Power distribution network fault positioning method and device based on graph convolutional neural network

Abstract

The invention discloses a power distribution network fault positioning method and equipment based on a graph convolutional neural network, belongs to the technical field of power distribution network fault positioning, and is used for solving the technical problems that a traditional fault positioning method depends on feature information of a single node, neglects the overall structure of a power distribution network and is difficult to deal with a complex and changeable power distribution network. The method comprises the steps of performing data preprocessing on a collected data set to obtain a target data set; constructing an optimization matrix based on the target data set and the topological structure diagram of the to-be-tested power distribution network; performing node feature extraction on the topological structure diagram to obtain a node feature vector; training a graph convolutional neural network through the optimization matrix and the node feature vectors, and performing sample balance adjustment to obtain a power distribution network fault positioning model; and inputting the to-be-detected data into the power distribution network fault positioning model for identification to obtain state parameters of each node, and positioning a fault node according to the node state parameters. According to the invention, the accuracy and robustness of power distribution network fault positioning are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network fault location, and in particular to a distribution network fault location method and device based on a graph convolutional neural network. Background Art

[0002] For a long time, the construction of power systems has been plagued by the problem of "focusing on power sources and neglecting power grids, focusing on transmission and neglecting distribution networks", resulting in relatively lagging construction of distribution networks, which has affected the reliability and safety of power supply to a certain extent.

[0003] In the past, fault location in distribution networks was determined through a series of data analysis to determine the specific location of the fault. However, with the continuous upgrading of electrical equipment, the structure of the distribution network has become increasingly complex. The widespread use of intelligent equipment and high-power devices has caused mutual interference between devices, which in turn caused data information distortion. Traditional distribution network fault location methods mainly rely on the characteristic information of a single node for analysis, ignoring the overall structure of the distribution network. This method can still work when the data collection is complete and accurate. Once the collected data is disturbed and distorted, the accuracy of fault location will be greatly reduced. Therefore, traditional fault location methods are difficult to cope with complex and changeable distribution networks. Summary of the invention

[0004] The embodiments of the present invention provide a distribution network fault location method and device based on a graph convolutional neural network, which are used to solve the following technical problems: the traditional fault location method relies on the characteristic information of a single node, ignores the overall structure of the distribution network, and is difficult to cope with complex and changeable distribution networks.

[0005] The embodiment of the present invention adopts the following technical solutions: On the one hand, an embodiment of the present invention provides a distribution network fault location method based on a graph convolutional neural network, the method comprising: performing data preprocessing on a collected data set to obtain a target data set; Based on the target data set and the topological structure diagram of the distribution network to be tested, constructing an optimization matrix, specifically including: obtaining each power device in the distribution network to be tested and their connection relationship, obtaining the topological structure of the distribution network to be tested; mapping each power device as a node, mapping the connection relationship as an edge, and constructing the topological structure diagram of the distribution network to be tested; generating a topological matrix according to the topological structure diagram; analyzing the target data set to construct a node relationship matrix; constructing the optimization matrix based on the topological matrix and the node relationship matrix, specifically including: Based on the fault data in the target data set, count the node types that fail simultaneously in all fault events; calculate the concurrent fault probability of any two nodes failing simultaneously based on the statistical data; construct a node relationship matrix based on the concurrent fault probability; perform a point multiplication operation on the node relationship matrix and the topology matrix to obtain the optimization matrix; Extracting node features from the topological structure graph to obtain a node feature vector; The graph convolutional neural network is trained by using the optimization matrix and the node feature vector, and sample balance adjustment is performed to obtain a distribution network fault location model; The data to be tested is input into the distribution network fault location model for identification, the state parameters of each node are obtained, and the faulty node is located according to the node state parameters.

[0006] In a feasible implementation manner, data preprocessing is performed on the collected data set to obtain a target data set, specifically including: Obtain historical operation data of each node in the distribution network to be tested, extract key data from the historical operation data, and eliminate irrelevant data to form the data set; wherein the key data includes at least one or more of the following: three-phase voltage, current amplitude and phase angle; Perform data cleaning on the data set to identify missing values, outliers and duplicate values ​​in the data set and remove them; The cleaned data set is normalized and the data is mapped into a preset range to obtain the target data set.

[0007] In a feasible implementation manner, performing node feature extraction on the topological structure graph to obtain a node feature vector specifically includes: In the target data set, obtaining the operating parameters of each node in the topological structure diagram; according to , extract features from the operating parameters to obtain an overall feature vector; wherein y is the output feature value, x is the input parameter, W represents the initial weight matrix, and b is the initial bias matrix; wherein the dimension of W is (n-1)×n, and the dimension of b is n×1; the overall feature vector is an n×n dimensional matrix, and n is the total number of nodes in the topological structure diagram; The overall feature vector is divided into n one-dimensional matrices to obtain the node feature of each node; the node features are concatenated to obtain the node feature vector.

[0008] In a feasible implementation manner, a graph convolutional neural network is trained by using the optimization matrix and the node feature vector, and sample balance adjustment is performed to obtain a distribution network fault location model, which specifically includes: Using the optimized matrix as the adjacency matrix of the graph convolutional neural network and the node feature vectors as training samples to train the graph convolutional neural network; The graph convolutional neural network is subjected to sample balance adjustment through the Focal Loss loss function to optimize the graph convolutional neural network and obtain the distribution network fault location model.

[0009] In a feasible implementation, the formula of the graph convolutional neural network is: ; in, is the node state feature output by the graph convolutional neural network, is the input of the graph convolutional neural network, that is, the node feature vector, is the adjacency matrix; and are the parameters of the graph convolutional neural network, and ReLU is the activation function.

[0010] In a feasible implementation, the graph convolutional neural network is subjected to sample balance adjustment by using the Focal Loss function, specifically including: based on , calculate the Focal Loss value after each training; in, represents the weight factor, and , when the detection value is a positive sample, the weight factor is , when the detection value is a negative sample, the weight factor is ; represents the adjustment factor, R represents the hyperparameter of the adjustment factor, and , the larger the R, the greater the adjustment strength; According to the Focal Loss loss value, sample balance adjustment is performed on the graph convolutional neural network.

[0011] In a feasible implementation manner, the data to be tested is input into the distribution network fault location model to identify the state parameters of each node, and the faulty node is located according to the node state parameters, specifically including: Collecting the data to be tested in real time in the power distribution network to be tested; The data to be tested corresponding to each node of the topological structure diagram is cyclically input into the distribution network fault location model, and the node state parameters of each node are identified by the distribution network fault location model; Determine the node where the fault occurs according to the difference between the node status parameter and the normal parameter; Generate an early warning message and send the failed node to the recipient's terminal.

[0012] On the other hand, an embodiment of the present invention further provides a distribution network fault location device based on a graph convolutional neural network, the device comprising: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor so that the at least one processor can execute the distribution network fault location method based on graph convolutional neural network.

[0013] Compared with the prior art, the distribution network fault location method and device based on graph convolutional neural network provided by the embodiment of the present invention have the following beneficial effects: The distribution network fault location method based on graph convolutional neural network proposed in the present invention takes graph convolutional neural network as the core, and uses its powerful information aggregation ability to deeply mine the topological structure information of the distribution network. By fusing the characteristic information of the neighborhood nodes, the faulty nodes can be effectively discovered and located, thereby significantly improving the accuracy of fault location. Compared with traditional fault location methods, the distribution network fault location technology based on graph convolutional neural network makes full use of the spatial structure information of the distribution network and avoids the limitation of relying on the characteristic information of a single node. By combining with the characteristic information of the neighborhood nodes, the method based on graph convolutional neural network can effectively deal with the distortion problem of data information by deeply mining the topological structure information of the distribution network, thereby improving the accuracy and robustness of fault location. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings: Figure 1 A flow chart of a distribution network fault location method based on a graph convolutional neural network provided by an embodiment of the present invention; Figure 2 A schematic diagram of a distribution network topology structure diagram provided by an embodiment of the present invention; Figure 3 An example diagram of a relationship matrix of concurrent fault probabilities provided by an embodiment of the present invention; Figure 4 A schematic diagram of a distribution network fault location model architecture provided by an embodiment of the present invention; Figure 5 A schematic diagram of the structure of a distribution network fault location device based on a graph convolutional neural network provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0016] The embodiment of the present invention provides a distribution network fault location method based on graph convolutional neural network, such as Figure 1 As shown, the distribution network fault location method based on graph convolutional neural network specifically includes steps S101-S105: S101 . Preprocess the collected data set to obtain a target data set.

[0017] Specifically, firstly, the historical operation data of each node in the distribution network to be tested is obtained, and the key data is extracted from the historical operation data, and irrelevant data is eliminated to form a data set; wherein the key data includes at least one or more of the following: three-phase voltage, current amplitude and phase angle.

[0018] Furthermore, the data set is cleaned to identify missing values, outliers and duplicate values ​​in the data set and remove them. The cleaned data set is normalized to map the data into a preset range to obtain the target data set.

[0019] The present invention first performs data preprocessing on the data actually obtained. The required key information is extracted from the data, including three-phase voltage, current amplitude, phase angle, etc., and irrelevant data is eliminated. Then data cleaning is performed to identify missing values, abnormal values ​​and duplicate values ​​for elimination. Finally, the data is normalized and the data is uniformly mapped to a specific range to eliminate the dimensional differences between different features, ensure that each feature has the same weight in the model calculation, reduce the influence of singular values, and ensure that the convergence speed of subsequent model training is accelerated.

[0020] S102: construct an optimization matrix based on the target data set and the topological structure diagram of the distribution network to be tested.

[0021] Specifically, each power device in the distribution network to be tested and its connection relationship are obtained to obtain the topological structure of the distribution network to be tested.

[0022] Furthermore, each power device is mapped as a node, and the connection relationship is mapped as an edge to construct a topological structure diagram of the distribution network to be tested. The topological structure diagram is then represented in the form of a matrix, which is called a dependency matrix, i.e., a topological matrix.

[0023] As a feasible implementation method, Figure 2 A schematic diagram of a distribution network topology structure diagram provided by an embodiment of the present invention is as follows: Figure 2 As shown, the nodes with numbers identify the various power devices in the distribution network, and the edges represent the connection relationship between the power devices.

[0024] Furthermore, because the relative importance of each weight in the undirected graph of the distribution network is different, directly using the dependency matrix generated by the undirected graph will lead to inaccurate results. Therefore, the present invention analyzes the target data set and constructs a node relationship matrix; and constructs an optimization matrix based on the topological matrix and the node relationship matrix.

[0025] As a feasible implementation method, the specific process of constructing the optimization matrix is ​​as follows: based on the data in the target data set, the node types that have problems at the same time in all fault events are counted. Then, based on the counted node types, the probability of simultaneous failure of any node combination is calculated. This probability value reflects the correlation between nodes in the occurrence of failures. Then, based on the probability of concurrent failures, a node relationship matrix is ​​constructed, and finally the node relationship matrix is ​​multiplied element by element with the topology matrix, that is, the dot product operation, to obtain the optimization matrix.

[0026] In one embodiment, the present invention counts which nodes have problems at the same time in all the failure events in the target data set. Then, based on these statistical data, the probability of any two node combinations failing at the same time is further calculated. This probability value reflects the correlation between the nodes in the occurrence of failures. Subsequently, using these calculated probability values, a new relationship matrix is ​​constructed, such as Figure 3 As shown in the figure, the values ​​in the relationship matrix are the probabilities that the two nodes corresponding to the row and column fail at the same time. For example, the first element in the first row represents the probability that node 1 and node 1 fail at the same time. Since they are the same node, the probability is 1. Similarly, the values ​​on the diagonal are all 1. The second element in the first row represents the probability that when node 1 fails, node 2 also fails at the same time, which is 0.618. It should be noted here that the first element in the second row represents the probability that when node 2 fails, node 1 also fails. Due to the different dependencies between nodes, the fault propagation mode is also different. Therefore, although they are both probabilities that node 1 and node 2 fail at the same time, the values ​​of the elements are different due to the different order relationships.

[0027] This relationship matrix is ​​then multiplied element-by-element (i.e., dot multiplication in mathematical operations) with the dependency matrix of the undirected graph to generate an optimized matrix. This optimized matrix not only captures the direct dependencies between nodes, but also incorporates the probability information of their concurrent failures. This optimized matrix is ​​used as the adjacency matrix of the subsequent graph convolutional neural network to improve the accuracy and performance of the model.

[0028] S103: Extract node features from the topological structure graph to obtain a node feature vector.

[0029] Specifically, Figure 4 A schematic diagram of a distribution network fault location model architecture provided by an embodiment of the present invention is shown in FIG. Figure 4 As shown in the figure, the distribution network fault location model includes a backbone network for feature extraction, a graph convolution module and a fully connected layer Dense module.

[0030] After constructing the topology diagram, in the target data set, the operating parameters of each node in the topology diagram are obtained and input into the backbone network for feature extraction. The extraction process is as follows: according to , extract features from the operating parameters to obtain the overall feature vector; where y is the output eigenvalue, x is the input parameter, W represents the initial weight matrix, and b is the initial bias matrix; where the dimension of W is (n-1)×n, and the dimension of b is n×1; the overall feature vector is an n×n dimensional matrix, where n is the total number of nodes in the topological structure diagram.

[0031] The overall feature vector is divided into n one-dimensional matrices to obtain the node features of each node; the node features are concatenated to obtain the node feature vector.

[0032] In one embodiment, if the current initial weight matrix is ​​W0, the size is 12×11, and the current initial bias matrix is ​​b0, the size is 1×12. W0 is flipped, and its feature size becomes 11×12, and then it is spliced ​​with b0 to obtain a spliced ​​vector S of size 12×12. Then S is divided into S1, S2, ... S12, each with a feature size of 12×1, which is used to correspond to 12-dimensional features such as the amplitude and phase angle of the three-phase voltage and current.

[0033] S104, training the graph convolutional neural network by optimizing the matrix and the node feature vector, and performing sample balance adjustment to obtain a distribution network fault location model.

[0034] Specifically, the optimized matrix is ​​used as the adjacency matrix of the graph convolutional neural network, and the node feature vector is used as a training sample to train the graph convolutional neural network. The adjacency matrix is ​​a matrix form used in graph theory to represent the connection relationship between nodes in a graph. Each element of the adjacency matrix represents the relationship between the corresponding node pairs. This matrix form has the characteristics of simple structure and easy calculation, and can be used for the calculation of graph convolutional neural networks. The graph convolutional neural network can perform convolution operations on the nodes of the graph to extract the structural features of the graph. By learning the connection relationship in the adjacency matrix, potential patterns and information can be mined.

[0035] Furthermore, the Focal Loss loss function is used to adjust the sample balance of the graph convolutional neural network to optimize the graph convolutional neural network and obtain the distribution network fault location model.

[0036] The formula of graph convolutional neural network is: ;in, is the node state feature output by the graph convolutional neural network, is the input of the graph convolutional neural network, i.e., the node feature vector, is the adjacency matrix; and is the parameter of graph convolutional neural network, and ReLU is the activation function.

[0037] Furthermore, judging whether a node is faulty requires two data samples: faulty samples and non-faulty samples. However, the sample type ratio in the data set is generally unbalanced, with the number of non-faulty samples far exceeding the number of faulty samples. Since the majority class samples account for a high proportion of the total samples, the minority class samples are easily ignored, and the trained model is more biased towards the majority class. The proportion of faulty samples learned by the model decreases, resulting in a decrease in the performance of the model. In order to solve this problem, the present invention introduces Focal Loss as a loss function.

[0038] Through the Focal Loss function, the sample balance of the graph convolutional neural network is adjusted, including: based on , calculate the Focal Loss value after each training. Among them, represents the weight factor, and , when the detection value is a positive sample, the weight factor is , when the detection value is a negative sample, the weight factor is ; represents the adjustment factor, R represents the hyperparameter of the adjustment factor, and , the larger R is, the greater the adjustment is. Then, according to the Focal Loss value, the sample balance of the graph convolutional neural network is adjusted.

[0039] The introduction of the adjustment factor can adjust the contribution of the sample to the loss according to the difference between the predicted result and the true label. The larger the value, the higher the confidence of the classification, which reduces the loss weight, and vice versa, increases the loss weight, making the loss function tend to be difficult to distinguish samples. The larger the R, the greater the adjustment.

[0040] S105 , inputting the data to be tested into the distribution network fault location model for identification, obtaining the state parameters of each node, and locating the faulty node according to the node state parameters.

[0041] Specifically, the data to be tested is collected in real time in the distribution network to be tested, the data to be tested corresponding to each node of the topological structure diagram is cyclically input into the distribution network fault location model, and the node state parameters of each node are identified through the distribution network fault location model.

[0042] Furthermore, according to the difference between the node status parameters and the normal parameters, the faulty node is determined, and an early warning message is generated, and the faulty node information is sent to the receiver's terminal so that corresponding measures can be taken in time to ensure the safe operation of the distribution network.

[0043] In addition, the embodiment of the present invention also provides a distribution network fault location device based on a graph convolutional neural network, such as Figure 5 As shown in the figure, the distribution network fault location device based on graph convolutional neural network specifically includes: at least one processor; and a memory in communication with the at least one processor; wherein, The memory stores instructions executable by at least one processor to enable the at least one processor to perform: Perform data preprocessing on the collected data set to obtain the target data set; Based on the target data set and the topological structure diagram of the distribution network to be tested, construct an optimization matrix; Extracting node features from the topological structure graph to obtain a node feature vector; The graph convolutional neural network is trained by using the optimization matrix and the node feature vector, and sample balance adjustment is performed to obtain a distribution network fault location model; The data to be tested is input into the distribution network fault location model to identify the state parameters of each node, and the faulty node is located according to the node state parameters.

[0044] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0045] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0046] Each embodiment of the present invention is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0047] The above describes specific embodiments of the present invention. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0048] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the embodiments of the present invention may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the embodiments of the present invention should be included in the protection scope of the present invention.

Claims

1. A distribution network fault location method based on graph convolutional neural network, characterized in that: The method comprises: Perform data preprocessing on the collected data set to obtain the target data set; Based on the target data set and the topological structure diagram of the distribution network to be tested, constructing an optimization matrix, specifically including: obtaining each power device in the distribution network to be tested and their connection relationship, obtaining the topological structure of the distribution network to be tested; mapping each power device as a node, mapping the connection relationship as an edge, and constructing the topological structure diagram of the distribution network to be tested; generating a topological matrix according to the topological structure diagram; analyzing the target data set to construct a node relationship matrix; constructing the optimization matrix based on the topological matrix and the node relationship matrix, specifically including: Based on the fault data in the target data set, count the node types that fail simultaneously in all fault events; calculate the concurrent fault probability of any two nodes failing simultaneously based on the statistical data; construct a node relationship matrix based on the concurrent fault probability; perform a point multiplication operation on the node relationship matrix and the topology matrix to obtain the optimization matrix; Extracting node features from the topological structure graph to obtain a node feature vector; The graph convolutional neural network is trained by using the optimization matrix and the node feature vector, and sample balance adjustment is performed to obtain a distribution network fault location model; The data to be tested is input into the distribution network fault location model for identification, the state parameters of each node are obtained, and the faulty node is located according to the node state parameters.

2. A distribution network fault location method based on graph convolutional neural network according to claim 1, characterized in that: The collected data set is preprocessed to obtain the target data set, including: Obtain historical operation data of each node in the distribution network to be tested, extract key data from the historical operation data, and eliminate irrelevant data to form the data set; wherein the key data includes at least one or more of the following: three-phase voltage, current amplitude and phase angle; Perform data cleaning on the data set to identify missing values, outliers and duplicate values ​​in the data set and remove them; The cleaned data set is normalized and the data is mapped into a preset range to obtain the target data set.

3. A distribution network fault location method based on graph convolutional neural network according to claim 1, characterized in that: Node feature extraction is performed on the topological structure graph to obtain a node feature vector, specifically including: In the target data set, obtaining the operating parameters of each node in the topological structure diagram; according to , extract features from the operating parameters to obtain an overall feature vector; wherein y is the output feature value, x is the input parameter, W represents the initial weight matrix, and b is the initial bias matrix; wherein the dimension of W is (n-1)×n, and the dimension of b is n×1; the overall feature vector is an n×n dimensional matrix, and n is the total number of nodes in the topological structure diagram; The overall feature vector is divided into n one-dimensional matrices to obtain the node feature of each node; the node features are concatenated to obtain the node feature vector.

4. A distribution network fault location method based on graph convolutional neural network according to claim 1, characterized in that: The graph convolutional neural network is trained by the optimization matrix and the node feature vector, and sample balance adjustment is performed to obtain a distribution network fault location model, which specifically includes: Using the optimized matrix as the adjacency matrix of the graph convolutional neural network and the node feature vectors as training samples to train the graph convolutional neural network; The graph convolutional neural network is subjected to sample balance adjustment through the Focal Loss loss function to optimize the graph convolutional neural network and obtain the distribution network fault location model.

5. A distribution network fault location method based on graph convolutional neural network according to claim 4, characterized in that: The formula of the graph convolutional neural network is: ; in, is the node state feature output by the graph convolutional neural network, is the input of the graph convolutional neural network, that is, the node feature vector, is the adjacency matrix; and are the parameters of the graph convolutional neural network, and ReLU is the activation function.

6. A distribution network fault location method based on graph convolutional neural network according to claim 4, characterized in that: The Focal Loss function is used to adjust the sample balance of the graph convolutional neural network, specifically including: based on , calculate the Focal Loss value after each training; in, represents the weight factor, and , when the detection value is a positive sample, the weight factor is , when the detection value is a negative sample, the weight factor is ; represents the adjustment factor, R represents the hyperparameter of the adjustment factor, and , the larger the R, the greater the adjustment strength; According to the Focal Loss loss value, sample balance adjustment is performed on the graph convolutional neural network.

7. A distribution network fault location method based on graph convolutional neural network according to claim 1, characterized in that: Inputting the data to be tested into the distribution network fault location model to identify the state parameters of each node, and locating the faulty node according to the node state parameters, specifically includes: Collecting the data to be tested in real time in the power distribution network to be tested; The data to be tested corresponding to each node of the topological structure diagram is cyclically input into the distribution network fault location model, and the node state parameters of each node are identified by the distribution network fault location model; Determine the node where the fault occurs according to the difference between the node status parameter and the normal parameter; Generate an early warning message and send the failed node to the recipient's terminal.

8. A distribution network fault location device based on graph convolutional neural network, characterized in that: The device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, so that the at least one processor can execute the distribution network fault location method based on graph convolutional neural network according to any one of claims 1-7.

Images