Transient voltage safety assessment method and device based on graph neural network

Abstract

The invention relates to a transient voltage safety assessment method and device based on a graph neural network. The method comprises the steps that modeling is carried out on a power system network topology to obtain a power system diagram structure, and the power system diagram structure comprises a node set and a line set in a power system network; constructing an initial node feature vector corresponding to the node set and a first line feature vector corresponding to the line set; performing projection conversion on different feature vectors of heterogeneous node types in the initial node feature vectors according to the feature transformation matrix of the target type nodes to obtain isomorphic node feature vectors; and determining a transient voltage safety evaluation result based on the first line feature vector, the isomorphic node feature vector and a pre-constructed transient voltage safety evaluation model. An electric power system is abstracted into a graph structure model, isomorphic input data of an adaptive graph neural network is constructed, and efficient mapping of transient voltage safety assessment is achieved.

Description

Technical Field

[0001] This application relates to the field of power system dynamic analysis technology, and in particular to a transient voltage safety assessment method and apparatus based on graph neural networks. Background Technology

[0002] The increasing penetration of new energy sources in power systems has led to more significant volatility and uncertainty in transmission and distribution networks, making online safety and stability analysis and control of power grids increasingly important. Unlike model-driven analysis methods such as numerical simulation, transient energy functions, and bifurcation theory, data-driven machine learning offers the possibility of online transient safety analysis and control for complex, large-scale power grids.

[0003] However, how to use graph neural networks to perform transient voltage security assessments of power systems remains a challenge to be addressed. Summary of the Invention

[0004] Therefore, it is necessary to provide a graph neural network-based transient voltage safety assessment method and device that can establish a mapping relationship between heterogeneous data of the power grid and graph neural networks to address the above-mentioned technical problems.

[0005] In a first aspect, this application provides a transient voltage safety assessment method based on graph neural networks, the method comprising:

[0006] A power system network topology is modeled to obtain a power system graph structure, which includes: a set of nodes and a set of lines in the power system network;

[0007] Construct the initial node feature vector corresponding to the node set, and the first line feature vector corresponding to the line set;

[0008] Based on the feature transformation matrix of the target type node, the different feature vectors of heterogeneous node types in the initial node feature vector are projected and transformed to obtain homogenized node feature vectors.

[0009] Based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model, the transient voltage safety assessment result is determined. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the isomorphic sample feature vector of the sample power system. The first sample line feature vector and the isomorphic sample feature vector are obtained based on the graph structure of the sample power system.

[0010] In one embodiment, determining the transient voltage safety assessment result based on the first line feature vector, the isomorphic node feature vector, and a pre-constructed transient voltage safety assessment model includes:

[0011] Based on the first line feature vector, fault indication information of each line in the line set is added to obtain the second line feature vector;

[0012] Based on the second line feature vector, the isomorphic node feature vector, and the transient voltage safety assessment model, the transient voltage safety assessment result is determined.

[0013] In one embodiment, the first line feature vector includes the equivalent resistance, equivalent reactance, active power, and reactive power of each line; the step of adding fault indication information of each line in the line set to the first line feature vector to obtain the second line feature vector includes:

[0014] Based on the equivalent resistance, equivalent reactance, active power, and reactive power of each line, fault indication information of each line in the line set is added to obtain a second line feature vector.

[0015] In one embodiment, constructing the initial node feature vector corresponding to the node set includes:

[0016] Determine the node type of the node set;

[0017] If the node type is a new energy type, an initial node feature vector corresponding to the node set is constructed based on the active power, reactive power, and transient performance of the generator output.

[0018] If the node type is a non-new energy type, an initial node feature vector corresponding to the node set is constructed based on the active power and reactive power output by the generator.

[0019] In one embodiment, the method further comprises:

[0020] The first line feature vector and the initial node feature vector are normalized.

[0021] In one embodiment, the topological connection relationship in the graph structure is an n×n adjacency matrix. When there is a connection between any two nodes in the node set, the element corresponding to any two nodes in the adjacency matrix is ​​1; where n is the number of nodes in the node set.

[0022] Secondly, this application also provides a transient voltage safety assessment device based on graph neural networks, comprising:

[0023] The acquisition module is used to model the topology of the power system network to obtain the power system graph structure, which includes: a set of nodes and a set of lines in the power system network;

[0024] A construction module is used to construct the initial node feature vector corresponding to the node set and the first line feature vector corresponding to the line set.

[0025] The transformation module is used to project and transform the different feature vectors of heterogeneous node types in the initial node feature vector according to the feature transformation matrix of the target type node, so as to obtain homogenized node feature vectors.

[0026] The determination module is used to determine the transient voltage safety assessment result based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the isomorphic sample feature vector of the sample power system. The first sample line feature vector and the isomorphic sample feature vector are obtained based on the graph structure of the sample power system.

[0027] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0028] A power system network topology is modeled to obtain a power system graph structure, which includes: a set of nodes and a set of lines in the power system network;

[0029] Construct the initial node feature vector corresponding to the node set, and the first line feature vector corresponding to the line set;

[0030] Based on the feature transformation matrix of the target type node, the different feature vectors of heterogeneous node types in the initial node feature vector are projected and transformed to obtain homogenized node feature vectors.

[0031] Based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model, the transient voltage safety assessment result is determined. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the isomorphic sample feature vector of the sample power system. The first sample line feature vector and the isomorphic sample feature vector are obtained based on the graph structure of the sample power system.

[0032] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0033] A power system network topology is modeled to obtain a power system graph structure, which includes: a set of nodes and a set of lines in the power system network;

[0034] Construct the initial node feature vector corresponding to the node set, and the first line feature vector corresponding to the line set;

[0035] Based on the feature transformation matrix of the target type node, the different feature vectors of heterogeneous node types in the initial node feature vector are projected and transformed to obtain homogenized node feature vectors.

[0036] Based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model, the transient voltage safety assessment result is determined. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the isomorphic sample feature vector of the sample power system. The first sample line feature vector and the isomorphic sample feature vector are obtained based on the graph structure of the sample power system.

[0037] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0038] A power system network topology is modeled to obtain a power system graph structure, which includes: a set of nodes and a set of lines in the power system network;

[0039] Construct the initial node feature vector corresponding to the node set, and the first line feature vector corresponding to the line set;

[0040] Based on the feature transformation matrix of the target type node, the different feature vectors of heterogeneous node types in the initial node feature vector are projected and transformed to obtain homogenized node feature vectors.

[0041] Based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model, the transient voltage safety assessment result is determined. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the isomorphic sample feature vector of the sample power system. The first sample line feature vector and the isomorphic sample feature vector are obtained based on the graph structure of the sample power system.

[0042] The aforementioned transient voltage safety assessment method and apparatus based on graph neural networks model the power system network topology to obtain a power system graph structure, which includes: a set of nodes and a set of lines in the power system network; construct initial node feature vectors corresponding to the node sets and first line feature vectors corresponding to the line sets; perform projection transformation on different feature vectors of heterogeneous node types in the initial node feature vectors according to the feature transformation matrix of the target type nodes to obtain homogenized node feature vectors; and determine the transient voltage safety assessment result based on the first line feature vector, the homogenized node feature vectors, and the pre-constructed transient voltage safety assessment model. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the homogenized sample feature vector of the sample power system. The first sample line feature vector and the homogenized sample feature vector are obtained based on the sample power system graph structure. By abstracting the power system into a graph structure model and using virtual isomorphism technology to unify the heterogeneous characteristics of nodes, isomorphic input data adapted to graph neural networks is constructed, achieving efficient mapping for transient voltage safety assessment. Through data-driven end-to-end prediction, computational efficiency is significantly improved while preserving physical interaction laws, supporting real-time safety analysis and intelligent decision-making in complex power grids, and providing reliable technical support for the stable operation of the system under high-proportion renewable energy access. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is an application environment diagram of a transient voltage safety assessment method based on graph neural networks in one embodiment;

[0045] Figure 2 This is a flowchart illustrating a transient voltage safety assessment method based on a graph neural network in one embodiment.

[0046] Figure 3 This is a flowchart illustrating a transient voltage safety assessment method based on a graph neural network in another embodiment.

[0047] Figure 4 This is a flowchart illustrating a transient voltage safety assessment method based on a graph neural network in another embodiment.

[0048] Figure 5 This is a flowchart illustrating a transient voltage safety assessment method based on a graph neural network in another embodiment.

[0049] Figure 6 This is a structural block diagram of a transient voltage safety assessment device based on a graph neural network in one embodiment. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0051] The transient voltage safety assessment method based on graph neural networks provided in this application can be applied to, for example... Figure 1 The application environment shown. The computer device can be a terminal, and its internal structure diagram can be as follows. Figure 1 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements a transient voltage safety assessment method based on a graph neural network. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device casing, or an external keyboard, touchpad or mouse.

[0052] Those skilled in the art will understand that Figure 1 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0053] In one embodiment, Figure 2As shown, a transient voltage safety assessment method based on graph neural networks is provided, which is then applied to... Figure 1 Taking the terminal in the example, the explanation includes:

[0054] S201, Model the power system network topology to obtain the power system graph structure, which includes the set of nodes and the set of lines in the power system network.

[0055] In this embodiment, the graph structure G can be represented as G = (V, E). First, the set of all buses is extracted from the steady-state operation data of the power system to form the node set V of the graph structure. The node set can be represented as: V = {v1, v2, ..., v...} n}, where each node v i This represents the busbars in a power system, such as substation buses and generator buses. The set of all transmission lines is extracted from the power system topology data, forming a graph-structured set E of lines. This set can be represented as: E = {e1, e2, ..., e...} m}, where each edge ek=(v i , v j ) represents node v i With v j The power transmission lines between them.

[0056] For example, suppose a power system contains 5 buses, then the node set is V={v1, v2, v3, v4, v5}; if there are line connections between buses v1 and v2, and between v1 and v3, then the line set includes e1=(v1, v2) and e2=(v1, v3).

[0057] Optionally, the topological connections in the graph structure are represented by an n×n adjacency matrix. When there is a connection between any two nodes in the node set, the elements corresponding to any two nodes in the adjacency matrix are 1; where n is the number of nodes in the node set.

[0058] In this embodiment, the adjacency matrix can be expressed as Equation 1:

[0059] (Equation 1)

[0060] For example, for the node set V = {v1, v2, v3, v4, v5} and the line set E = {(v1, v2), (v1, v3), (v2, v4), (v3, v5)}, the adjacency matrix is ​​as shown in Equation 2:

[0061] (Equation 2)

[0062] Furthermore, for each node vi, its neighbor set N(vi) is defined as the set of nodes directly connected by edges, N(v i )={v j |A ij =1,jeqi},For example: For node v1, its neighborhood node set is N(v1)={v2,v3}.

[0063] S202, construct the initial node feature vector corresponding to the node set, and the first line feature vector corresponding to the line set.

[0064] In this embodiment of the application, an initial node feature vector is determined based on the running status information of each node corresponding to the node set, and a first line feature vector is determined based on the running status information of each line corresponding to the line set.

[0065] For example, because the electrical equipment associated with each bus in a power system is different (such as loads, synchronous generators, etc.), the nodes in the corresponding graph structure data have heterogeneous characteristics. The nodes... The feature input vector is represented as ,line The feature input vector is represented as , and Let be the dimension of the feature vectors for this node and this line.

[0066] S203, based on the feature transformation matrix of the target type node, project and transform the different feature vectors of heterogeneous node types in the initial node feature vector to obtain the homogeneous node feature vector.

[0067] In this embodiment of the application, the feature transformation matrix can be Determined by the node type, the projection transformation of different feature vectors of heterogeneous node types in the initial node feature vector can be performed based on the feature transformation matrix of the target type node, as shown in Equation 3:

[0068] (Equation 3)

[0069] In this embodiment, the feature vectors of different heterogeneous node types are projected and transformed to obtain homogeneous node feature vectors, thus avoiding the inability to perform calculations due to different dimensions of different nodes.

[0070] S204. Based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model, the transient voltage safety assessment result is determined. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the isomorphic sample feature vector of the sample power system. The first sample line feature vector and the isomorphic sample feature vector are obtained based on the graph structure of the sample power system.

[0071] In this embodiment, an initial security assessment model can be pre-constructed based on a graph neural network. Further, the initial security assessment model is trained using historical feature vectors of the power system to obtain a transient voltage security assessment model. In this embodiment, the first line feature vector and the isomorphic node feature vector can be input into the transient voltage security assessment model to learn the mapping function for transient voltage security assessment and determine the transient voltage security assessment result.

[0072] Optionally, the model output can be a classification result or a continuous value. For example, the classification result can be "safe" or "unsafe," and the continuous value can be a voltage stability index. Depending on the specific application scenario, the output result can be used for decision support or early warning systems.

[0073] Optionally, the homogenized node feature vector and the first line feature vector can be used as input, and graph convolution operations can be used to aggregate the information of nodes and lines. Pooling operations can be used to reduce the size of the graph, extract global features, and then the output of the graph convolution layer can be passed to the fully connected layer for the final classification or regression task.

[0074] In the aforementioned transient voltage safety assessment method based on graph neural networks, the power system network topology is modeled to obtain the power system graph structure, which includes: a set of nodes and a set of lines in the power system network; initial node feature vectors corresponding to the node sets and first line feature vectors corresponding to the line sets are constructed; based on the feature transformation matrix of the target type nodes, different feature vectors of heterogeneous node types in the initial node feature vectors are projected and transformed to obtain homogenized node feature vectors; based on the first line feature vector, the homogenized node feature vectors, and the pre-constructed transient voltage safety assessment model, the transient voltage safety assessment result is determined. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the homogenized sample feature vector of the sample power system. The first sample line feature vector and the homogenized sample feature vector are obtained based on the sample power system graph structure. By abstracting the power system into a graph structure model and using virtual isomorphism technology to unify the heterogeneous characteristics of nodes, isomorphic input data adapted to graph neural networks is constructed, achieving efficient mapping for transient voltage safety assessment. Through data-driven end-to-end prediction, computational efficiency is significantly improved while preserving physical interaction laws, supporting real-time safety analysis and intelligent decision-making in complex power grids, and providing reliable technical support for the stable operation of the system under high-proportion renewable energy access.

[0075] In one embodiment, one implementation of S204 above is provided, such as... Figure 3 As shown, the above-mentioned "determining the transient voltage safety assessment result based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model" includes:

[0076] S301, based on the first line feature vector, add fault indication information of each line in the line set to obtain the second line feature vector.

[0077] Among them, the fault indication information is used to indicate whether the line is a line with a anticipated fault or a normal line.

[0078] In this embodiment of the application, the first line feature vector is the basic feature vector. Based on the fault information of each line, one-hot encoding can be introduced, that is, the fault indication information of each line in the line set can be added to obtain the second line feature vector.

[0079] In this embodiment, the fault indication information can be identified as fault=1 or fault=0, where fault=1 indicates that the line is designated as an N-1 anticipated fault line, and fault=0 indicates that the line is a normal line. It should be noted that the N-1 anticipated fault line is a classic scenario in power system security analysis, used to assess the impact on the power grid operation when a fault occurs in a certain transmission line (or transformer, bus, etc.) in the system.

[0080] Optionally, based on the equivalent resistance, equivalent reactance, active power, and reactive power of each line, fault indication information of each line in the line set can be added to obtain a second line feature vector. In this embodiment, the equivalent resistance, equivalent reactance, active power, and reactive power of each line can be determined by the acquisition equipment installed in the power grid system. Based on this, fault indication information (fault) is added to obtain the second line feature vector. Exemplarily, the second line feature vector can be represented as a 5-dimensional vector. ,in These are the equivalent resistance and reactance of the transmission line, respectively. It refers to the active and reactive power transmitted by the power transmission line.

[0081] Optionally, when the line has N-1 anticipated faults, the faulty component is removed from the system, regardless of its fault type. The anticipated fault set is generated by performing offline calculations on the system beforehand.

[0082] S302, based on the second line feature vector, the isomorphic node feature vector, and the transient voltage safety assessment model, the transient voltage safety assessment result is determined.

[0083] In this embodiment, an initial security assessment model can be pre-constructed based on a graph neural network. Furthermore, the initial security assessment model is trained using historical feature vectors of the power system to obtain a transient voltage security assessment model. In this embodiment, the second line feature vector and the isomorphic node feature vector can be input into the transient voltage security assessment model to learn the mapping function for transient voltage security assessment.

[0084] In the above-mentioned embodiments, fault indication information is introduced into the line feature vector to determine the second line feature vector, which makes the description of the line more comprehensive and improves the comprehensiveness and accuracy of the second line feature vector.

[0085] In one embodiment, one implementation of S202 above is provided, such as... Figure 4 As shown, the aforementioned "constructing the initial node feature vector corresponding to the node set" includes:

[0086] S401, Determine the node type of the node set.

[0087] S402, if the node type is a new energy type, then based on the active power, reactive power, and transient performance of the generator output, construct the initial node feature vector corresponding to the node set.

[0088] S403, if the node type is a non-new energy type, then construct the initial node feature vector corresponding to the node set based on the active power and reactive power output by the generator.

[0089] The node types in the node set can include new energy types and non-new energy types.

[0090] In this embodiment of the application, the node set may include different types of nodes. Non-new energy type nodes may be synchronous generator nodes. Basic characteristics may include active power, reactive power, terminal voltage, etc. The feature vector of new energy nodes may be based on the synchronous generator characteristics, with the addition of transient characteristics, such as inverter control parameters, fault ride-through capability flags, etc.

[0091] In the above application embodiments, different node types are allowed to carry differentiated features, such as transient parameters of new energy nodes, to avoid information loss due to truncation or padding; and when adding a new node type, only the corresponding projection matrix needs to be added, without modifying the model architecture.

[0092] In one embodiment, Figure 5 As shown, the above-mentioned transient voltage safety assessment method based on graph neural networks, after step S202, further includes:

[0093] S205, normalize the first line feature vector and the initial node feature vector.

[0094] In this embodiment, before projecting and transforming the different feature vectors of heterogeneous node types in the initial node feature vector according to the feature transformation matrix of the target type node, the first line feature vector and the initial node feature vector are normalized. For example, as shown in Equation 4:

[0095] (Equation 4)

[0096] in, and These are the maximum and minimum values ​​of the i-th input feature, respectively. After feature normalization, all features in the graph data are mapped to the interval [0,1].

[0097] In the above-mentioned embodiments, normalization calculation eliminates the dimensional differences between power system feature data (such as the significant differences in the magnitudes of voltage, power, and impedance), linearly maps multidimensional features to a unified interval, balances the weights of multi-source features, and avoids the model from being biased towards large-value features due to differences in numerical scale; accelerates gradient descent convergence, enabling the graph neural network to stably adjust parameters during training; suppresses the risk of numerical overflow, preventing extreme values ​​in matrix operations from causing computational anomalies; and at the same time, by weakening the dependence on absolute quantities, it enhances the model's generalization ability to unknown power grid structures and operating conditions, laying the foundation for efficient fusion of heterogeneous power system data and faithful modeling of physical laws.

[0098] In one embodiment, a complete transient voltage safety assessment method based on graph neural networks is provided, including:

[0099] S1. Model the power system network topology to obtain the power system graph structure, which includes the set of nodes and the set of lines in the power system network.

[0100] S2, determine the node type of the node set.

[0101] S3. If the node type is a new energy type, then based on the active power, reactive power output of the generator, and transient performance, construct the initial node feature vector corresponding to the normalized node set; if the node type is a non-new energy type, then based on the active power and reactive power output of the generator, construct the initial node feature vector corresponding to the normalized node set.

[0102] S4, construct the first feature vector corresponding to the set of routes.

[0103] S5, normalize the first line feature vector and the initial node feature vector.

[0104] S6. Based on the feature transformation matrix of the target type node, project and transform the different feature vectors of heterogeneous node types in the initial node feature vector to obtain the homogeneous node feature vector.

[0105] S7. Based on the equivalent resistance, equivalent reactance, active power and reactive power of each line, add fault indication information of each line in the line set to obtain the second line feature vector.

[0106] S8. Based on the second line feature vector, the isomorphic node feature vector, and the transient voltage safety assessment model, the transient voltage safety assessment result is determined; the transient voltage safety assessment model is constructed based on a graph neural network.

[0107] In the aforementioned transient voltage safety assessment method based on graph neural networks, the power system network topology is modeled to obtain the power system graph structure, which includes: a set of nodes and a set of lines in the power system network; initial node feature vectors corresponding to the node sets and first line feature vectors corresponding to the line sets are constructed; based on the feature transformation matrix of the target type nodes, different feature vectors of heterogeneous node types in the initial node feature vectors are projected and transformed to obtain homogenized node feature vectors; based on the first line feature vector, the homogenized node feature vectors, and the pre-constructed transient voltage safety assessment model, the transient voltage safety assessment result is determined. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the homogenized sample feature vector of the sample power system. The first sample line feature vector and the homogenized sample feature vector are obtained based on the sample power system graph structure. By abstracting the power system into a graph structure model and using virtual isomorphism technology to unify the heterogeneous characteristics of nodes, isomorphic input data adapted to graph neural networks is constructed, achieving efficient mapping for transient voltage safety assessment. Through data-driven end-to-end prediction, computational efficiency is significantly improved while preserving physical interaction laws, supporting real-time safety analysis and intelligent decision-making in complex power grids, and providing reliable technical support for the stable operation of the system under high-proportion renewable energy access.

[0108] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0109] Based on the same inventive concept, this application also provides a graph neural network-based transient voltage safety assessment device for implementing the graph neural network-based transient voltage safety assessment method described above. The solution provided by this device is similar to the implementation described in the above method. Therefore, the specific limitations in one or more graph neural network-based transient voltage safety assessment device embodiments provided below can be found in the limitations of the graph neural network-based transient voltage safety assessment method described above, and will not be repeated here.

[0110] In one embodiment, Figure 6As shown, a transient voltage safety assessment device based on graph neural networks is provided, comprising: an acquisition module 10, a construction module 11, a conversion module 12, and a determination module 13, wherein:

[0111] The acquisition module 10 is used to model the power system network topology to obtain the power system graph structure, which includes the set of nodes and the set of lines in the power system network.

[0112] Module 11 is used to construct the initial node feature vector corresponding to the node set and the first line feature vector corresponding to the line set.

[0113] The transformation module 12 is used to project and transform the different feature vectors of heterogeneous node types in the initial node feature vector according to the feature transformation matrix of the target type node, so as to obtain the homogeneous node feature vector.

[0114] The determination module 13 is used to determine the transient voltage safety assessment result based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the isomorphic sample feature vector of the sample power system. The first sample line feature vector and the isomorphic sample feature vector are obtained based on the graph structure of the sample power system.

[0115] In one embodiment, the determining module 13 includes: an adding unit and a first determining unit, wherein:

[0116] An additional unit is used to add fault indication information of each line in the line set to the first line feature vector to obtain a second line feature vector.

[0117] The first determining unit is used to determine the transient voltage safety assessment result based on the second line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model.

[0118] In one embodiment, the aforementioned adding unit is specifically used to add fault indication information of each line in the line set to the equivalent resistance, equivalent reactance, active power and reactive power of each line, so as to obtain a second line feature vector.

[0119] In one embodiment, the above-mentioned construction module includes: a second determining unit and a construction unit, wherein:

[0120] The second determining unit is used to determine the node type of the node set.

[0121] The construction unit is used to construct the initial node feature vector corresponding to the node set based on the active power, reactive power, and transient performance of the generator when the node type is a new energy type; and to construct the initial node feature vector corresponding to the node set based on the active power and reactive power of the generator when the node type is a non-new energy type.

[0122] In one embodiment, the aforementioned transient voltage safety assessment device based on graph neural networks further includes: a processing module for normalizing the first line feature vector and the initial node feature vector.

[0123] In one embodiment, the topological connection relationship in the graph structure described above is an n×n adjacency matrix. When there is a connection between any two nodes in the node set, the elements corresponding to any two nodes in the adjacency matrix are 1; where n is the number of nodes in the node set.

[0124] Each module in the aforementioned graph neural network-based transient voltage safety assessment device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0125] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0126] Modeling the power system network topology is used to obtain the power system graph structure, which includes the set of nodes and the set of lines in the power system network.

[0127] Construct the initial node feature vector corresponding to the node set, and the first line feature vector corresponding to the line set;

[0128] Based on the feature transformation matrix of the target type node, the different feature vectors of heterogeneous node types in the initial node feature vector are projected and transformed to obtain the homogenized node feature vector;

[0129] Based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model, the transient voltage safety assessment result is determined. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the isomorphic sample feature vector of the sample power system. The first sample line feature vector and the isomorphic sample feature vector are obtained based on the graph structure of the sample power system.

[0130] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0131] Based on the first line feature vector, fault indication information of each line in the line set is added to obtain the second line feature vector;

[0132] Based on the second line feature vector, the isomorphic node feature vector, and the transient voltage safety assessment model, the transient voltage safety assessment result is determined.

[0133] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0134] Based on the equivalent resistance, equivalent reactance, active power, and reactive power of each line, fault indication information of each line in the line set is added to obtain the second line feature vector.

[0135] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0136] Determine the node type of the node set;

[0137] If the node type is a new energy type, then based on the active power, reactive power, and transient performance of the generator output, an initial node feature vector corresponding to the node set is constructed.

[0138] If the node type is a non-new energy type, then the initial node feature vector corresponding to the node set is constructed based on the active power and reactive power output of the generator.

[0139] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0140] The feature vectors of the first line and the initial node feature vectors are normalized.

[0141] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0142] The topological connections in a graph structure are represented by an n×n adjacency matrix. When there is a connection between any two nodes in the node set, the corresponding elements of any two nodes in the adjacency matrix are 1; where n is the number of nodes in the node set.

[0143] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0144] Modeling the power system network topology is used to obtain the power system graph structure, which includes the set of nodes and the set of lines in the power system network.

[0145] Construct the initial node feature vector corresponding to the node set, and the first line feature vector corresponding to the line set;

[0146] Based on the feature transformation matrix of the target type node, the different feature vectors of heterogeneous node types in the initial node feature vector are projected and transformed to obtain the homogenized node feature vector;

[0147] Based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model, the transient voltage safety assessment result is determined. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the isomorphic sample feature vector of the sample power system. The first sample line feature vector and the isomorphic sample feature vector are obtained based on the graph structure of the sample power system.

[0148] In one embodiment, when the computer program is executed by a processor, it further performs the following steps:

[0149] Based on the first line feature vector, fault indication information of each line in the line set is added to obtain the second line feature vector;

[0150] Based on the second line feature vector, the isomorphic node feature vector, and the transient voltage safety assessment model, the transient voltage safety assessment result is determined.

[0151] In one embodiment, when the computer program is executed by a processor, it further performs the following steps:

[0152] Based on the equivalent resistance, equivalent reactance, active power, and reactive power of each line, fault indication information of each line in the line set is added to obtain the second line feature vector.

[0153] In one embodiment, when the computer program is executed by a processor, it further performs the following steps:

[0154] Determine the node type of the node set;

[0155] If the node type is a new energy type, then based on the active power, reactive power, and transient performance of the generator output, an initial node feature vector corresponding to the node set is constructed.

[0156] If the node type is a non-new energy type, then the initial node feature vector corresponding to the node set is constructed based on the active power and reactive power output of the generator.

[0157] In one embodiment, when the computer program is executed by a processor, it further performs the following steps:

[0158] The feature vectors of the first line and the initial node feature vectors are normalized.

[0159] In one embodiment, when the computer program is executed by a processor, it further performs the following steps:

[0160] The topological connections in a graph structure are represented by an n×n adjacency matrix. When there is a connection between any two nodes in the node set, the corresponding elements of any two nodes in the adjacency matrix are 1; where n is the number of nodes in the node set.

[0161] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0162] Modeling the power system network topology is used to obtain the power system graph structure, which includes the set of nodes and the set of lines in the power system network.

[0163] Construct the initial node feature vector corresponding to the node set, and the first line feature vector corresponding to the line set;

[0164] Based on the feature transformation matrix of the target type node, the different feature vectors of heterogeneous node types in the initial node feature vector are projected and transformed to obtain the homogenized node feature vector;

[0165] Based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model, the transient voltage safety assessment result is determined. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the isomorphic sample feature vector of the sample power system. The first sample line feature vector and the isomorphic sample feature vector are obtained based on the graph structure of the sample power system.

[0166] In one embodiment, when the computer program is executed by a processor, it further performs the following steps:

[0167] Based on the first line feature vector, fault indication information of each line in the line set is added to obtain the second line feature vector;

[0168] Based on the second line feature vector, the isomorphic node feature vector, and the transient voltage safety assessment model, the transient voltage safety assessment result is determined.

[0169] In one embodiment, when the computer program is executed by a processor, it further performs the following steps:

[0170] Based on the equivalent resistance, equivalent reactance, active power, and reactive power of each line, fault indication information of each line in the line set is added to obtain the second line feature vector.

[0171] In one embodiment, when the computer program is executed by a processor, it further performs the following steps:

[0172] Determine the node type of the node set;

[0173] If the node type is a new energy type, then based on the active power, reactive power, and transient performance of the generator output, an initial node feature vector corresponding to the node set is constructed.

[0174] If the node type is a non-new energy type, then the initial node feature vector corresponding to the node set is constructed based on the active power and reactive power output of the generator.

[0175] In one embodiment, when the computer program is executed by a processor, it further performs the following steps:

[0176] The feature vectors of the first line and the initial node feature vectors are normalized.

[0177] In one embodiment, when the computer program is executed by a processor, it further performs the following steps:

[0178] The topological connections in a graph structure are represented by an n×n adjacency matrix. When there is a connection between any two nodes in the node set, the corresponding elements of any two nodes in the adjacency matrix are 1; where n is the number of nodes in the node set.

[0179] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), data processing logic devices based on quantum computing, and the like.

[0180] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0181] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A transient voltage safety assessment method based on graph neural networks, characterized in that, The method includes: A power system network topology is modeled to obtain a power system graph structure, which includes: a set of nodes and a set of lines in the power system network; Construct the initial node feature vector corresponding to the node set, and the first line feature vector corresponding to the line set; Based on the feature transformation matrix of the target type node, the different feature vectors of heterogeneous node types in the initial node feature vector are projected and transformed to obtain homogenized node feature vectors. Based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model, the transient voltage safety assessment result is determined. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the isomorphic sample feature vector of the sample power system. The first sample line feature vector and the isomorphic sample feature vector are obtained based on the graph structure of the sample power system.

2. The method according to claim 1, characterized in that, The determination of transient voltage safety assessment results based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model includes: Based on the first line feature vector, fault indication information of each line in the line set is added to obtain the second line feature vector; Based on the second line feature vector, the isomorphic node feature vector, and the transient voltage safety assessment model, the transient voltage safety assessment result is determined.

3. The method according to claim 2, characterized in that, The first line feature vector includes the equivalent resistance, equivalent reactance, active power, and reactive power of each line; the second line feature vector is obtained by adding fault indication information of each line in the line set to the first line feature vector, including: Based on the equivalent resistance, equivalent reactance, active power, and reactive power of each line, fault indication information of each line in the line set is added to obtain a second line feature vector.

4. The method according to claim 1, characterized in that, The construction of the initial node feature vector corresponding to the node set includes: Determine the node type of the node set; If the node type is a new energy type, then based on the active power, reactive power, and transient performance of the generator output, an initial node feature vector corresponding to the node set is constructed. If the node type is a non-new energy type, then an initial node feature vector corresponding to the node set is constructed based on the active power and reactive power output by the generator.

5. The method according to claim 1, characterized in that, The method further includes: The first line feature vector and the initial node feature vector are normalized.

6. The method according to any one of claims 1-5, characterized in that, The topological connections in the graph structure are represented by an n×n adjacency matrix. When any two nodes in the node set are connected, the corresponding elements of those two nodes in the adjacency matrix are 1; where n is the number of nodes in the node set.

7. A transient voltage safety assessment device based on graph neural networks, characterized in that, The device includes: The acquisition module is used to model the topology of the power system network to obtain the power system graph structure, which includes: a set of nodes and a set of lines in the power system network; A construction module is used to construct the initial node feature vector corresponding to the node set and the first line feature vector corresponding to the line set. The transformation module is used to project and transform the different feature vectors of heterogeneous node types in the initial node feature vector according to the feature transformation matrix of the target type node, so as to obtain homogenized node feature vectors. The determination module is used to determine the transient voltage safety assessment result based on the first line feature vector, the isomorphic node feature vector, and the pre-constructed transient voltage safety assessment model. The transient voltage safety assessment model is obtained by training a graph neural network based on the first sample line feature vector and the isomorphic sample feature vector of the sample power system. The first sample line feature vector and the isomorphic sample feature vector are obtained based on the graph structure of the sample power system.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images