the target new energy equipment group to obtain a corresponding time-varying graph structure includes: Acquire real-time operation data of each node in the target new energy equipment group and real-time meteorological data of the area where the target new energy equipment group is located; Time-aligning the real-time operation data and the real-time meteorological data of each node to obtain dynamic attribute data of each node; The directed weighted graph structure is updated in real time based on the dynamic attribute data of each node to obtain a corresponding time-varying graph structure.

5. The new energy equipment capacity configuration method according to claim 1, characterized in that: The step of modeling the directed weighted graph structure to obtain a first multimodal graph neural network model, and adjusting parameters of the first multimodal graph neural network model based on the time-varying graph structure to obtain a second multimodal graph neural network model includes: Performing feature extraction on the directed weighted graph structure to obtain node features of each node and edge features of each edge, wherein the node features include power generation performance data, and the edge features include energy transmission intensity; Training a preset graph convolutional network based on the node features of each node and the edge features of each edge to obtain a first multimodal graph neural network model; Extracting features from the time-varying graph structure to obtain real-time node features of each node and real-time edge features of each edge, wherein the real-time node features include real-time operation data and real-time meteorological data, and the real-time edge features include real-time energy transmission intensity; Based on the real-time node features of each node and the real-time edge features of each edge, the parameters of the first multimodal graph neural network model are adjusted to obtain a second multimodal graph neural network model.

6. The new energy equipment capacity configuration method according to claim 5, characterized in that: The step of adjusting parameters of the first multimodal graph neural network model based on the real-time node feature of each node and the real-time edge feature of each edge to obtain a second multimodal graph neural network model includes: Based on the real-time node features of each node and the real-time edge features of each edge, an incremental learning mechanism is used to adjust the parameters of the first multimodal graph neural network model to obtain a second multimodal graph neural network model.

7. The new energy equipment capacity configuration method according to claim 1, characterized in that: The step of performing multimodal feature fusion on the time-varying graph structure based on the second multimodal graph neural network model to obtain a corresponding embedding vector set includes: Extracting data from the time-varying composition graph structure to obtain multimodal data of each node, wherein the multimodal data includes real-time operation data and real-time meteorological data; The real-time operation data and the real-time meteorological data of each node are input into the second multimodal graph neural network model for modality alignment and semantic mapping to obtain a corresponding embedding vector set.

8. The new energy equipment capacity configuration method according to claim 1, characterized in that: The method of constructing a capacity planning model based on the embedded vector set with the power generation of the target new energy equipment group matching the power demand as the goal, and solving the capacity planning model to obtain a distributed capacity configuration strategy for the target new energy equipment group includes: Performing data extraction on the embedded vector set to obtain the power generation efficiency of each node in the target new energy equipment group; Based on the power generation efficiency of each node, construct an objective function with the goal of minimizing the deviation between the power generation and the power demand of the target new energy equipment group, and characterize the objective function as the capacity planning model; The capacity planning model is solved by a solver based on preset constraints to obtain a capacity configuration strategy for each node in the target new energy equipment group, wherein the preset constraints include a non-negative capacity constraint and a maximum capacity constraint.

9. The new energy equipment capacity configuration method according to claim 8, characterized in that: The capacity planning model is characterized by the following formula: in, represents the power generation efficiency of the i-th node, represents the planned capacity of the i-th node, Indicates the electricity demand, Indicates the number of nodes.

10. A new energy equipment capacity configuration system based on graph neural network, characterized in that: include: A static structure determination module, used to construct a corresponding directed weighted graph structure based on the topological data of the target new energy equipment group; A dynamic structure determination module, used to update the directed weighted graph structure based on the real-time operation data of the target new energy equipment group to obtain a corresponding time-varying composition graph structure; A parameter adjustment module, used to model the directed weighted graph structure to obtain a first multimodal graph neural network model, and adjust the parameters of the first multimodal graph neural network model based on the time-varying graph structure to obtain a second multimodal graph neural network model; A multimodal fusion module, used for performing multimodal feature fusion on the time-varying graph structure based on the second multimodal graph neural network model to obtain a corresponding embedding vector set; The capacity configuration strategy determination module is used to construct a capacity planning model based on the embedded vector set with the goal of matching the power generation of the target new energy equipment group with the power demand, and solve the capacity planning model to obtain the distributed capacity configuration strategy of the target new energy equipment group.

Images