Method and device for identifying in-home topology node, computer device and storage medium

Abstract

The application relates to a household access topology node identification method and device, computer equipment and a storage medium. The method comprises the following steps: determining a mutation node and an affected weight in a current weighted topology graph according to the difference between a plurality of historical weighted topology graphs corresponding to household side data in an electric network and the current weighted topology graph; constructing a weighted adjacency matrix based on the mutation node, the weight between nodes having a connection relationship in the historical weighted topology graph, the affected weight and a binary adjacency matrix of the historical weighted topology graph; inputting the weighted adjacency matrix and an original feature matrix into a preset reasoning model for reasoning to obtain a node embedding vector; determining the identification information of each node in the current weighted topology graph according to the node embedding vector; and the identification information comprises at least one of a subregion, an identity label and a function category. The method can efficiently and accurately identify household access topology nodes.

Description

Technical Field

[0001] This application relates to the field of power grid topology technology, and in particular to a method, apparatus, computer equipment, and storage medium for identifying inbound topology nodes. Background Technology

[0002] In smart home, smart grid and IoT scenarios, the entry topology nodes will change dynamically. For example, various smart devices (such as home appliances, sensors and smart switches) will be frequently connected, removed or moved. These dynamic changes will make the entry network topology highly dynamic, which will make it much more difficult to accurately identify entry topology nodes and perceive network status.

[0003] In existing technologies, the identification of inbound topology nodes mainly relies on static models based on neural networks. The core idea is to collect the electrical characteristics or communication data of the nodes to construct a fixed topology map; use a pre-trained neural network model to identify and classify the nodes in the fixed topology map; and adopt an incremental learning method of full model retraining or global regularization to adapt when facing topology changes.

[0004] However, the above-mentioned method for identifying inbound topology nodes suffers from low efficiency. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, apparatus, computer equipment, and storage medium for identifying inbound topology nodes that can improve the identification efficiency of inbound topology nodes, in order to address the aforementioned technical problems.

[0006] Firstly, this application provides a method for identifying inbound topology nodes, including:

[0007] Based on the differences between multiple historical weighted topology graphs and the current weighted topology graph corresponding to the data from the household side of the power grid, determine the mutation nodes and affected weights in the current weighted topology graph;

[0008] A weighted adjacency matrix is ​​constructed based on mutated nodes, the weights of nodes with connections in the historical weighted topology, the affected weights, and the binary adjacency matrix of the historical weighted topology. The binary adjacency matrix is ​​obtained from the communication handshake protocol data in the inbound data of the historical weighted topology. The communication handshake protocol data includes link layer address allocation information and routing table information.

[0009] The weighted adjacency matrix and the original feature matrix are input into a preset inference model for inference to obtain the node embedding vector. The original feature matrix is ​​obtained based on the electrical feature data in the household-side data of the historical weighted topology. The electrical feature data includes current harmonic component data, voltage phase offset data, and power factor time series data. The node embedding vector includes the node's own electrical features, the node's connection relationship in the power grid, and the features and weights between neighboring nodes.

[0010] Based on the node embedding vector, determine the identification information of each node in the current weighted topology graph; the identification information includes at least one of the following: the sub-region to which it belongs, the identity label, and the functional category.

[0011] In one embodiment, determining the abrupt change nodes and affected weights in the current weighted topology based on the differences between multiple historical weighted topology maps corresponding to the data from the power grid's inbound side and the current weighted topology map includes:

[0012] Based on the differences between multiple historical weighted topology graphs and the current weighted topology graph corresponding to the data from the household side of the power grid, the mutation nodes in the current weighted topology graph are determined.

[0013] Centered on the mutation node, locate the affected weights in the current weighted topology graph along the message passing path, and update the affected weights according to the weight parameters affected by topology mutations in the preset inference model.

[0014] In one embodiment, the method further includes:

[0015] The legitimacy of the mutation nodes is verified to obtain the overall confidence level of the mutation nodes;

[0016] Under the condition that the overall confidence level meets the preset conditions, the abnormality score of the mutation node is obtained based on the anomaly analysis model.

[0017] If the anomaly degree is less than the preset value, the following steps are performed: locating the affected weights in the current weighted topology graph along the message passing path with the mutation node as the center, and updating the affected weights according to the weight parameters affected by the topology mutation in the preset inference model.

[0018] In one embodiment, the above-mentioned legality verification of the mutation node to obtain the overall confidence level of the mutation node includes:

[0019] Extract the harmonic feature vectors of the mutation nodes and determine the cosine similarity between the harmonic feature vectors of the mutation nodes and the template feature vectors in the known device library;

[0020] Based on the relationship between cosine similarity and a set threshold, the matching parameters between the electrical characteristics of the mutation node and the electrical characteristics of known devices in the known device library are determined.

[0021] Verify the compliance of field lengths and the standardization of service type declarations in the communication handshake protocol data to obtain the protocol stability parameters of the communication messages;

[0022] The overall confidence level of mutation nodes is determined based on cosine similarity, matching parameters, and protocol stability parameters.

[0023] In one embodiment, determining the abrupt change nodes in the current weighted topology based on the differences between multiple historical weighted topology maps corresponding to the data from the power grid's inbound side and the current weighted topology map includes:

[0024] By comparing the current weighted topology with the historical weighted topology, candidate mutation nodes are obtained;

[0025] For the common connection edges in the current weighted topology and the historical weighted topology, calculate the sum of the absolute values ​​of the current weight of the common connection edge in the current weighted topology and the historical weight in the historical weighted topology.

[0026] The connected components of the current weighted topology and the historical weighted topology are identified by the depth-first search algorithm, and the total number of split events and the total number of merge events of the connected components are counted.

[0027] The mutation nodes in the current weighted topology are determined based on the number of candidate mutation nodes, the sum of the absolute values ​​of their weights, the total number of split events, and the total number of merge events.

[0028] In one embodiment, constructing a weighted adjacency matrix includes:

[0029] For each pair of connected nodes in the binary adjacency matrix, obtain the voltage vector difference and current vector difference between the connected nodes;

[0030] The weights between nodes with connections are determined based on the voltage vector difference and current vector difference between the nodes.

[0031] A weighted adjacency matrix is ​​constructed based on the weights of nodes with connections and the binary adjacency matrix.

[0032] In one embodiment, determining the weights between connected nodes based on the voltage vector difference and current vector difference of the connected nodes includes:

[0033] The magnitude of the ratio of voltage vector difference to current vector difference is determined as the electrical impedance distance between nodes that are connected.

[0034] The reciprocal of the electrical impedance distance between connected nodes is used as the weight between the connected nodes.

[0035] Secondly, this application also provides a device for identifying inbound topology nodes, comprising:

[0036] The determination module is used to determine the mutation nodes and affected weights in the current weighted topology map based on the differences between multiple historical weighted topology maps corresponding to the data from the household side of the power grid.

[0037] The module is used to construct a weighted adjacency matrix based on mutated nodes, the weights between nodes with connections in the historical weighted topology graph, the affected weights, and the binary adjacency matrix of the historical weighted topology graph. The binary adjacency matrix is ​​obtained from the communication handshake protocol data in the inbound data of the historical weighted topology graph. The communication handshake protocol data includes link layer address allocation information and routing table information.

[0038] The inference module is used to input the weighted adjacency matrix and the original feature matrix into a preset inference model for inference to obtain the node embedding vector. The original feature matrix is ​​obtained from the electrical feature data in the inlet-side data of the historical weighted topology. The electrical feature data includes current harmonic component data, voltage phase offset data, and power factor time series data. The node embedding vector includes the node's own electrical features, the node's connection relationship in the power grid, and the features and weights between neighboring nodes.

[0039] The identification module is used to determine the identification information of each node in the current weighted topology graph based on the node embedding vector; the identification information includes at least one of the following: the sub-region to which it belongs, the identity label, and the functional category.

[0040] 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:

[0041] Based on the differences between multiple historical weighted topology graphs and the current weighted topology graph corresponding to the data from the household side of the power grid, determine the mutation nodes and affected weights in the current weighted topology graph;

[0042] A weighted adjacency matrix is ​​constructed based on mutated nodes, the weights of nodes with connections in the historical weighted topology, the affected weights, and the binary adjacency matrix of the historical weighted topology. The binary adjacency matrix is ​​obtained from the communication handshake protocol data in the inbound data of the historical weighted topology. The communication handshake protocol data includes link layer address allocation information and routing table information.

[0043] The weighted adjacency matrix and the original feature matrix are input into a preset inference model for inference to obtain the node embedding vector. The original feature matrix is ​​obtained based on the electrical feature data in the household-side data of the historical weighted topology. The electrical feature data includes current harmonic component data, voltage phase offset data, and power factor time series data. The node embedding vector includes the node's own electrical features, the node's connection relationship in the power grid, and the features and weights between neighboring nodes.

[0044] Based on the node embedding vector, determine the identification information of each node in the current weighted topology graph; the identification information includes at least one of the following: the sub-region to which it belongs, the identity label, and the functional category.

[0045] 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:

[0046] Based on the differences between multiple historical weighted topology graphs and the current weighted topology graph corresponding to the data from the household side of the power grid, determine the mutation nodes and affected weights in the current weighted topology graph;

[0047] A weighted adjacency matrix is ​​constructed based on mutated nodes, the weights of nodes with connections in the historical weighted topology, the affected weights, and the binary adjacency matrix of the historical weighted topology. The binary adjacency matrix is ​​obtained from the communication handshake protocol data in the inbound data of the historical weighted topology. The communication handshake protocol data includes link layer address allocation information and routing table information.

[0048] The weighted adjacency matrix and the original feature matrix are input into a preset inference model for inference to obtain the node embedding vector. The original feature matrix is ​​obtained based on the electrical feature data in the household-side data of the historical weighted topology. The electrical feature data includes current harmonic component data, voltage phase offset data, and power factor time series data. The node embedding vector includes the node's own electrical features, the node's connection relationship in the power grid, and the features and weights between neighboring nodes.

[0049] Based on the node embedding vector, determine the identification information of each node in the current weighted topology graph; the identification information includes at least one of the following: the sub-region to which it belongs, the identity label, and the functional category.

[0050] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0051] Based on the differences between multiple historical weighted topology graphs and the current weighted topology graph corresponding to the data from the household side of the power grid, determine the mutation nodes and affected weights in the current weighted topology graph;

[0052] A weighted adjacency matrix is ​​constructed based on mutated nodes, the weights of nodes with connections in the historical weighted topology, the affected weights, and the binary adjacency matrix of the historical weighted topology. The binary adjacency matrix is ​​obtained from the communication handshake protocol data in the inbound data of the historical weighted topology. The communication handshake protocol data includes link layer address allocation information and routing table information.

[0053] The weighted adjacency matrix and the original feature matrix are input into a preset inference model for inference to obtain the node embedding vector. The original feature matrix is ​​obtained based on the electrical feature data in the household-side data of the historical weighted topology. The electrical feature data includes current harmonic component data, voltage phase offset data, and power factor time series data. The node embedding vector includes the node's own electrical features, the node's connection relationship in the power grid, and the features and weights between neighboring nodes.

[0054] Based on the node embedding vector, determine the identification information of each node in the current weighted topology graph; the identification information includes at least one of the following: the sub-region to which it belongs, the identity label, and the functional category.

[0055] The aforementioned method, device, computer equipment, and storage medium for identifying inbound topology nodes accurately locates mutated nodes and affected weights by comparing the differences between historical and current weighted topology graphs. It also constructs a weighted adjacency matrix that fits the actual topology structure by combining communication handshake protocols. Simultaneously, it integrates multi-dimensional electrical features to form an original feature matrix. Then, it generates a node embedding vector that integrates the node's own electrical characteristics, topological connection relationships, and neighboring node feature weights based on an inference model. Finally, it effectively identifies the sub-region, identity label, and functional category of the node. Overall, it can make full use of dynamic topology change information and multi-source electrical data, significantly improving the identification efficiency, accuracy, robustness, and refinement of inbound power grid node identification, and better adapting to highly dynamic inbound network topology environments. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is an application environment diagram of the method for identifying inbound topology nodes in one embodiment;

[0058] Figure 2 This is a flowchart illustrating a method for identifying inbound topology nodes in one embodiment;

[0059] Figure 3This is a flowchart illustrating the method for identifying inbound topology nodes in another embodiment;

[0060] Figure 4 This is a flowchart illustrating the method for identifying inbound topology nodes in another embodiment;

[0061] Figure 5 This is a flowchart illustrating the method for identifying inbound topology nodes in another embodiment;

[0062] Figure 6 This is a flowchart illustrating the method for identifying inbound topology nodes in another embodiment;

[0063] Figure 7 This is a flowchart illustrating the method for identifying inbound topology nodes in another embodiment;

[0064] Figure 8 This is a flowchart illustrating the method for identifying inbound topology nodes in another embodiment;

[0065] Figure 9 This is a flowchart illustrating the method for identifying inbound topology nodes in another embodiment;

[0066] Figure 10 This is a structural block diagram of an identification device for an inbound topology node in one embodiment;

[0067] Figure 11 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0068] 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.

[0069] In smart home, smart grid and IoT scenarios, the entry topology nodes will change dynamically. For example, various smart devices (such as home appliances, sensors and smart switches) will be frequently connected, removed or moved. These dynamic changes will make the entry network topology highly dynamic, which will make it much more difficult to accurately identify entry topology nodes and perceive network status.

[0070] In existing technologies, the identification of inbound topology nodes mainly relies on static models based on neural networks. The core idea is to collect the electrical characteristics or communication data of the nodes to construct a fixed topology map; use a pre-trained neural network model to identify and classify the nodes in the fixed topology map; and adopt an incremental learning method of full model retraining or global regularization to adapt when facing topology changes.

[0071] However, the aforementioned methods for identifying inbound topology nodes suffer from low efficiency. Therefore, this application provides a method for identifying inbound topology nodes to address these issues.

[0072] The method for identifying inbound topology nodes provided in this application embodiment can be applied to, for example, Figure 1 In the application environment shown, the application environment includes a data storage system 102 and a server 104. The data storage system 102 stores the data that the server 104 needs to process. The data storage system 102 can be integrated onto the server 104, or it can be located in the cloud or on other network servers. The server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The data storage system 102 can communicate with the server 104. For example, the server 104 can send a data request to the data storage system 102 to retrieve data, and then process the retrieved data.

[0073] In other possible implementations, the method for identifying inbound topology nodes provided in this application can also be applied to terminals. Terminals can be, but are not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can be smart speakers, smart TVs, smart air conditioners, smart vehicle devices, etc. Portable wearable devices can be smartwatches, smart bracelets, head-mounted devices, etc.

[0074] In one exemplary embodiment, such as Figure 2 As shown, a method for identifying inbound topology nodes is provided, which can be applied to... Figure 1 Taking server 104 as an example, the explanation includes:

[0075] S201. Based on the differences between multiple historical weighted topology diagrams and the current weighted topology diagram corresponding to the data from the household side in the power grid, determine the mutation nodes and affected weights in the current weighted topology diagram.

[0076] The data received from the user includes electrical characteristic data and communication handshake protocol data. Electrical characteristic data includes current harmonic component data, voltage phase offset data, and power factor time series data. The current harmonic component data primarily extracts the harmonic components of the 3rd, 5th, and 7th harmonics. Voltage phase offset data refers to the voltage phase offset relative to the zero-crossing point. This electrical characteristic data is generally collected from the user's main incoming line or branch monitoring points and obtained through high-precision energy metering chips (e.g., HLW8032, ATT7053, etc.). The communication handshake protocol data includes link-layer address allocation information, routing table information, device identification codes, and service type declarations. The link-layer address allocation information supports protocols such as DLT645, Modbus, and HomePlugAV2. The communication handshake protocol data can be extracted from the communication process between smart devices (e.g., smart home appliances, distributed energy sources, energy storage devices, etc.) and the gateway.

[0077] The weighted topology graph represents the connection relationships between various users in the power grid, as well as the connection weights between associated users. Optionally, for historical data from the user side of the power grid, a historical weighted topology graph corresponding to the user side data can be constructed, and for current data from the user side of the power grid, a current weighted topology graph corresponding to the user side data can be constructed.

[0078] In this embodiment, historical data of the power grid's inbound side can be obtained in advance, and a historical weighted topology map corresponding to the inbound side data can be constructed based on the historical data of the power grid's inbound side. The historical weighted topology map is stored in a database for later use. When it is necessary to identify inbound topology nodes based on the current data of the power grid's inbound side, the current data of the power grid's inbound side can be obtained, and a current weighted topology map corresponding to the inbound side data can be constructed based on the current data of the power grid's inbound side. Then, the difference between the historical weighted topology map and the current weighted topology map is analyzed to obtain the mutation nodes and affected weights in the current weighted topology map.

[0079] Optionally, if historical data from multiple different times at the power grid's inbound side are obtained in advance, multiple historical weighted topology maps corresponding to the inbound side data can be constructed based on the historical data from multiple different times at the power grid's inbound side. Then, based on the differences between the multiple historical weighted topology maps and the current weighted topology map, the mutation nodes and affected weights in the current weighted topology map can be determined.

[0080] S202. Based on the mutation node, the weights between nodes with connections in the historical weighted topology graph, the affected weights, and the binary adjacency matrix of the historical weighted topology graph, construct a weighted adjacency matrix; the binary adjacency matrix is ​​obtained from the communication handshake protocol data in the inbound data of the historical weighted topology graph; the communication handshake protocol data includes link layer address allocation information and routing table information.

[0081] In this embodiment, based on the information of dynamically changing mutation nodes in the inbound network, combined with the original connection weights between nodes that already have connections in the historical weighted topology map, and the affected weight data generated by the dynamic changes in the topology, a binary adjacency matrix is ​​constructed based on the communication handshake protocol data extracted from the inbound side data of the historical weighted topology map. The communication handshake protocol data specifically includes link layer address allocation information and routing table information. On this basis, the weighted adjacency matrix is ​​constructed by integrating multiple types of key information.

[0082] S203. Input the weighted adjacency matrix and the original feature matrix into the preset inference model for inference to obtain the node embedding vector; the original feature matrix is ​​obtained based on the electrical feature data in the household side data of the historical weighted topology map; the electrical feature data includes current harmonic component data, voltage phase offset data and power factor time series data; the node embedding vector includes the node's own electrical features, the node's connection relationship in the power grid, and the features and weights between neighboring nodes.

[0083] The preset inference model aggregates neighbor features through weighted averaging, introduces nonlinearity through the ReLU activation function, and finally outputs a 128-dimensional node embedding vector. The preset inference model can be trained from multiple historical weighted adjacency matrices and multiple historical original feature matrices.

[0084] In this embodiment, after obtaining the weighted adjacency matrix, an original feature matrix can be constructed based on the electrical feature data in the inbound side data of the historical weighted topology map. Then, the weighted adjacency matrix and the original feature matrix are input into a preset inference model for inference to obtain a node embedding vector that includes the node's own electrical features, the node's connection relationship in the power grid, and the features and weights between neighboring nodes.

[0085] Optionally, the weighted adjacency matrix constructed by the mutation node, historical connection weight, affected weight, and binary adjacency matrix, together with the original feature matrix generated from the electrical feature data in the inbound topology data of the historical weighted topology map, is input into the preset inference model for inference calculation, thereby outputting the corresponding node embedding vector. The original feature matrix used here comes from multi-dimensional electrical feature information such as current harmonic component data, voltage phase offset data based on zero crossover point, and power factor time series data collected from the inbound side. The finally generated node embedding vector can integrate and characterize the node's own electrical characteristics, the node's connection relationship in the inbound power grid topology, and the association characteristics and connection weight information between each neighboring node.

[0086] Optionally, during the training of the preset inference model, regularization strength calculation is required to train the preset physical strength model based on the regularization strength calculation results. The calculation process includes:

[0087] (1) Structural difference normalization: ,in, The 95th percentile (typically 0.8) was determined for offline experiments.

[0088] (2) Equipment power classification: High power equipment (>2000W) P_class=1.0, medium power (200-2000W) P_class=0.6, low power (<200W) P_class=0.3;

[0089] (3) Historical stability index: S_hist = number of unmutated cycles / 100;

[0090] (4) Load constraint: When CPU utilization > 70%, the upper limit of regularization strength. _max=0.05; otherwise _max=0.15;

[0091] (5) Final strength: =min ( + P_class+ S_hist, _max)( =0.5、 =0.3、 =0.2).

[0092] (6) Regularization application: L2 regularization is adopted, and constraints are applied only to the updated weight subset to determine the regularization loss term. = || ||², Overall loss = + ( (For node classification, cross-entropy loss).

[0093] S204. Based on the node embedding vector, determine the identification information of each node in the current weighted topology graph; the identification information includes at least one of the following: the sub-region to which it belongs, the identity label, and the functional category.

[0094] In this embodiment, after the inference and generation of the node embedding vector is completed, the network nodes in the current weighted topology are comprehensively judged and classified based on the multi-dimensional representation information contained in the vector, and then the identification information corresponding to each node is determined. This identification information can be used to clearly characterize the attributes and roles of the node in the home access topology, specifically including at least one of the sub-region to which the node belongs in the power grid topology, the identity label used to distinguish individuals, and the functional category reflecting the purpose of the equipment, thereby realizing the refined identification and labeling of dynamic home access topology nodes.

[0095] In this embodiment, the difference between historical and current weighted topology graphs is compared to accurately locate mutated nodes and affected weights. A weighted adjacency matrix that fits the actual topology structure is constructed by combining the communication handshake protocol. At the same time, multi-dimensional electrical features are integrated to form an original feature matrix. Then, a node embedding vector is generated based on the inference model, which integrates the node's own electrical characteristics, topological connection relationships and neighboring node feature weights. Finally, the sub-region, identity label and functional category of the node are effectively identified. Overall, it can make full use of the dynamic topology change information and multi-source electrical data, significantly improve the identification efficiency, accuracy, robustness and refinement of the identification of the grid node, and better adapt to the highly dynamic grid topology environment.

[0096] In this embodiment, in the above Figure 2 Based on the illustrated embodiment, a detailed process will be explained to determine the abrupt change nodes and affected weights in the current weighted topology map by considering the differences between multiple historical weighted topology maps and the current weighted topology map corresponding to data from the grid's inbound side. In an exemplary embodiment, such as Figure 3 As shown, the above method also includes:

[0097] S301. Based on the differences between the multiple historical weighted topology maps and the current weighted topology map corresponding to the data from the household side of the power grid, determine the mutation nodes in the current weighted topology map.

[0098] In this embodiment, by comprehensively comparing and analyzing multiple historical weighted topology maps corresponding to the grid access side data with the weighted topology map generated at the current moment, the focus is on extracting and comparing the differences between nodes in terms of existence status and access location, the differences in the connection relationship between physical links and communication associations between nodes, and the weight differences corresponding to the link association strength and interaction degree. By combining the above multi-dimensional difference information, the abrupt nodes that have undergone dynamic changes in the current weighted topology map can be accurately determined.

[0099] Optionally, the following provides a specific implementation method for determining the abrupt change nodes and affected weights in the current weighted topology map based on the differences between multiple historical weighted topology maps corresponding to the data from the user side of the power grid and the current weighted topology map. See [link to implementation details]. Figure 4 The aforementioned S301 includes:

[0100] S3011. Compare the current weighted topology with the historical weighted topology to obtain candidate mutation nodes.

[0101] Among them, candidate mutation nodes include candidate newly added nodes and candidate removed nodes.

[0102] In this embodiment, after obtaining the current weighted topology and the historical weighted topology, the current weighted topology can be compared with the historical weighted topology to obtain candidate mutation nodes.

[0103] S3012. For the common connection edges in the current weighted topology graph and the historical weighted topology graph, calculate the sum of the absolute values ​​of the current weight of the common connection edge in the current weighted topology graph and the historical weight in the historical weighted topology graph.

[0104] In this embodiment, common connection edges that coexist and remain connected in the current weighted topology graph and the historical weighted topology graph are first selected. For each determined common connection edge, its current weight in the current weighted topology graph and its historical weight in the historical weighted topology graph are extracted. Then, the current weight and the historical weight are numerically summed and the absolute value is taken to complete the calculation of the sum of the absolute values ​​of the weights of the corresponding common connection edges.

[0105] S3013. Identify the connected components of the current weighted topology and the historical weighted topology using a depth-first search algorithm, and count the total number of split events and the total number of merge events of the connected components.

[0106] In this embodiment, a depth-first search (DFS) algorithm is used to traverse and analyze the current weighted topology and the historical weighted topology, respectively, to accurately identify the connected components and their inclusion ranges in the two graphs. Based on this, the changes in connectivity of the same group of nodes under the historical and current topologies are compared and analyzed. The total number of splitting events caused by dynamic changes in the topology leading to the splitting of connected components, as well as the total number of merging events where multiple independent connected components aggregate to form larger connected components, are determined and counted one by one. This fully depicts the connectivity evolution characteristics of the inbound network topology during dynamic changes.

[0107] S3014. Based on the number of candidate mutation nodes, the sum of absolute weights, the total number of split events, and the total number of merge events, determine the mutation nodes in the current weighted topology graph.

[0108] In this embodiment, after obtaining the number of candidate mutation nodes, the sum of their absolute weights, the total number of splitting events, and the total number of merging events, the structural difference metric can be calculated according to the following formula (1). :

[0109]

[0110] in, , , and These are the correction factors, =0.4、 =0.3, γ=0.2, =0.1, This refers to the number of newly added nodes among the candidate mutation nodes. This refers to the number of nodes removed from the mutated nodes. It refers to the sum of the absolute values ​​of the weights. It refers to the sum of the total number of split events and the total number of merge events.

[0111] In this embodiment, the structural difference metric value is calculated using the above formula (1). Subsequently, if the structural difference metric If the value is greater than a preset threshold (e.g., 0.15), then the candidate mutation node is determined to be the mutation node in the current weighted topology graph.

[0112] S302. Using the mutation node as the center, locate the affected weights in the current weighted topology graph along the message passing path, and update the affected weights according to the weight parameters affected by topology mutations in the preset inference model.

[0113] Among them, the weight parameters affected by topological mutations in the preset inference model refer to the historical importance of the evaluation parameters (I(w_i)) and the topological correlation (A(w_i)).

[0114] In this embodiment, the identified mutation node is used as the core central node. The message transmission and data interaction paths in the power grid access topology are traced and the scope is located layer by layer to determine the affected weights in the current weighted topology graph that have changed due to node mutation. Based on this, the affected weights obtained above are adaptively updated by combining the weight parameters in the preset inference model that are sensitive to topological dynamic changes and easily affected by topological mutations. This allows the weight values ​​to accurately reflect the actual correlation strength after the topology has changed dynamically, providing a weight basis that is more in line with the real network state for subsequent topology modeling and node inference.

[0115] Optionally, after locating the affected weights in the current weighted topology graph along the message passing path, the sensitivity score S(w_i) of each affected weight to topology changes can be calculated, and a filtering threshold can be dynamically set in conjunction with the real-time system load. Select S(w_i)> The weighted subset is used as the object to be updated.

[0116] Optionally, based on the weight parameters affected by topological mutations in the preset inference model, update the affected weights, including:

[0117] (1) If the topological correlation (A (w_i)) is high and the historical importance of the evaluation parameter (I (w_i)) is low, then the affected weights are updated based on the standard learning rate;

[0118] (2) If the topological correlation (A (w_i)) is high and the historical importance of the evaluation parameter (I (w_i)) is high, then the affected weights are updated based on halving the learning rate;

[0119] (3) If the topological correlation degree (A (w_i)) is low and the historical importance of the evaluation parameter (I (w_i)) is high, then the affected weights will not be updated based on freezing.

[0120] (4) If the topological correlation (A (w_i)) is low and the historical importance of the evaluation parameter (I (w_i)) is low, the affected weights remain unchanged.

[0121] Safety rollback mechanism: Before the update, a weight shadow copy is created. After the update, a small verification dataset is used for quick evaluation. If the historical recognition accuracy drops by more than 2% or the new node recognition accuracy is lower than 85%, the weight will be automatically rolled back to the backup weight.

[0122] In this embodiment, by comprehensively comparing the multidimensional differences in nodes, connections, and weights between historical and current weighted topologies, abrupt changes in the topology can be accurately and efficiently identified, enabling rapid perception of dynamic network changes. Based on this, the affected weights are located along the message passing path centered on the abrupt changes, and updated in conjunction with the weight parameters affected by topology changes in the inference model. This ensures the accuracy of the affected range location and allows the weight values ​​to adapt to the real network state after the topology change in a timely manner, effectively improving the timeliness and reliability of weight information. At the same time, it avoids redundant calculation and updates of global weights, significantly improving the efficiency and robustness of dynamic inbound network topology processing.

[0123] In this embodiment, in the above Figure 3 Based on the embodiments shown, such as Figure 5 As shown, the above method also includes:

[0124] S303. Verify the legitimacy of the mutation node to obtain the overall confidence level of the mutation node.

[0125] The legitimacy verification includes: electrical fingerprint matching verification, protocol integrity verification, and multi-cycle consistency analysis.

[0126] In this embodiment, after identifying the mutation node in the current weighted topology graph, the access behavior, identity information and data interaction legality of the mutation node are comprehensively verified by combining the node's historical connection characteristics, communication behavior patterns, electrical characteristic consistency and topology structure rationality and other multi-dimensional verification rules. The weighted evaluation and quantitative judgment of various verification indicators are carried out to finally obtain a comprehensive confidence level that can reflect the true credibility of the mutation node.

[0127] Optionally, the following provides a specific implementation method for validating the legitimacy of mutation nodes and obtaining their overall confidence level, such as... Figure 6 As shown, the above S303 includes:

[0128] S3031. Extract the harmonic feature vector of the mutation node and determine the cosine similarity between the harmonic feature vector of the mutation node and the template feature vector in the known device library.

[0129] In this embodiment, electrical fingerprint matching verification can be performed on the mutation node. The process includes: extracting the harmonic feature vector of the mutation node and determining the cosine similarity between the harmonic feature vector of the mutation node and the template feature vector in the known device library.

[0130] S3032. Based on the relationship between cosine similarity and a set threshold, determine the matching parameters between the electrical characteristics of the mutation node and the electrical characteristics of known devices in the known device library.

[0131] The preset value can be 0.85.

[0132] In this embodiment, after obtaining the cosine similarity, if the cosine similarity is less than a set threshold, the matching parameters between the electrical characteristics of the mutation node and the electrical characteristics of the known devices in the known device library can be determined, and if the cosine similarity is not less than the set threshold, no further processing is required.

[0133] S3033. Verify the compliance of field lengths and service type declarations in the communication handshake protocol data to obtain the protocol stability parameters of the communication message.

[0134] In this embodiment, the compliance of the length of the communication handshake protocol data field and the standardization of the service type declaration are verified to obtain the protocol stability parameters of the communication message.

[0135] Optionally, multi-period consistency analysis should also be performed on the mutation nodes. The analysis process includes: analyzing whether the device identification code, power factor, and communication behavior pattern remain stable within three consecutive sampling periods.

[0136] S3034. Based on cosine similarity, matching parameters, and protocol stability parameters, determine the overall confidence level of the mutation node.

[0137] In this embodiment, after obtaining the cosine similarity, matching parameter, and protocol stability parameter, three verification weights can be assigned to the cosine similarity, matching parameter, and protocol stability parameter respectively. The overall confidence of the mutation node is determined by the sum of the product of cosine similarity and its corresponding verification weight, the product of matching parameter and its corresponding verification weight, and the product of protocol stability parameter and its corresponding verification weight.

[0138] S304. Under the condition that the overall confidence level meets the preset conditions, the anomaly score of the mutation node is obtained based on the anomaly analysis model.

[0139] The preset condition can be that the overall confidence level is greater than a preset value, and the preset value can be 0.75.

[0140] The anomaly analysis model can be a Gaussian mixture model (GMM_base).

[0141] In this embodiment, when the overall confidence level meets the preset conditions, the communication handshake protocol data and electrical characteristic data of the mutation node are input into the anomaly analysis model to perform anomaly analysis on the mutation node and obtain the anomaly score of the mutation node.

[0142] Optionally, if the anomaly score of the mutated node exceeds a preset value (generally 2.5), it indicates that the mutated node is an abnormal device and the update is refused.

[0143] The aforementioned S302 includes:

[0144] If the anomaly score is less than the preset value, the affected weights in the current weighted topology graph are located along the message passing path with the mutation node as the center, and the affected weights are updated according to the weight parameters affected by the topology mutation in the preset inference model.

[0145] In this embodiment, if the anomaly degree is less than a preset value, it indicates that the mutated node is a normal node. At this time, the steps of locating the affected weights in the current weighted topology graph along the message passing path with the mutated node as the center and updating the affected weights according to the weight parameters affected by the topology mutation in the preset inference model can be performed.

[0146] In this embodiment, by first verifying the legitimacy of mutated nodes and obtaining a comprehensive confidence level, and then using an anomaly analysis model to calculate the anomaly score for nodes that meet the confidence level conditions, the weight location and update steps are only executed when the anomaly score is lower than a preset threshold. This effectively filters out interference from illegal nodes and highly abnormal nodes, avoids redundant calculations triggered by invalid or abnormal topology changes, improves the accuracy and reliability of weight updates, and reduces unnecessary model calculation overhead, making the entire dynamic processing of the inbound network topology more rigorous, efficient, and stable.

[0147] In this embodiment, in the above Figure 2 Based on the illustrated embodiment, the detailed process of constructing a weighted adjacency matrix will be explained. In an exemplary embodiment, such as Figure 7 As shown, the above method also includes:

[0148] S401. For each pair of nodes in the binary adjacency matrix that have a connection relationship, obtain the voltage vector difference and current vector difference of the nodes that have a connection relationship.

[0149] In this embodiment, after constructing a binary adjacency matrix based on the link layer address allocation information and routing table information in the inbound-side data of the historical weighted topology map, the voltage vector difference and current vector difference of the connected nodes can be obtained for each pair of connected nodes in the binary adjacency matrix.

[0150] S402. Determine the weights between nodes that are connected based on the voltage vector difference and current vector difference.

[0151] In this embodiment, after extracting the electrical quantities of each pair of connected nodes in the binary adjacency matrix, the voltage vector difference and current vector difference between the nodes are calculated to fully explore the electrical correlation strength and transmission characteristics reflected by the differences in electrical phase and amplitude between the nodes. The vector difference is normalized and quantized in combination with the electrical characteristic constraints of the power grid at the household entrance. Based on this, the correlation weight between connected nodes is calculated and determined so that the weight values ​​can truly reflect the degree of electrical coupling between nodes and the actual interaction state of the link.

[0152] Optionally, the following provides a specific implementation method for determining the weights between connected nodes based on the voltage vector difference and current vector difference of the connected nodes. See [link to implementation details]. Figure 8 The aforementioned S402 includes:

[0153] S4021. The magnitude of the ratio of voltage vector difference to current vector difference is determined as the electrical impedance distance between nodes that are connected.

[0154] S4022. Take the reciprocal of the electrical impedance distance between nodes that are connected, and use it as the weight between nodes that are connected.

[0155] In this embodiment, after obtaining the voltage vector difference and current vector difference of the nodes with a connection relationship, the magnitude of the ratio of the voltage vector difference and the current vector difference can be determined as the electrical impedance distance between the nodes with a connection relationship, and the reciprocal of the electrical impedance distance between the nodes with a connection relationship can be used as the weight between the nodes with a connection relationship.

[0156] S403. Construct a weighted adjacency matrix based on the weights between nodes with connections and the binary adjacency matrix.

[0157] In this embodiment, after obtaining the weights between nodes with connections, the weights between nodes with connections can be combined with the binary adjacency matrix to construct a weighted adjacency matrix.

[0158] In this embodiment, the voltage vector difference and current vector difference are extracted and calculated for each pair of nodes with a connection relationship in the binary adjacency matrix. Based on this, the association weight between nodes is dynamically determined. Then, a weighted adjacency matrix is ​​constructed by combining the topological connection relationship of the binary adjacency matrix. This can closely integrate the electrical quantity characteristics of the power grid physical layer with the topological structure, so that the weight in the adjacency matrix no longer depends solely on the communication connection relationship, but truly reflects the degree of electrical coupling and link transmission characteristics between nodes. This effectively improves the accuracy and physical meaning of the weighted adjacency matrix, and provides data support that is more in line with the actual power grid operation status for subsequent node identification and topology reasoning.

[0159] In one exemplary embodiment, see Figure 9 It also provides a method for identifying inbound topology nodes, including:

[0160] T1. Compare the current weighted topology with the historical weighted topology to obtain candidate mutation nodes;

[0161] T2. For the common connection edges in the current weighted topology and the historical weighted topology, calculate the sum of the absolute values ​​of the current weight of the common connection edge in the current weighted topology and the historical weight in the historical weighted topology.

[0162] T3. Identify the connected components of the current weighted topology and the historical weighted topology using the depth-first search algorithm, and count the total number of split events and the total number of merge events of the connected components;

[0163] T4. Determine the mutation nodes in the current weighted topology graph based on the number of candidate mutation nodes, the sum of the absolute values ​​of their weights, the total number of split events, and the total number of merge events.

[0164] T5. Extract the harmonic feature vector of the mutation node and determine the cosine similarity between the harmonic feature vector of the mutation node and the template feature vector in the known equipment library.

[0165] T6. Based on the relationship between cosine similarity and a set threshold, determine the matching parameters between the electrical characteristics of the mutation node and the electrical characteristics of known devices in the known device library;

[0166] T7. Verify the compliance of field lengths and service type declarations in the communication handshake protocol data to obtain the protocol stability parameters of the communication message;

[0167] T8. Determine the overall confidence level of the mutation node based on cosine similarity, matching parameters, and protocol stability parameters;

[0168] T9. Under the condition that the overall confidence level meets the preset conditions, the anomaly score of the mutation node is obtained based on the anomaly analysis model.

[0169] T10. If the anomaly is less than the preset value, locate the affected weights in the current weighted topology graph along the message passing path with the mutation node as the center, and update the affected weights according to the weight parameters affected by topology mutation in the preset inference model.

[0170] T11. Based on the mutation node, the weights of nodes with connections in the historical weighted topology graph, the affected weights, and the binary adjacency matrix of the historical weighted topology graph, for each pair of nodes with connections in the binary adjacency matrix, obtain the voltage vector difference and current vector difference of the nodes with connections; the binary adjacency matrix is ​​obtained from the communication handshake protocol data in the ingress data of the historical weighted topology graph; the communication handshake protocol data includes link layer address allocation information and routing table information;

[0171] T12. The ratio of voltage vector difference to current vector difference is used as the electrical impedance distance between nodes that are connected.

[0172] T13. Take the reciprocal of the electrical impedance distance between nodes that are connected, and use it as the weight between nodes that are connected;

[0173] T14. Construct a weighted adjacency matrix based on the weights between nodes with connections and the binary adjacency matrix;

[0174] T15. Input the weighted adjacency matrix and the original feature matrix into the preset inference model for inference to obtain the node embedding vector; the original feature matrix is ​​obtained based on the electrical feature data in the household side data of the historical weighted topology map; the electrical feature data includes current harmonic component data, voltage phase offset data and power factor time series data; the node embedding vector includes the node's own electrical features, the node's connection relationship in the power grid, and the features and weights between neighboring nodes;

[0175] T16. Based on the node embedding vector, determine the identification information of each node in the current weighted topology graph; the identification information includes at least one of the following: the sub-region to which it belongs, the identity label, and the functional category.

[0176] It should be noted that the descriptions of T1-T16 above can be found in the relevant descriptions in the above embodiments, and their effects are similar, so they will not be repeated here.

[0177] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0178] Based on the same inventive concept, this application also provides an access topology node identification device for implementing the above-described method for identifying access topology nodes. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more access topology node identification device embodiments provided below can be found in the limitations of the access topology node identification method described above, and will not be repeated here.

[0179] In one exemplary embodiment, such as Figure 10 As shown, a device for identifying inbound topology nodes is provided, comprising: a determination module 10, a construction module 11, a reasoning module 12, and an identification module 13, wherein:

[0180] The determination module 10 is used to determine the mutation nodes and affected weights in the current weighted topology map based on the differences between multiple historical weighted topology maps and the current weighted topology map corresponding to the data from the household side in the power grid.

[0181] Module 11 is used to construct a weighted adjacency matrix based on mutation nodes, the weights between nodes with connections in the historical weighted topology graph, the affected weights, and the binary adjacency matrix of the historical weighted topology graph. The binary adjacency matrix is ​​obtained from the communication handshake protocol data in the inbound data of the historical weighted topology graph. The communication handshake protocol data includes link layer address allocation information and routing table information.

[0182] The reasoning module 12 is used to input the weighted adjacency matrix and the original feature matrix into a preset reasoning model for reasoning to obtain the node embedding vector. The original feature matrix is ​​obtained based on the electrical feature data in the inbound side data of the historical weighted topology map. The electrical feature data includes current harmonic component data, voltage phase offset data, and power factor time series data. The node embedding vector includes the node's own electrical features, the node's connection relationship in the power grid, and the features and weights between neighboring nodes.

[0183] The identification module 13 is used to determine the identification information of each node in the current weighted topology graph based on the node embedding vector; the identification information includes at least one of the sub-region to which it belongs, identity label and functional category.

[0184] In an exemplary embodiment, the determining module 10 includes:

[0185] The determination unit is specifically used to determine the mutation nodes in the current weighted topology map based on the differences between multiple historical weighted topology maps and the current weighted topology map corresponding to the data from the household side in the power grid.

[0186] The update unit is specifically used to locate the affected weights in the current weighted topology graph along the message passing path, with the mutation node as the center, and update the affected weights according to the weight parameters affected by the topology mutation in the preset inference model.

[0187] In an exemplary embodiment, the determining module 10 further includes:

[0188] The verification unit is specifically used to verify the legitimacy of the mutation node and obtain the overall confidence level of the mutation node;

[0189] The analysis unit is specifically used to obtain the anomaly score of the mutation node based on the anomaly analysis model, provided that the overall confidence level meets the preset conditions.

[0190] The aforementioned update unit is further configured to, when the anomaly degree is less than a preset value, locate the affected weights in the current weighted topology graph along the message passing path with the mutation node as the center, and update the affected weights according to the weight parameters affected by topology mutations in the preset inference model.

[0191] In an exemplary embodiment, the verification unit is further configured to extract the harmonic feature vector of the mutation node, determine the cosine similarity between the harmonic feature vector of the mutation node and the template feature vector in the known device library; determine the matching parameters between the electrical features of the mutation node and the electrical features of known devices in the known device library based on the relationship between the cosine similarity and a set threshold; verify the compliance of the field length and the standardization of the service type declaration of the communication handshake protocol data to obtain the protocol stability parameters of the communication message; and determine the comprehensive confidence level of the mutation node based on the cosine similarity, the matching parameters, and the protocol stability parameters.

[0192] In an exemplary embodiment, the determining unit is further configured to compare the current weighted topology with the historical weighted topology to obtain candidate mutation nodes; calculate the sum of the absolute values ​​of the current weight and the historical weight of the common connection edges in the current weighted topology and the historical weights in the historical weighted topology for the common connection edges in the current weighted topology; identify the connected components of the current weighted topology and the historical weighted topology using a depth-first search algorithm, and count the total number of split events and the total number of merge events of the connected components; and determine the mutation nodes in the current weighted topology based on the number of candidate mutation nodes, the sum of the absolute values ​​of the weights, the total number of split events, and the total number of merge events.

[0193] In one exemplary embodiment, the above-described apparatus further includes:

[0194] The acquisition module is used to acquire the voltage vector difference and current vector difference of each pair of connected nodes in the binary adjacency matrix.

[0195] The determination module is used to determine the weights between nodes that are connected based on the voltage vector difference and current vector difference of the nodes that are connected.

[0196] The building module is used to construct a weighted adjacency matrix based on the weights and binary adjacency matrices between nodes with existing connections.

[0197] In one exemplary embodiment, the determining module further includes:

[0198] The first determining unit is specifically used to determine the magnitude of the ratio of voltage vector difference to current vector difference as the electrical impedance distance between nodes that have a connection relationship;

[0199] The second determining unit is specifically used to take the reciprocal of the electrical impedance distance between nodes that are connected, and use it as the weight between the nodes that are connected.

[0200] Each module in the aforementioned identification device for the inbound topology node can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0201] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 11 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O 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, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores data from the power grid's inbound side. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a method for identifying inbound topology nodes.

[0202] Those skilled in the art will understand that Figure 11 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.

[0203] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0204] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0205] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0206] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0207] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory 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 many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0208] 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 application.

[0209] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for identifying inbound topology nodes, characterized in that, The method includes: Based on the differences between multiple historical weighted topology graphs and the current weighted topology graph corresponding to the data from the household side of the power grid, the mutation nodes and affected weights in the current weighted topology graph are determined. Based on the mutated node, the weights between nodes with connections in the historical weighted topology, the affected weights, and the binary adjacency matrix of the historical weighted topology, a weighted adjacency matrix is ​​constructed; the binary adjacency matrix is ​​obtained from the communication handshake protocol data in the inbound data of the historical weighted topology; the communication handshake protocol data includes link layer address allocation information and routing table information; The weighted adjacency matrix and the original feature matrix are input into a preset inference model for inference to obtain a node embedding vector. The original feature matrix is ​​obtained based on the electrical feature data in the household-side data of the historical weighted topology. The electrical feature data includes current harmonic component data, voltage phase offset data, and power factor time series data. The node embedding vector includes the node's own electrical features, the node's connection relationship in the power grid, and the features and weights between neighboring nodes. Based on the node embedding vector, the identification information of each node in the current weighted topology graph is determined; the identification information includes at least one of the sub-region to which it belongs, identity label, and functional category.

2. The method according to claim 1, characterized in that, The step of determining the abrupt change nodes and affected weights in the current weighted topology based on the differences between multiple historical weighted topology maps corresponding to the data from the power grid's inbound side includes: Based on the differences between multiple historical weighted topology graphs and the current weighted topology graph corresponding to the data from the household side of the power grid, the mutation nodes in the current weighted topology graph are determined; Centered on the mutation node, the affected weights in the current weighted topology graph are located along the message passing path, and the affected weights are updated according to the weight parameters affected by topology mutations in the preset inference model.

3. The method according to claim 2, characterized in that, The method further includes: The legality of the mutation node is verified to obtain the overall confidence level of the mutation node; When the overall confidence level meets the preset conditions, the anomaly score of the mutation node is obtained based on the anomaly analysis model. If the anomaly degree is less than a preset value, the following steps are performed: locating the affected weights in the current weighted topology graph along the message passing path with the mutation node as the center, and updating the affected weights according to the weight parameters affected by the topology mutation in the preset inference model.

4. The method according to claim 3, characterized in that, The process of validating the mutation node to obtain its overall confidence level includes: Extract the harmonic feature vector of the mutation node and determine the cosine similarity between the harmonic feature vector of the mutation node and the template feature vector in the known device library; Based on the relationship between the cosine similarity and the set threshold, the matching parameters between the electrical characteristics of the mutation node and the electrical characteristics of the known devices in the known device library are determined; Verify the compliance of field lengths and the standardization of service type declarations in the communication handshake protocol data to obtain the protocol stability parameters of the communication message; The overall confidence level of the mutation node is determined based on the cosine similarity, the matching parameters, and the protocol stability parameters.

5. The method according to claim 2, characterized in that, The step of determining the abrupt change nodes in the current weighted topology map based on the differences between multiple historical weighted topology maps and the current weighted topology map corresponding to the data from the power grid's inbound side includes: The current weighted topology is compared with the historical weighted topology to obtain candidate mutation nodes; For the common connection edges in the current weighted topology and the historical weighted topology, calculate the sum of the absolute values ​​of the current weight of the common connection edge in the current weighted topology and the historical weight in the historical weighted topology; The connected components of the current weighted topology and the historical weighted topology are identified by a depth-first search algorithm, and the total number of split events and the total number of merge events of the connected components are counted. The mutation nodes in the current weighted topology are determined based on the number of candidate mutation nodes, the sum of the absolute values ​​of the weights, the total number of split events, and the total number of merge events.

6. The method according to any one of claims 1-5, characterized in that, Constructing the weighted adjacency matrix includes: For each pair of nodes with a connection relationship in the binary adjacency matrix, obtain the voltage vector difference and current vector difference of the nodes with a connection relationship; The weights between the connected nodes are determined based on the voltage vector difference and current vector difference of the connected nodes. Based on the weights between the nodes with the connection relationship and the binary adjacency matrix, the weighted adjacency matrix is ​​constructed.

7. The method according to claim 5, characterized in that, Determining the weights between the connected nodes based on the voltage vector difference and current vector difference of the connected nodes includes: The magnitude of the ratio of the voltage vector difference to the current vector difference is determined as the electrical impedance distance between the connected nodes; The reciprocal of the electrical impedance distance between the connected nodes is used as the weight between the connected nodes.

8. A device for identifying an inbound topology node, characterized in that, The device includes: The determination module is used to determine the mutation nodes and affected weights in the current weighted topology map based on the differences between multiple historical weighted topology maps and the current weighted topology map corresponding to the data from the household side of the power grid. The construction module is used to construct a weighted adjacency matrix based on the mutated node, the weights between nodes with connections in the historical weighted topology graph, the affected weights, and the binary adjacency matrix of the historical weighted topology graph; the binary adjacency matrix is ​​obtained based on the communication handshake protocol data in the inbound data of the historical weighted topology graph; the communication handshake protocol data includes link layer address allocation information and routing table information; The inference module is used to input the weighted adjacency matrix and the original feature matrix into a preset inference model for inference to obtain a node embedding vector. The original feature matrix is ​​obtained based on the electrical feature data in the household-side data of the historical weighted topology. The electrical feature data includes current harmonic component data, voltage phase offset data, and power factor time series data. The node embedding vector includes the node's own electrical features, the node's connection relationship in the power grid, and the features and weights between neighboring nodes. The identification module is used to determine the identification information of each node in the current weighted topology graph based on the node embedding vector; the identification information includes at least one of the sub-region to which it belongs, identity label, and functional category.

9. 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 7.

10. 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 7.

Images