Power distribution network topology correction method fusing topology anomaly identification and structure reliable inference

Abstract

The application relates to a power distribution network topology correction method combining topology anomaly identification and structure reliable inference, and the steps comprise: identifying a topology structure abnormal area, identifying a candidate edge and an unobserved node; a structure behavior joint feature tensor considering the behavior cooperation between nodes, the adjacent structure similarity and the mutation synchronism is constructed, and is input into a topology connection reliability scoring model to obtain an edge reliability score; the edge reliability score obtained by the topology connection reliability scoring model is combined with the candidate edge and the unobserved node, the initial topology structure is reliably corrected, and a final topology structure is generated. Compared with the prior art, the application realizes comprehensive repair of the missing edge, the wrong edge and the unmodeled node in the original topology structure, and significantly improves the integrity and reliability of the power distribution network topology structure.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power distribution network topology identification, in particular to a power distribution network topology correction method fusing topology abnormality identification and structure reliable inference. BACKGROUND

[0002] With the large number of distributed energy access and the continuous improvement of power distribution automation level, the structure of power distribution network is becoming more and more complex, showing the characteristics of many points, long lines and flexible structure. The traditional topology identification method relying on drawings and manual checking has been difficult to meet the real-time and accuracy requirements. In recent years, the application of graph neural network (GNN) in topology identification has made certain progress, but still faces two key challenges: one is that part of the edge-node connection relationship is limited by imperfect collection device deployment, and there are data abnormalities or missing problems; the other is that there are a large number of transient disturbances and data fluctuations under complex working conditions, and the traditional identification method based on single time or static graph structure is easy to misjudge.

[0003] At the same time, in the process of power distribution network topology identification, the attributes and connection relationships of nodes (such as ring network cabinets, switch stations, etc.) are partially observable and partially unobservable. How to infer the unknown node relationship under the condition that part of the node information is known is the key to improve the accuracy and robustness of topology identification. Therefore, a new auxiliary verification method fusing time sequence information mining and graph structure propagation mechanism is urgently needed to effectively supplement and correct the results of the main identification model such as graph neural network. SUMMARY

[0004] The purpose of the present application is to provide a power distribution network topology correction method fusing topology abnormality identification and structure reliable inference.

[0005] The purpose of the present application can be realized by the following technical solutions:

[0006] A power distribution network topology correction method fusing topology abnormality identification and structure reliable inference, the method steps comprising:

[0007] Identify the abnormal area of topology structure, and generate candidate edges and unobserved nodes;

[0008] Construct a structure behavior joint feature tensor considering the behavior cooperation, adjacency structure similarity and mutation synchronicity between nodes, and input the topology connection reliability scoring model to obtain edge reliability score;

[0009] Combine the edge reliability score obtained by the topology connection reliability scoring model and the candidate edges and unobserved nodes to perform reliable correction on the initial topology structure, and generate the final topology structure.

[0010] As a preferred technical solution, the identification of the abnormal area of topology structure and the generation of candidate edges and unobserved nodes are as follows:

[0011] The power residual and voltage residual of each node in the calculation topology are calculated, the mean and standard deviation of the residual time sequence are calculated in a sliding window, and dynamic stability is judged;

[0012] The nodes whose mean and standard deviation of the residual in the sliding window interval are greater than the set threshold are determined as abnormal nodes, and are added to the abnormal node set;

[0013] The similarity matrix is constructed by using dynamic time warping distance for abnormal nodes, and the potential connection edge between two nodes is obtained by using spectral clustering algorithm;

[0014] The graph structure completion is modeled as a low-rank matrix completion problem, and the singular value thresholding algorithm is used for iterative solution to obtain the completed adjacency matrix;

[0015] The non-zero elements higher than the threshold in the completed adjacency matrix are determined as potential connections, i.e. candidate edges; if the nodes corresponding to a row or column of the completed adjacency matrix are all zero or approximately zero, the node is determined as a potential unobserved node.

[0016] As a preferred technical solution, the spectral clustering algorithm is used to obtain the potential connection edge between two nodes, which is specifically as follows:

[0017] The node behavior characteristics are constructed based on the node current, voltage and power factor, and the dynamic time warping distance between nodes is calculated based on the node behavior characteristics;

[0018] The similarity matrix is calculated based on the dynamic time warping distance between nodes, and a Laplacian matrix is further constructed;

[0019] The eigenvectors corresponding to the first k The smallest non-zero eigenvalue are solved to form a spectral embedding matrix;

[0020] Each row of the spectral embedding matrix is clustered, and nodes with similar characteristics are divided into the same cluster;

[0021] For any two nodes, if they belong to the same cluster but are not connected in the existing topology, it is determined that there may be a potential connection edge between the two nodes, and the candidate edge set is added.

[0022] As a preferred technical solution, the graph structure completion is modeled as a low-rank matrix completion problem, which is represented as:

[0023]

[0024] In the formula, is the original observed adjacency matrix; is the adjacency matrix to be completed; is the observation position projection operator; is the observation set; is a Frobenius norm; is a nuclear norm; is a regularizer.

[0025] As a preferred technical solution, the topological connection credibility scoring model models the connection credibility of the candidate edge by using a graph attention network: a linear transformation is used to project the structure-behavior joint feature tensor into a high-dimensional space to obtain an embedded feature representation of the edge; the embedded vector is input into an attention mechanism for scoring, and then the attention score is input into an activation function to obtain the final edge credibility.

[0026] As a preferred technical solution, the structure-behavior joint feature tensor specifically includes: current correlation, voltage correlation, mutation synchronization degree, mirror change rate, and adjacent coincidence rate between each pair of candidate nodes.

[0027] As a preferred technical solution, the mutation synchronization degree is calculated as follows:

[0028] For each node, a set of mutation events is constructed If the current or voltage change at any time t is greater than a set threshold, the time is added to the set of mutation events;

[0029] For two nodes, the sets of mutation events , Matching is performed, and if the first time , there is a second time such that the absolute value of the difference between the first time and the second time is less than or equal to the tolerance synchronization error, it is considered to be a pair of synchronous mutations, and the mutation synchronization degree is defined as:

[0030]

[0031] In the formula, denotes the set of mutation events of node i,j ; and is a constant used to prevent division by zero.

[0032] As a preferred technical solution, the mirror change rate is used to detect whether the current mutation of node i corresponds to a change in the opposite direction of the voltage of node j , and is calculated as follows:

[0033] The set of mutation times of the detection node is obtained, and for each mutation point, it is determined whether the current and voltage change directions of the node are opposite, and the number of times that satisfy the condition is counted and divided by the total number of mutations:

[0034]

[0035] In the formula, is a set of mutation moments of nodes i . and are respectively current and voltage change amounts of nodes i at moment t . is a sign function, that is, the positive and negative signs of the return value.

[0036] As a preferred technical solution, the edge credibility score obtained by combining the topology connection credibility scoring model is used to modify the credibility of the initial topology structure, specifically as follows:

[0037] Based on the scoring results of each candidate edge, the candidate edges with scoring results lower than the set scoring threshold are removed;

[0038] The nodes that are still completely isolated after edge modification are screened for structure matching prediction, and each unobserved node is combined with all nodes j in the original topology structure in the current topology. For each candidate edge , the topology connection credibility scoring model is used to calculate its scoring result, and the connection relationship is judged;

[0039] Based on the connection relationship judgment result, the edge structure is adjusted and completed to obtain the modified topology graph.

[0040] As a preferred technical solution, the connection relationship judgment is specifically as follows:

[0041] For each unobserved node, if the maximum candidate edge scoring result is less than the first scoring threshold, the unobserved node is not connected;

[0042] If there is a node such that the candidate edge scoring result is greater than or equal to the second scoring threshold, and the number of nodes satisfying the condition is less than the set number, then the nodes satisfying the condition are connected;

[0043] If there is a node such that the candidate edge scoring result is greater than or equal to the second scoring threshold, and the number of nodes satisfying the condition is greater than or equal to the set number, then the set number of nodes with the highest score are connected;

[0044] If there is no node with a candidate edge scoring result greater than or equal to the second scoring threshold, but there is a node such that the candidate edge scoring result is between the first scoring threshold and the second scoring threshold, then the node with the highest score is connected.

[0045] Compared with the prior art, the present application has the following beneficial effects:

[0046] 1) The application proposes a power distribution network topology correction method combining topology anomaly recognition and structure reliable reasoning. By introducing a score-driven edge reliability judgment mechanism and combining a structured access strategy for unobserved nodes, accurate correction and closed-loop completion of the power distribution network topology structure are achieved. While ensuring low computational resource consumption, the missing edges, incorrect edges and unmodeled nodes in the original topology structure are comprehensively repaired, significantly improving the integrity and reliability of the power distribution network topology structure.

[0047] 2) The application identifies suspected unobserved or unmodeled nodes in the main model through methods such as graph structure completion and spectral clustering. Based on the residual indicators of power and voltage, abnormal nodes are screened, a similarity matrix is constructed, and spectral clustering algorithm is used to find nodes with similar behavior but unconnected structure to mine possible implicit connections in abnormal areas. Finally, the topology structure of the reconstructed graph is modeled as a low-rank matrix completion problem, and the singular value thresholding algorithm is used for iterative solution, which can locate the anomaly from the power flow and voltage behavior and generate the set of edges to be completed and possible missed nodes. BRIEF DESCRIPTION OF DRAWINGS

[0048] Figure 1 The application is a topology auxiliary verification and correction method based on operation data driving process schematic diagram;

[0049] Figure 2 The application is a topology structure anomaly recognition and candidate information extraction process schematic diagram;

[0050] Figure 3 The application is a topology connection reliability score model construction process schematic diagram;

[0051] Figure 4 The application is a structure correction mechanism and topology output process schematic diagram. DETAILED DESCRIPTION

[0052] The application will be described in detail below in conjunction with the drawings and specific embodiments. The present embodiment is implemented on the basis of the technical solution of the application, and detailed implementation and specific operation process are given, but the protection scope of the application is not limited to the following examples.

[0053] Example 1

[0054] The application proposes a topology auxiliary verification and correction method based on operation data driving, which aims to find the structural omissions, connection errors and unobserved devices in the main recognition topology result of the power distribution network, and proposes reliable correction suggestions. As shown in the figure, the method consists of three steps: Figure 1

[0055] ​Step 1: Topology anomaly identification and candidate information extraction: Use power flow residual, voltage imbalance and behavior pattern inconsistency, combined with spectral clustering and graph completion algorithm, to locate suspected structural damage area, get candidate edges and unobserved nodes.

[0056] Step 2: Topology connection credibility scoring model construction: Construct behavior + structure joint feature tensor, use graph attention network (GAT) to output edge credibility score, quantify connection reliability.

[0057] Step 3: Topology structure correction mechanism and topology output: Based on the scoring and structure reasoning results, optimize node connection, output the corrected topology structure and credible graph atlas results.

[0058] Among them, the specific implementation process of each step is as follows:

[0059] Step 1: Topology structure anomaly identification and candidate information extraction. The goal of this step is to detect the inconsistent area with the topology structure in the operation data, locate the anomaly from the power flow and voltage behavior, and generate the set of edges to be completed and the possible missed nodes. The specific implementation process is as shown in Figure 2

[0060] 1.1) Multi-source residual consistency analysis

[0061] Construct residual index based on power and voltage, and use sliding statistical method to select abnormal nodes.

[0062] Where the power residual is calculated by:

[0063]

[0064] In the formula: is the active power residual of node i at time t ; is all adjacent nodes i that transmit power to ; k is the power transmitted by node i to node ; i is the power output by node j to adjacent node ; i is the power of local load of node

[0065] The voltage residual is calculated by:

[0066]

[0067] In the formula: is the voltage residual of node i at time t ​the voltage residual error of the node the voltage value of the node i the set of adjacent nodes of the node i the number of adjacent nodes.

[0068] Then, the mean and standard deviation of the residual error time series are calculated in a fixed-length sliding window to realize dynamic stability discrimination.

[0069] wherein, for the sliding window, the window length is set to , the current time is t , and the window interval is:

[0070]

[0071] The average residual error in the window interval is calculated to measure the overall power fluctuation level:

[0072]

[0073] The standard deviation is calculated, and the larger the standard deviation, the worse the stability:

[0074]

[0075] If the current time satisfies the following formula:

[0076]

[0077] wherein, is the average residual error threshold, is the fluctuation amplitude threshold.

[0078] then the node i is determined as an abnormal node, and is added to the abnormal node set , i.e.:

[0079]

[0080] The obtained abnormal node set will focus on completing their connections, finding and closely related but missed connections in subsequent steps, and will also focus on whether these abnormal nodes have become isolated nodes due to measurement problems.

[0081] 1.2) Behavior pattern clustering and spectral clustering

[0082] In order to further mine the implicit connections that may exist in the abnormal area, the behavior patterns between nodes need to be clustered. A similarity matrix constructed based on dynamic time warping (DTW) is adopted, and a spectral clustering algorithm is used to find nodes with similar behavior but unconnected structure.

[0083] ​​First, construct the node behavior features as follows:

[0084]

[0085] wherein: , , The distribution is the node current, voltage and power factor, T is the time length.

[0086] Then calculate the dynamic time warping (DTW) distance:

[0087]

[0088] wherein: is the time alignment path, is the vector length.

[0089] Calculate the similarity matrix :

[0090]

[0091] wherein: is the Gaussian kernel width coefficient.

[0092] Then construct the Laplacian matrix L :

[0093]

[0094] wherein: S is the similarity matrix, that is, ; D is the degree matrix, and is the diagonal matrix, ;

[0095] Spectral decomposition and eigenvector extraction are performed, and the following equation is solved to obtain:

[0096]

[0097] wherein, L is the Laplacian matrix; x is the eigenvector (corresponding to a dimension in the graph structure); is the eigenvalue, that is, the strength of the tension in this direction.

[0098] Solve the first k minimum non-zero eigenvalue corresponding to the eigenvector , to form the spectral embedding matrix:

[0099]

[0100] wherein each row represents the coordinates of the i-th node in the spectral space. i

[0101] Finally, the k-means algorithm is used to cluster each row of the spectral embedding matrix X and divide nodes with similar features into the same cluster.

[0102] For any two nodes i and j , if they belong to the same cluster but are not connected in the existing topology, it is determined that there may be a potential connection between the two nodes, and the candidate edge set is added.

[0103] The candidate edges output by spectral clustering are important inputs or prior information for low-rank matrix completion. Low-rank completion can optimize and improve the network structure more comprehensively and strictly based on these preliminary candidates.

[0104] 1.3)Low-rank matrix completion based on singular value thresholding

[0105] The goal of graph structure completion is to complete the missing connection edges in the adjacency matrix based on the partially known node connection relationships and similarities, and to reconstruct the topology of the graph.

[0106] This problem can be modeled as a low-rank matrix completion problem and solved iteratively using the singular value thresholding (SVT) algorithm.

[0107] Let the original observed adjacency matrix be , but some elements are not observable.

[0108] Define the observation set as , then construct the following optimization problem:

[0109]

[0110] In the formula, is the adjacency matrix to be completed; is the observation position projection operator; is the Frobenius norm; is the nuclear norm, which is the sum of all singular values, reflecting the rank of the matrix; is the regularization coefficient, .

[0111] The singular value thresholding (SVT) algorithm is used to solve the problem iteratively as follows:

[0112] 1.3.1)Algorithm initialization, set the initial matrix , set the threshold parameter, step size parameter, convergence tolerance, and maximum number of iterations K . ​​

[0113] 1.3.2) For each iteration k = 1, 2, 3,... K ; for matrix Perform singular value decomposition:

[0114]

[0115] where , is an orthogonal matrix; is a diagonal matrix of singular values, r is the effective rank, and each represents the strong principal component amplitude of the current graph structure.

[0116] Continue soft-thresholding operation for all singular values:

[0117]

[0118] where if , the singular value is set to 0 to suppress noise and redundant structure; is the sparse singular value matrix.

[0119] Reconstruct the adjacency matrix using the sparse singular values:

[0120]

[0121] where is the complete adjacency matrix estimate of the k + 1th iteration.

[0122] Introduce the reconstruction error into the next iteration:

[0123]

[0124] where is the error that preserves the observation positions; is the step size parameter.

[0125] If one of the following conditions is met, terminate the iteration:

[0126]

[0127] where is the convergence tolerance.

[0128] Finally, output the complete adjacency matrix , and set the non-zero elements in above the threshold to 0 If all the nodes in a row or column are zero or near zero, it is considered that the node is a potential unobserved node.

[0129] The candidate edge set obtained by low-rank matrix completion contains not only the possible connections discovered by spectral clustering method, but also the connections newly inferred by the completion algorithm according to global consistency.

[0130] In this step, for topology anomaly identification and candidate information extraction, first, abnormal nodes are found out through multi-source residual consistency analysis, and then through behavior pattern clustering and spectral clustering on these abnormal nodes and surrounding nodes, it can be found that some nodes are not connected together in the original network, but their power consumption patterns are similar (at this time, there may be a real but missed connection between them), the connection of these nodes is added to the candidate edge set, and finally, based on the singular value thresholding low-rank matrix completion, the candidate connections found by behavior pattern clustering and spectral clustering based on known connections and the similarity between nodes are filled in the algorithm iteration, which can integrate the information of the whole network to fill in the missing connection relationship, and even some previously unnoticed island nodes can be found.

[0131] Step 2: Topology connection credibility scoring model construction. This step aims to establish a credibility scoring function for each edge in the candidate edge set ( i,j ) to quantify its possibility as a real connection edge. The procedure of this step is shown in Figure 3 .

[0132] By constructing a structure and behavior joint feature tensor and introducing a graph attention mechanism (GAT), this model can comprehensively judge the behavior coordination, adjacent structure similarity and mutation synchronicity between nodes and output a score value. The higher the score value, the greater the possibility of the existence of the connection edge in the real topology structure.

[0133] 2.1) For each pair of candidate nodes , construct a five-dimensional feature:

[0134]

[0135] The calculation methods of each index are as follows:

[0136] is the current correlation, that is, the correlation coefficient of the current between nodes i,j .

[0137]

[0138] In the formula, and are the current time series between nodes i,j .Cov Covariance; Standard deviation.

[0139] Voltage correlation, i.e. the correlation coefficient of the voltage between nodes i,j .

[0140]

[0141] where, and are the voltage time series between nodes i,j .

[0142] Mutation synchronization degree is calculated as follows:

[0143] For each node, construct a mutation event set , if at any time t , it satisfies:

[0144]

[0145] then add the time to the mutation event set , i.e.

[0146]

[0147] For the mutation event set i,j of node , perform matching: if , there exists such that ( is the tolerance synchronization error) then it is considered as a pair of synchronous mutations, and the mutation synchronization degree is defined as:

[0148]

[0149] where, is used to prevent division by zero.

[0150] Mirror change rate is used to detect whether the current mutation of node i corresponds to the opposite change of the voltage direction of node j , and is calculated as follows:

[0151] Get the mutation time set i of the detection node ;

[0152] At each mutation point, determine whether the following formula is satisfied:

[0153]

[0154] where, and are the node i At time t , the current and voltage change, is a sign function, which returns the sign of the value.

[0155] Finally, the number of times that meet the condition is divided by the total number of mutations:

[0156]

[0157] where, is an indicator function, which takes 1 when the condition is met, and 0 otherwise, is the number of parameters in the mutation time set.

[0158] is the adjacent coincidence rate, which is obtained by calculating the intersection and union of the adjacent sets of two nodes.

[0159]

[0160] where, and are the adjacent node sets of nodes i and j, respectively.

[0161] 2.2) Connection credibility score model construction

[0162] This scoring mechanism uses a lightweight graph attention network (GAT) to model the connection credibility of candidate edges ( i,j ). The core process includes feature mapping, attention calculation, weight normalization, and final score output.

[0163] (2.2.1) Feature mapping

[0164] Use linear transformation to project low-dimensional input features into high-dimensional space to enhance the nonlinear interaction ability between different scoring factors:

[0165]

[0166] where, is a learnable parameter matrix; is the embedding feature representation of edge .

[0167] (2.2.2) Attention score calculation

[0168] Input the embedding vector into the attention mechanism for scoring:

[0169]

[0170] where, is the learnable attention weight vector; is the nonlinear activation function, denotes the original attention value of edge .

[0171] (2.2.3) Final score output

[0172] The attention score is sent into a sigmoid activation function to compress it into the interval [0, 1] to get the final edge confidence:

[0173]

[0174] where, is the optional edge score output layer; is the sigmoid function, is the probability of the existence of the candidate edge.

[0175] Step 3: Structure correction mechanism and topology output The purpose of this step is to modify the initial topology structure based on the candidate structure, combine the output results of the scoring model with the potential edges and unknown node information identified in the graph structure completion process, and generate a complete and reliable output topology structure. The process is shown in Figure 4 :

[0176] 3.1) Input information

[0177] Candidate edge set: = { (from step 1 structure completion process)}; Unobserved node (candidate node) set:

[0178] { (from step 1 structure completion process)}; = Edge score result of each edge: (from step 2 scoring result);

[0179] Original topology graph structure: .

[0180]

[0181] 3.2) Edge correction mechanism

[0182] According to the scoring result of each candidate edge obtained in step 2 , the following strategy is used.

[0183] When , it is determined to be a real edge, and the completion is performed. If , it is determined to be a false edge, and the completion is not performed.​If so, the candidate edge is determined, and manual review or further verification is performed. If If so, the edge is determined to be non-existent, and the elimination process is performed.

[0184] 3.3) Unobserved node connection mechanism

[0185] In step 1, a batch of suspected unobserved or unmodeled nodes in the main model are identified through methods such as graph structure completion and spectral clustering. These nodes have obvious power and voltage activity in their behavior time series data, but are not connected to any known nodes. Therefore, in the structure modification stage, their connection edges need to be supplemented to achieve the structural closure of the graph.

[0186] (3.3.1) First, filter the nodes that are completely isolated after the edge modification. These nodes are not connected in the candidate edge completion and must be structure matching predicted.

[0187] (3.3.2) For each unobserved node Combine all in the current topology, establish a candidate edge set, and for each candidate edge , use the scoring method in step 2 to calculate its score result .

[0188] Based on the score results of each candidate edge, construct the score table as follows:

[0189]

[0190] Determine the connection relationship based on the mechanism shown in the following table.

[0191] Table 1 Connection relationship judgment table

[0192]

[0193] 3.4) Structure graph update and result output

[0194] According to the above process, adjust the edge structure and complete the final output of the modified topology graph:

[0195]

[0196]

[0197]

[0198] In the formula, is the final output of the power distribution network topology graph, including all nodes and the modified connection relationship; is the node set of the final topology graph, including the original nodes and the identified unobserved nodes; A set of edges of the final topology graph, including original reserved edges, completed edges and predicted access edges; A set of nodes that have been modeled in the original topology structure; A set of unobserved nodes identified in the structure completion process; A set of all edges (connection relationships) existing in the original topology graph; A set of edges that are determined to be low in credibility and should be removed in the scoring process; A set of edges that are determined to be completed in the candidate edge scoring process with a score higher than the credibility threshold; A set of access edges generated by predicting unobserved nodes to ensure structural connectivity.

[0199] The method of the present application realizes accurate correction and closed-loop completion of the power distribution network topology structure by introducing a scoring-driven edge credibility judgment mechanism and combining a structured access strategy for unobserved nodes. Compared with existing topology identification models, the present application has obvious improvements in multiple key indicators, as shown in the following table.

[0200] Table 2 Comparison of the present application scheme and existing topology identification model identification structure

[0201]

[0202] As can be seen from the results, the present application ensures low computational resource consumption while achieving comprehensive repair of missing edges, incorrect edges and unmodeled nodes in the original topology structure, significantly improving the integrity and credibility of the power distribution network topology structure. The method has high deployability and can be integrated into existing power distribution network topology identification systems as an auxiliary verification and correction module, and is particularly suitable for substation structure modeling optimization tasks for automation operation scenarios.

[0203] Embodiment 2

[0204] As a second aspect of the present application, the present application also provides an electronic device comprising: one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the power distribution network topology correction method as described above. In addition to the above-mentioned processors, memories and interfaces, any device with data processing capability in the embodiment usually includes other hardware according to the actual function of the device with data processing capability, and will not be described here.

[0205] Embodiment 3

[0206] As a third aspect of the present application, the present application also provides a computer readable storage medium having stored thereon computer instructions which, when executed by a processor, implement the power grid topology correction method of fusing topology anomaly identification and structure trusted reasoning as described above. The computer readable storage medium can be an internal storage unit of any data processing capable device, such as a hard disk or a memory, of any of the aforementioned embodiments. The computer readable storage medium can also be an external storage device, such as a plug-in hard disk, a Smart Media Card (SMC), an SD card, a Flash Card, etc. equipped on the device. Further, the computer readable storage medium can include both the internal storage unit of any data processing capable device and the external storage device. The computer readable storage medium is used to store the computer program and other programs and data required by the data processing capable device, and can also be used to temporarily store data that has been output or will be output.

[0207] The above functions, if implemented in the form of software function units and sold or used as independent products, can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application, in essence or the part that contributes to the prior art, or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0208] The preferred embodiments of the present application are described in detail above. It should be understood that those skilled in the art can make many modifications and changes without creative labor based on the concept of the present application. Therefore, any technical solution that can be obtained by logical analysis, reasoning or limited experiments by those skilled in the art based on the prior art and the concept of the present application shall be within the protection scope determined by the claims.

Claims

1. A method for topology correction in distribution networks that integrates topology anomaly identification and structural reliability reasoning, characterized in that, The method steps include: Identify regions with abnormal topological structures, and identify and generate candidate edges and unobserved nodes; A joint feature tensor of structural behavior considering the behavioral coordination, adjacency structure similarity and mutation synchronization between nodes is constructed and input into the topological connection credibility scoring model to obtain the edge credibility score; the joint feature tensor of structural behavior specifically includes: current correlation, voltage correlation, mutation synchronization, mirror change rate and adjacency overlap rate between each pair of candidate nodes. The mutation synchronization degree is calculated as follows: a set of mutation events is constructed for each node. If at any time t the change in current or voltage is greater than a set threshold, then that time is added to the set of sudden events. The set of mutation events for two nodes , Execute the match, if in the first moment There exists a second moment. Make the first moment With the second moment If the absolute value of the difference is less than or equal to the tolerance synchronization error, then they are considered a pair of synchronization mutations. The mutation synchronization degree is defined as: In the formula, Represents a node i,j The set of mutation events; The constant is used to prevent division by zero; The mirror change rate is used to detect nodes. i Does the current surge correspond to a node? j The voltage changes in opposite directions, and the calculation method is as follows: Obtain the set of abrupt change moments for the detection nodes. For each abrupt change moment, determine whether the current and voltage changes in the node are in opposite directions, and count the number of times this condition is met, dividing by the total number of abrupt changes. In the formula, For nodes i The set of abrupt change moments; and They are nodes i At any moment t The changes in current and voltage; This is a sign function, meaning the return value is positive or negative. By combining the edge credibility score obtained from the topology connection credibility scoring model with the candidate edges and unobserved nodes, the initial topology structure is credibly corrected to generate the final topology structure.

2. The distribution network topology correction method integrating topology anomaly identification and structural reliability reasoning as described in claim 1, characterized in that, The process of identifying abnormal regions in the topological structure and generating candidate edges and unobserved nodes is as follows: Calculate the power residual and voltage residual of each node in the topology, perform mean and standard deviation statistics on the residual time series using a sliding window, and determine dynamic stability. Nodes whose mean and standard deviation of residuals within the sliding window interval are both greater than a set threshold are identified as abnormal nodes and added to the abnormal node set. A similarity matrix is ​​constructed for abnormal nodes using dynamic time-normalized distance, and potential connection edges between two nodes are obtained through spectral clustering algorithm. The graph structure completion problem is modeled as a low-rank matrix completion problem, and the singular value thresholding algorithm is used for iterative solution to obtain the completed adjacency matrix. Non-zero elements above a threshold in the completed adjacency matrix are considered to have potential connections, i.e., candidate edges. If all nodes in a row or column of the completed adjacency matrix are zero or their absolute values ​​are less than a set threshold, then that node is considered a potential unobserved node.

3. The distribution network topology correction method integrating topology anomaly identification and structural reliability reasoning according to claim 2, characterized in that, The process of obtaining potential connection edges between two nodes using spectral clustering algorithm is as follows: Node behavior characteristics are constructed based on node current, voltage, and power factor, and the dynamic time-warped distance between nodes is calculated based on the node behavior characteristics. The similarity matrix is ​​calculated based on the dynamic time-normalized distance between nodes, and the Laplacian matrix is ​​further constructed. Before solving the Laplace matrix k The eigenvectors corresponding to the smallest non-zero eigenvalues ​​constitute the spectral embedding matrix; Cluster each row of the spectral embedding matrix, grouping nodes with similar features into the same cluster; For any two nodes, if they belong to the same cluster but are not connected in the existing topology, then it is determined that there may be a potential connection edge between the two nodes, and they are added to the candidate edge set.

4. The distribution network topology correction method integrating topology anomaly identification and structural reliability reasoning according to claim 2, characterized in that, The method of modeling graph structure completion as a low-rank matrix completion problem is expressed as follows: In the formula, This is the original observation adjacency matrix; The adjacency matrix to be completed; Projection operator for observation position; For the observation set; It is the Frobenius norm; For nuclear norm; is the regularization coefficient.

5. The distribution network topology correction method integrating topology anomaly identification and structural reliability reasoning according to claim 1, characterized in that, The topological connection credibility scoring model uses a graph attention network to model the connection credibility of candidate edges: a linear transformation is used to project the joint feature tensor of structural behavior onto a high-dimensional space to obtain the embedded feature representation of the edge; The embedded vector is input into the attention mechanism for scoring, and then the attention score is input into the activation function to obtain the final edge credibility.

6. The distribution network topology correction method integrating topology anomaly identification and structural reliability reasoning according to claim 1, characterized in that, The edge credibility score obtained by combining the topology connection credibility scoring model with candidate edges and unobserved nodes is used to perform credibility correction on the initial topology structure, as follows: Based on the score of each candidate edge, edge correction is performed, and candidate edges with scores below a set score threshold are removed. For nodes that remain completely isolated after edge correction, perform structure matching prediction and identify each unobserved node. With all the nodes in the original topology of the current topology j Combined, for each candidate edge The scoring results are calculated using a topology connectivity credibility scoring model, and the connectivity relationships are determined. Based on the results of the connection relationship judgment, the edge structure is adjusted and completed to obtain the corrected topology graph.

7. The distribution network topology correction method integrating topology anomaly identification and structural reliability reasoning according to claim 6, characterized in that, The connection relationship determination is as follows: For each unobserved node, if the score of the largest candidate edge is less than the first score threshold, then the unobserved node is not connected. If there exists a node such that the candidate edge score is greater than or equal to the second score threshold, and the number of nodes that meet the condition is less than the set number, then connect the nodes that meet the condition. If there exists a node such that the candidate edge score is greater than or equal to the second score threshold, and the number of nodes that meet the condition is greater than or equal to the set number, then connect the set number of nodes with the highest scores. If there are no nodes whose candidate edge scores are greater than or equal to the second score threshold, but there are nodes whose candidate edge scores are between the first and second score thresholds, then connect the node with the highest score.

Images