Power distribution network fault determination method and device, equipment and storage medium

Abstract

The embodiment of the invention provides a power distribution network fault determination method and device, equipment and a storage medium. According to the method, first normalized data of effective data in operation state data corresponding to at least one node in a to-be-processed power distribution network is determined, then the first normalized data of the at least one node is input into a preset detection model, and a target fault probability corresponding to a target data feature of the at least one node is obtained, the detection model is determined based on the node information of at least one node in the to-be-processed power distribution network and the connection relationship between the nodes, and finally determining the target position information corresponding to the fault node in the to-be-processed power distribution network according to the target fault probability of the at least one node and the position information of each node. According to the method provided by the invention, the fault identification speed is accelerated, and accurate positioning of the power distribution network fault is realized.

Description

Technical Field

[0001] The present application relates to the field of electric power technology, and in particular to a method, apparatus, device, and storage medium for determining a distribution network fault. Background Art

[0002] As power systems continue to expand and become more complex, fault location in distribution networks is becoming increasingly important. Fault location can quickly and accurately identify the location of faults in distribution networks, helping personnel quickly address them, reducing outages and improving power supply reliability.

[0003] Traditional fault location methods include the impedance method, voltage drop method, and traveling wave method. The impedance method mainly determines the fault location by measuring the impedance between the fault point and the fault protection device; the voltage drop method mainly determines the fault location by measuring the voltage change before and after the fault and the electrical characteristics of the line; the traveling wave method mainly calculates the fault location by measuring the time difference between the arrival of the traveling wave at sensors at different locations.

[0004] However, traditional fault location methods have the problems of slow fault identification speed and low fault location accuracy. Summary of the Invention

[0005] The embodiments of the present application provide a method, apparatus, device and storage medium for determining distribution network faults, so as to achieve the technical problem of accelerating fault identification speed and improving fault location accuracy.

[0006] In a first aspect, an embodiment of the present application provides a method for determining a distribution network fault, comprising:

[0007] Determining first normalized data of valid data in the operating status data corresponding to at least one node in the distribution network to be processed;

[0008] Inputting the first normalized data of the at least one node into a preset detection model to obtain a target fault probability corresponding to a target data feature of the at least one node, wherein the detection model is determined based on node information of the at least one node in the distribution network to be processed and a connection relationship between each node;

[0009] The target location information corresponding to the node with fault in the distribution network to be processed is determined according to the target failure probability of the at least one node and the location information of each node.

[0010] In one or more embodiments, before inputting the first normalized data of the at least one node into a preset detection model to obtain a target failure probability corresponding to a target data feature of the at least one node, the method further includes:

[0011] Determining a degree matrix and an adjacency matrix of the distribution network to be processed according to node information of at least one node in the distribution network to be processed and a connection relationship between each node;

[0012] Determining a Laplace matrix of the power distribution network to be processed and second normalized data corresponding to the Laplace matrix according to the degree matrix and the adjacency matrix;

[0013] The detection model is determined based on the second normalized data corresponding to the Laplacian matrix and the initial convolution kernel in the preset graph convolutional neural network model.

[0014] In one or more embodiments, determining the detection model based on the second normalized data corresponding to the Laplacian matrix and an initial convolution kernel in a preset graph convolutional neural network model includes:

[0015] Determine a first characteristic matrix corresponding to the second normalized data corresponding to the Laplace matrix;

[0016] Determining a spectral convolution kernel according to the first feature matrix and an initial convolution kernel in the graph convolutional neural network model;

[0017] Performing Chebyshev processing on the spectral convolution kernel to obtain a Chebyshev convolution kernel;

[0018] The detection model is determined according to the Chebyshev convolution kernel and the second normalized data.

[0019] In one or more embodiments, determining the detection model according to the Chebyshev convolution kernel and the second normalized data includes:

[0020] Determining model convolution information of the graph convolutional neural network model according to the Chebyshev convolution kernel and the second normalized data;

[0021] According to at least one historical power grid topology map and the labeled data features corresponding to the at least one historical power grid topology map, the graph convolutional neural network model corresponding to the model convolution information is trained to obtain the detection model.

[0022] In one or more embodiments, the step of training a graph convolutional neural network model corresponding to the model convolution information based on at least one historical power grid topology map and the annotated data features corresponding to the at least one historical power grid topology map to obtain the detection model includes:

[0023] For each historical power grid topology graph, converting the labeled data features of at least one node in the historical power grid topology graph into one-hot encoded data;

[0024] According to each one-hot encoded data and the at least one historical power grid topology map, a graph convolutional neural network model corresponding to the model convolution information is trained to obtain the detection model.

[0025] In one or more embodiments, determining target location information corresponding to a node with a fault in the distribution network to be processed based on the target failure probability of the at least one node and the location information of each node includes:

[0026] The node with the highest probability of target failure is regarded as the node with failure;

[0027] The location information of the node having the fault is determined as the target location information.

[0028] In one or more embodiments, determining first normalized data of valid data in the operating status data corresponding to at least one node in the distribution network to be processed includes:

[0029] Obtaining operating status data corresponding to at least one node in the distribution network to be processed;

[0030] For each node, determining valid data from the operating status data corresponding to the node;

[0031] The valid data corresponding to each node is normalized to obtain the first normalized data of each node.

[0032] In a second aspect, an embodiment of the present application provides a device for determining a distribution network fault, comprising:

[0033] A first determining module is used to determine first normalized data of valid data in the operating status data corresponding to at least one node in the distribution network to be processed;

[0034] a processing module, configured to input the first normalized data of the at least one node into a preset detection model to obtain a target failure probability corresponding to a target data feature of the at least one node, wherein the detection model is determined based on node information of the at least one node in the distribution network to be processed and a connection relationship between each node;

[0035] The second determining module is configured to determine target location information corresponding to a node with a fault in the distribution network to be processed according to the target failure probability of the at least one node and the location information of each node.

[0036] In one or more embodiments, before inputting the first normalized data of the at least one node into a preset detection model to obtain a target failure probability corresponding to a target data feature of the at least one node, the processing module is further configured to:

[0037] Determining a degree matrix and an adjacency matrix of the distribution network to be processed according to node information of at least one node in the distribution network to be processed and a connection relationship between each node;

[0038] Determining a Laplace matrix of the power distribution network to be processed and second normalized data corresponding to the Laplace matrix according to the degree matrix and the adjacency matrix;

[0039] The detection model is determined based on the second normalized data corresponding to the Laplacian matrix and the initial convolution kernel in the preset graph convolutional neural network model.

[0040] In one or more embodiments, the detection model is determined based on the second normalized data corresponding to the Laplacian matrix and the initial convolution kernel in the preset graph convolutional neural network model, and the processing module is specifically used to:

[0041] Determine a first characteristic matrix corresponding to the second normalized data corresponding to the Laplace matrix;

[0042] Determining a spectral convolution kernel according to the first feature matrix and an initial convolution kernel in the graph convolutional neural network model;

[0043] Performing Chebyshev processing on the spectral convolution kernel to obtain a Chebyshev convolution kernel;

[0044] The detection model is determined according to the Chebyshev convolution kernel and the second normalized data.

[0045] In one or more embodiments, the detection model is determined based on the Chebyshev convolution kernel and the second normalized data, and the processing module is specifically configured to:

[0046] Determining model convolution information of the graph convolutional neural network model according to the Chebyshev convolution kernel and the second normalized data;

[0047] According to at least one historical power grid topology map and the labeled data features corresponding to the at least one historical power grid topology map, the graph convolutional neural network model corresponding to the model convolution information is trained to obtain the detection model.

[0048] In one or more embodiments, the graph convolutional neural network model corresponding to the model convolution information is trained based on at least one historical power grid topology map and the annotated data features corresponding to the at least one historical power grid topology map to obtain the detection model, and the processing module is specifically used to:

[0049] For each historical power grid topology graph, converting the labeled data features of at least one node in the historical power grid topology graph into one-hot encoded data;

[0050] According to each one-hot encoded data and the at least one historical power grid topology map, a graph convolutional neural network model corresponding to the model convolution information is trained to obtain the detection model.

[0051] In one or more embodiments, the second determining module is specifically configured to:

[0052] The node with the highest probability of target failure is regarded as the node with failure;

[0053] The location information of the node having the fault is determined as the target location information.

[0054] In one or more embodiments, the first determining module is specifically configured to:

[0055] Obtaining operating status data corresponding to at least one node in the distribution network to be processed;

[0056] For each node, determining valid data from the operating status data corresponding to the node;

[0057] The valid data corresponding to each node is normalized to obtain the first normalized data of each node.

[0058] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a memory, a processor;

[0059] The memory stores computer-executable instructions;

[0060] The processor executes the computer-executable instructions stored in the memory, so that the processor is used to implement the method described in the first aspect and any one of the embodiments when executing.

[0061] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the method described in the first aspect and any one of the embodiments above.

[0062] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, is used to implement the method for determining a distribution network fault as described in the first aspect and various possible implementations of the first aspect.

[0063] The present application provides a method, apparatus, device, and storage medium for determining distribution network faults. The method first determines first normalized data of valid data in operating status data corresponding to at least one node in a distribution network to be processed, then inputs the first normalized data of the at least one node into a preset detection model to obtain a target fault probability corresponding to a target data feature of the at least one node, wherein the detection model is determined based on node information of the at least one node in the distribution network to be processed and the connection relationship between each node, and finally determines target location information corresponding to a node with a fault in the distribution network to be processed based on the target fault probability of the at least one node and the location information of each node. In the above method, by normalizing the valid data in the operating status data corresponding to at least one node in the distribution network, the first normalized data of at least one node is obtained and input into the preset detection model, which can effectively eliminate the influence of the data dimension, provide more accurate and standardized input for subsequent fault location, and help improve the training efficiency and fault location accuracy of subsequent models; using the detection model constructed based on the information of at least one node in the distribution network to be processed and the connection relationship between each node, the topological structure and operating characteristics of the distribution network can be fully explored to achieve intelligent prediction of fault probability; by analyzing the target fault probability and location information of the node, the node location where the fault occurs can be locked more quickly, reducing the time delay in traditional manual positioning and improving the fault processing speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0065] Figure 1 Schematic diagram of the process of determining a distribution network fault provided in the embodiment of the present application Figure 1 ;

[0066] Figure 2 A schematic diagram of the structure of the distribution network to be processed provided in an embodiment of the present application;

[0067] Figure 3 Schematic diagram of the process of determining a distribution network fault provided in the embodiment of the present application Figure 2 ;

[0068] Figure 4 Schematic diagram of the process of determining a distribution network fault provided in the embodiment of the present application Figure 3 ;

[0069] Figure 5 A schematic diagram of the structure of a device for determining a distribution network fault provided in an embodiment of the present application;

[0070] Figure 6A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0071] The above drawings illustrate specific embodiments of the present application, which will be described in more detail below. These drawings and the textual description are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0072] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0073] Before introducing the embodiments of the present application, the application background of the embodiments of the present application is first explained:

[0074] As power systems continue to expand and become more complex, fault location in distribution networks is becoming increasingly important. Fault location can quickly and accurately identify the location of faults in distribution networks, helping personnel quickly address them, reducing outages and improving power supply reliability.

[0075] Traditional fault location methods include the impedance method, voltage drop method, and traveling wave method. The impedance method mainly determines the fault location by measuring the impedance between the fault point and the fault protection device; the voltage drop method mainly determines the fault location by measuring the voltage change before and after the fault and the electrical characteristics of the line; the traveling wave method mainly calculates the fault location by measuring the time difference between the arrival of the traveling wave at sensors at different locations.

[0076] However, when faced with complex distribution network topologies, traditional fault location methods often exhibit limitations such as dependence on distribution network topology, sensitivity to noise and data loss, and high computational complexity, resulting in slow fault identification and low fault location accuracy.

[0077] The method for determining distribution network faults provided by the present application is intended to solve the above technical problems of the prior art. The inventive concept of the present application is as follows: When faced with a complex distribution network topology, the traditional fault location method has the problems of slow fault identification speed and low fault location accuracy. The determination of the fault is mainly to analyze the operating data of the distribution network and determine the fault location in the distribution network topology. If a detection model can be constructed, by inputting the operating data of the distribution network to be processed and analyzing the operating data of the distribution network to be processed through the model, the location information of the distribution network fault can be quickly identified. Therefore, the present application first determines the first normalized data of the valid data in the operating status data corresponding to at least one node in the distribution network to be processed, and then determines the detection model based on the node information of at least one node in the distribution network to be processed and the connection relationship between each node, and obtains the target fault probability corresponding to the target data feature of at least one node through the detection model, and then determines the target location information corresponding to the node with fault in the distribution network to be processed in combination with the location information of each node.

[0078] The execution subject of the embodiments of the present application is an electronic device, which can be a terminal device, such as a laptop computer, a desktop computer, a tablet computer, etc., or a server. In actual applications, whether the electronic device is a terminal device or a server can be determined based on actual conditions and is not specifically limited to this.

[0079] The following specific embodiments describe in detail the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0080] Figure 1 Schematic diagram of the process of determining a distribution network fault provided in the embodiment of the present application Figure 1 .

[0081] like Figure 1 As shown, the method for determining a distribution network fault includes the following steps:

[0082] S110. Determine first normalized data of valid data in operating status data corresponding to at least one node in the distribution network to be processed.

[0083] In this step, the operating status data corresponding to at least one node in the distribution network to be processed includes valid data, and the valid data in the operating status data corresponding to at least one node in the distribution network to be processed is normalized to determine first normalized data.

[0084] Exemplarily, at least one node in the distribution network to be processed generally represents a key connection point or equipment in the power system, such as a substation node, a load node, a switch node, a device node, etc.

[0085] The valid data in the operating status data corresponding to at least one node in the power distribution network to be processed may be valid values of three-phase voltage data and three-phase current data of the at least one node.

[0086] In one possible implementation, determining the first normalized data of valid data in the operating status data corresponding to at least one node in the distribution network to be processed can be normalizing the valid data by performing standard deviation normalization to obtain first normalized data that conforms to the standard normal distribution.

[0087] S120: Inputting first normalized data of at least one node into a preset detection model to obtain a target failure probability corresponding to a target data feature of the at least one node;

[0088] The detection model is determined based on the node information of at least one node in the distribution network to be processed and the connection relationship between each node.

[0089] In this step, a detection model can be determined based on the node information of at least one node in the distribution network to be processed and the connection relationship between each node. The first normalized data of at least one node is used as the input of the pre-set detection model. The target data features of at least one node are extracted and processed through the detection model to obtain the target fault probability corresponding to the target data features of at least one node.

[0090] Exemplarily, the node information of at least one node in the distribution network to be processed includes the admittance matrix of the node, the load parameter of the node, the node number, and the location information of the node.

[0091] In one possible implementation, the detection model can perform feature extraction on the first normalized data of at least one node, and based on the target data features corresponding to the first normalized data of at least one node, perform a binary classification solution through the normalization function Softmax function to obtain the target fault probability corresponding to the target data features of at least one node.

[0092] In addition, the connection relationship between each node is defined and input in the form of a start node and a target node. The numbers of all the start nodes and target nodes in the connection relationship between each node are extracted, the repeated node numbers in the node numbers are removed, and the number of unique node numbers in the node numbers is recorded. The total number of nodes in the distribution network to be processed can be obtained.

[0093] S130. Determine target location information corresponding to a node with a fault in the distribution network to be processed according to a target failure probability of at least one node and location information of each node.

[0094] In this step, the faulty nodes in the distribution network to be processed can be determined based on the target failure probability of at least one node, and the target location information corresponding to the faulty nodes in the distribution network to be processed can be determined based on the location information of each node.

[0095] Exemplarily, the location information of each node is used to indicate the location corresponding to the number of each node in the power distribution network to be processed.

[0096] For example, Figure 2 The schematic diagram of the structure of the distribution network to be processed provided in the embodiment of the present application is as follows: Figure 2 As shown, the distribution network to be processed includes 33 nodes, and the connection relationship between each node is in the form of a solid line or a dotted line. The solid line represents the actual connection, and the dotted line represents the tie switch. The position corresponding to the number of each node is the location information of each node.

[0097] In a possible implementation, a possible implementation of step S130 further includes the following steps:

[0098] Step 1: The node with the highest target failure probability is regarded as the node with failure.

[0099] Exemplarily, the target failure probability of at least one node is compared, and the node with the highest target failure probability is determined to be the node with the failure.

[0100] For example, Figure 2 For example, among the 33 nodes, it is determined that the nodes with target failure probabilities not equal to 0 are node 1, node 3, and node 5. The target failure probability of node 1 is 0.4, the target failure probability of node 3 is 0.8, and the target failure probability of node 5 is 0.3. The node with the largest target failure probability is node 3. Node 3 is then regarded as the node with a fault in the distribution network to be processed.

[0101] Step 2: Determine the location information of the faulty node as the target location information.

[0102] Exemplarily, corresponding location information can be obtained based on the node with the fault, and the location information is determined as the target location information.

[0103] Exemplarily, the location information of the node with the fault (ie, the location corresponding to the node number) is the target location information.

[0104] For example, based on the above example, the node with a fault is node 3, and the position corresponding to node 3 is the target position information.

[0105] The method for determining a distribution network fault provided in an embodiment of the present application first determines the first normalized data of valid data in the operating status data corresponding to at least one node in the distribution network to be processed, and then inputs the first normalized data of the at least one node into a preset detection model to obtain a target fault probability corresponding to the target data feature of the at least one node, wherein the detection model is determined based on the node information of at least one node in the distribution network to be processed and the connection relationship between each node, and finally, based on the target failure probability of the at least one node and the location information of each node, the target position information corresponding to the node with a fault in the distribution network to be processed is determined. In this embodiment, by normalizing the valid data in the operating status data corresponding to at least one node in the distribution network, the first normalized data of at least one node is obtained and input into the preset detection model, which can effectively eliminate the influence of the data dimension, provide more accurate and standardized input for subsequent fault location, and help improve the training efficiency and fault location accuracy of subsequent models; using the detection model constructed based on the information of at least one node in the distribution network to be processed and the connection relationship between each node, the topological structure and operating characteristics of the distribution network can be fully explored to achieve intelligent prediction of fault probability; by analyzing the target fault probability and location information of the node, the node location where the fault occurs can be locked more quickly, reducing the time delay in traditional manual positioning, and improving the fault processing speed.

[0106] Based on the above embodiments, Figure 3 Schematic diagram of the process of determining a distribution network fault provided in the embodiment of the present application Figure 2 .like Figure 3 As shown, before the above step S120, the method for determining a distribution network fault further includes the following steps:

[0107] S310 : Determine a degree matrix and an adjacency matrix of the distribution network to be processed according to node information of at least one node in the distribution network to be processed and connection relationships between nodes.

[0108] In this step, the node information of at least one node in the distribution network to be processed and the connection relationship between each node are converted into the form of a degree matrix and an adjacency matrix to determine the degree matrix and adjacency matrix of the distribution network to be processed.

[0109] For example, the degree matrix is a diagonal matrix that represents the degree of each node (ie, the number of nodes connected to the node), and the adjacency matrix is a symmetric matrix that indicates the connection relationship between each node.

[0110] In one possible implementation, the matrix element A of the adjacency matrix A ij Indicates whether there is a connection relationship between node i and node j. When A ij= 0, it means that there is no connection between node i and node j. ij =1, it indicates that there is a connection relationship between node i and node j.

[0111] The diagonal elements D of the degree matrix D ii represents the degree of node i and can be defined as follows:

[0112]

[0113] Where N represents the total number of nodes.

[0114] In addition, at least one distribution network graph to be processed corresponding to the distribution network to be processed may be determined based on the node information of at least one node in the distribution network to be processed and the connection relationship between the nodes.

[0115] S320: Determine the Laplace matrix of the distribution network to be processed according to the degree matrix and the adjacency matrix, and determine second normalized data corresponding to the Laplace matrix.

[0116] In this step, the degree matrix and the adjacency matrix are subtracted to determine the Laplace matrix of the distribution network to be processed, and the Laplace matrix of the distribution network to be processed is normalized to obtain second normalized data corresponding to the Laplace matrix.

[0117] Exemplarily, the Laplace matrix of the distribution network to be processed is used to describe the topological structure of the distribution network to be processed and the connection relationship between each node.

[0118] In one possible implementation, the Laplacian matrix L can be obtained by subtracting the adjacency matrix A from the degree matrix D.

[0119] Normalize the Laplace matrix, that is, divide each node by its own degree for normalization, which corresponds to a matrix multiplication of the rows and columns, that is:

[0120] Δ=D -1 / 2 LD -1 / 2 =D -1 / 2 (DA)D -1 / 2 =ID -1 / 2 AD -1 / 2

[0121] Wherein, Δ represents the normalized Laplace matrix, that is, the second normalized data corresponding to the Laplace matrix, and I represents the identity matrix.

[0122] S330. Determine a detection model based on the second normalized data corresponding to the Laplace matrix and the initial convolution kernel in the preset graph convolutional neural network model.

[0123] In this step, the convolution kernel in the detection model is determined according to the second normalized data corresponding to the Laplace matrix and the initial convolution kernel in the preset graph convolutional neural network model, thereby determining the detection model.

[0124] In a possible implementation, a possible implementation of step S330 further includes the following steps:

[0125] S331. Determine a first characteristic matrix corresponding to the second normalized data corresponding to the Laplace matrix.

[0126] Exemplarily, the eigenvector corresponding to the second normalized data corresponding to the Laplace matrix is calculated, the eigenvector is converted into a matrix representation, and the first eigenmatrix corresponding to the second normalized data corresponding to the Laplace matrix is determined.

[0127] In a possible implementation, the eigenvalues and eigenvectors corresponding to the second normalized data corresponding to the Laplace matrix may be obtained by solving the second normalized data and the characteristic equation corresponding to the Laplace matrix.

[0128] S332. Determine a spectral convolution kernel based on the first feature matrix and the initial convolution kernel in the graph convolutional neural network model.

[0129] Exemplarily, the spectral decomposition of the second normalized data is obtained according to the first feature matrix, and then the spectral convolution kernel is determined according to the initial convolution kernel in the graph convolutional neural network model and the spectral decomposition of the second normalized data.

[0130] In one possible implementation, the spectral decomposition of the Laplacian matrix is defined as:

[0131] Δ=UΛU T

[0132] Where U represents the first characteristic matrix and Λ represents the eigenvalue matrix.

[0133] The general form of the convolution operation in the graph convolutional neural network model can be defined by the spectral decomposition of the Laplacian matrix:

[0134]

[0135] Among them, g represents the initial convolution kernel, It represents the element-by-element multiplication of matrices, that is, the elements at corresponding positions of two matrices are multiplied, and f represents the distribution network diagram to be processed.

[0136] According to the general form of the convolution operation in the graph convolutional neural network model, the spectral convolution kernel g(Λ)=(U T g).

[0137] S333. Perform Chebyshev processing on the spectral convolution kernel to obtain a Chebyshev convolution kernel.

[0138] For example, the Chebyshev polynomial T is used. k (x) is used to approximate the spectral convolution kernel g(Λ) and obtain the Chebyshev convolution kernel.

[0139] In one possible implementation, the Chebyshev convolution kernel is defined as:

[0140]

[0141] Among them, θ k Represents the coefficient to be learned, T k (x) represents the Chebyshev polynomial, which satisfies the recurrence relation:

[0142] T0(x)=1,T1(x)=x,T k (x) = 2xT k-1 (x)-T k-2 (x),

[0143] is the scaled eigenvalue matrix:

[0144]

[0145] Among them, Λ Max Indicates the largest eigenvalue among the eigenvalues.

[0146] S334. Determine a detection model based on the Chebyshev convolution kernel and the second normalized data.

[0147] Exemplarily, a Chebyshev convolution kernel is determined as the convolution kernel in the detection model, and a detection model based on the Chebyshev convolution kernel is determined according to the second normalized data.

[0148] In a possible implementation, a possible implementation of step S334 further includes the following steps:

[0149] S1. Determine the model convolution information of the graph convolutional neural network model according to the Chebyshev convolution kernel and the second normalized data.

[0150] Exemplarily, a Chebyshev convolution kernel is determined as a convolution kernel in a graph convolutional neural network model, and model convolution information based on the Chebyshev convolution kernel is determined according to the second normalized data.

[0151] In one possible implementation, the model convolution information based on the Chebyshev convolution kernel, that is, the convolution operation, can be expressed as:

[0152]

[0153] Finally it simplifies to:

[0154]

[0155] S2. Based on at least one historical power grid topology map and the labeled data features corresponding to at least one historical power grid topology map, a graph convolutional neural network model corresponding to the model convolution information is trained to obtain a detection model.

[0156] Exemplarily, the annotation data feature corresponding to at least one historical power grid topology map is used to indicate the annotation of location information of a faulty node contained in a node set corresponding to at least one historical power grid topology map.

[0157] In one possible implementation, at least one historical power grid topology map is used as the input of a graph convolutional neural network model to obtain the predicted location information of the faulty node. The predicted location information of the faulty node is compared with the location information of the faulty node marked in the marked data features, and the graph convolutional neural network model corresponding to the model convolution information is trained to obtain a detection model.

[0158] In a possible implementation, a possible implementation of step S2 further includes the following steps:

[0159] Step 1: For each historical power grid topology graph, the labeled data features of at least one node in the historical power grid topology graph are converted into one-hot encoded data.

[0160] Exemplarily, the length of the one-hot encoded data is the same as the number of nodes in each historical power grid topology graph. A one-hot encoded data of 1 indicates that the node has failed, and the one-hot encoded data of other nodes that have not failed is 0.

[0161] In a possible implementation, location information of a faulty node in at least one node in the historical power grid topology graph may be determined based on the one-hot encoded data.

[0162] Step 2: Based on each one-hot encoded data and at least one historical power grid topology map, the graph convolutional neural network model corresponding to the model convolution information is trained to obtain a detection model.

[0163] Exemplarily, at least one historical power grid topology map is used as the input of the graph convolutional neural network model corresponding to the model convolution information, each one-hot encoded data is compared with the location information of the predicted faulty node output by the graph convolutional neural network model, the graph convolutional neural network model corresponding to the model convolution information is trained based on the comparison results, and the trained graph convolutional neural network model is used as the detection model.

[0164] In one possible implementation, the cross entropy loss between the predicted location information of the faulty node and the one-hot encoded data corresponding to the labeled data features of at least one node is calculated, and back propagation is performed to update the weights of the graph convolutional neural network model until the location information of the faulty node predicted by the graph convolutional neural network model is close to the one-hot encoded data (that is, the probability of the faulty node is 1, and the probability of other nodes is 0). The training is then completed and the trained graph convolutional neural network model is used as the detection model.

[0165] The method for determining a distribution network fault provided in an embodiment of the present application first determines the degree matrix and adjacency matrix of the distribution network to be processed based on the node information of at least one node in the distribution network to be processed and the connection relationship between each node. Then, based on the degree matrix and the adjacency matrix, the Laplace matrix of the distribution network to be processed and the second normalized data corresponding to the Laplace matrix are determined. Finally, the detection model is determined based on the second normalized data corresponding to the Laplace matrix and the initial convolution kernel in the preset graph convolutional neural network model. In this embodiment, by determining the degree matrix, adjacency matrix and Laplace matrix based on the node information of at least one node in the distribution network and the connection relationship between each node, it is helpful to deeply understand the topological structure of the distribution network and the relationship between each node, thereby providing richer and more accurate data support for subsequent fault analysis and detection; when constructing a detection model based on the Laplace matrix and the second normalized data, the graph convolutional neural network can make full use of the graph structure characteristics of the distribution network to effectively model the connection relationship between nodes and each node; by performing graph convolution processing on the topological structure of the distribution network, the relationship and abnormal changes between each node in the distribution network can be captured more finely. When a fault occurs, the detection model can accurately identify the location of the fault node, improve the accuracy of distribution network fault location, and use the detection model to achieve more automated and intelligent fault identification and fault location, thereby improving the self-healing ability and power supply reliability of the distribution network.

[0166] Based on the above embodiments, Figure 4 Schematic diagram of the process of determining a distribution network fault provided in the embodiment of the present application Figure 3 .like Figure 4 As shown, a possible implementation of the above step S110 further includes the following steps:

[0167] S410: Obtain operating status data corresponding to at least one node in the distribution network to be processed.

[0168] In this step, at least one node in the distribution network to be processed corresponds to operating status data, and the operating status data corresponding to the at least one node can be obtained for subsequent analysis of the at least one node.

[0169] Exemplarily, the operating status data includes measurement data of three-phase voltage (UA, UB, UC) and three-phase current (IA, IB, IC) of at least one node.

[0170] In a possible implementation, the operating status data corresponding to at least one node in the distribution network to be processed may be obtained by real-time collection through monitoring equipment such as a voltage transformer, a current transformer, and a smart meter.

[0171] S420 : For each node, determine valid data from the operating status data corresponding to the node.

[0172] In this step, for each node, the valid value corresponding to the running status data corresponding to the node is calculated and determined as valid data.

[0173] In one possible implementation, the effective values of the three-phase voltage and the three-phase current are used to measure the strength of the AC signal. The effective value of the three-phase voltage (Ua, Ub, Uc) can be obtained by taking the square root of the mean of the sum of the squares of the three-phase voltage, and the effective value of the three-phase current (Ia, Ib, Ic) can be obtained by taking the square root of the mean of the sum of the squares of the three-phase current.

[0174] Table 1 Valid data (taking the first three nodes as an example)

[0175]

[0176] S430 : Normalize the valid data corresponding to each node to obtain first normalized data of each node.

[0177] In this step, there is valid data corresponding to each node, and the valid data corresponding to each node is normalized, and the valid data after normalization of each node is used as the first normalized data of each node.

[0178] In a possible implementation, normalizing the valid data corresponding to each node to obtain the first normalized data of each node can be normalizing the valid data corresponding to each node by standard deviation to obtain the first normalized data that conforms to the standard normal distribution.

[0179] The standard deviation normalization formula can be expressed as follows:

[0180]

[0181] Among them, x is a set of valid data corresponding to each node, μ is the mean of the set of data, and σ is the standard deviation of the set of data.

[0182] Based on the above example, taking the first three nodes as an example, Table 2 shows the first normalized data (three-phase voltage and three-phase current) obtained after normalization.

[0183] Table 2 First normalized data (taking the first three nodes as an example)

[0184]

[0185] The method for determining a distribution network fault provided in an embodiment of the present application first obtains operating status data corresponding to at least one node in the distribution network to be processed, then determines valid data from the operating status data corresponding to each node for each node, and finally normalizes the valid data corresponding to each node to obtain first normalized data for each node. In this embodiment, by obtaining operating status data corresponding to at least one node in the distribution network to be processed and filtering out valid data, invalid or abnormal data can be effectively removed, which helps to ensure the data quality of subsequent processing stages, avoid errors introduced by invalid data, and improve the accuracy of fault location; by normalizing the valid data corresponding to each node, the dimensional differences between the data of each node can be eliminated, thereby improving the reliability of the analysis results.

[0186] Based on the above embodiments, the following are device embodiments involved in this application:

[0187] Figure 5 This is a schematic diagram of the structure of the device for determining a distribution network fault provided in an embodiment of the present application. Figure 5 As shown, the distribution network fault determination device 500 includes:

[0188] A first determining module 510 is configured to determine first normalized data of valid data in the operating status data corresponding to at least one node in the distribution network to be processed;

[0189] a processing module 520 configured to input the first normalized data of at least one node into a preset detection model to obtain a target failure probability corresponding to a target data feature of the at least one node, wherein the detection model is determined based on node information of at least one node in the distribution network to be processed and a connection relationship between the nodes;

[0190] The second determining module 530 is configured to determine target location information corresponding to a node with a fault in the distribution network to be processed according to a target failure probability of at least one node and location information of each node.

[0191] In one or more embodiments, before inputting the first normalized data of at least one node into a preset detection model to obtain a target failure probability corresponding to a target data feature of the at least one node, the processing module 520 is further configured to:

[0192] Determining a degree matrix and an adjacency matrix of the distribution network to be processed according to node information of at least one node in the distribution network to be processed and a connection relationship between each node;

[0193] Determining a Laplace matrix of the distribution network to be processed and second normalized data corresponding to the Laplace matrix according to the degree matrix and the adjacency matrix;

[0194] The detection model is determined according to the second normalized data corresponding to the Laplace matrix and the initial convolution kernel in the preset graph convolutional neural network model.

[0195] In one or more embodiments, the detection model is determined based on the second normalized data corresponding to the Laplacian matrix and the initial convolution kernel in the preset graph convolutional neural network model. The processing module 520 is specifically configured to:

[0196] Determine a first characteristic matrix corresponding to the second normalized data corresponding to the Laplace matrix;

[0197] Determine a spectral convolution kernel according to the first feature matrix and the initial convolution kernel in the graph convolutional neural network model;

[0198] Perform Chebyshev processing on the spectral convolution kernel to obtain the Chebyshev convolution kernel;

[0199] The detection model is determined based on the Chebyshev convolution kernel and the second normalized data.

[0200] In one or more embodiments, a detection model is determined based on the Chebyshev convolution kernel and the second normalized data, and the processing module 520 is specifically configured to:

[0201] Determine the model convolution information of the graph convolutional neural network model based on the Chebyshev convolution kernel and the second normalized data;

[0202] According to at least one historical power grid topology map and the labeled data features corresponding to at least one historical power grid topology map, a graph convolutional neural network model corresponding to the model convolution information is trained to obtain a detection model.

[0203] In one or more embodiments, based on at least one historical power grid topology map and the annotated data features corresponding to at least one historical power grid topology map, a graph convolutional neural network model corresponding to the model convolution information is trained to obtain a detection model. The processing module 520 is specifically configured to:

[0204] For each historical power grid topology graph, converting the labeled data features of at least one node in the historical power grid topology graph into one-hot encoded data;

[0205] According to each one-hot encoded data and at least one historical power grid topology map, a graph convolutional neural network model corresponding to the model convolution information is trained to obtain a detection model.

[0206] In one or more embodiments, the second determining module 530 is specifically configured to:

[0207] The node with the highest probability of target failure is regarded as the node with failure;

[0208] The location information of the node having the fault is determined as the target location information.

[0209] In one or more embodiments, the first determining module 510 is specifically configured to:

[0210] Obtaining operating status data corresponding to at least one node in the distribution network to be processed;

[0211] For each node, determine the valid data from the running status data corresponding to the node;

[0212] The valid data corresponding to each node is normalized to obtain the first normalized data of each node.

[0213] Based on the above embodiments, Figure 6 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present application. Figure 6 As shown, the electronic device 600 includes: a processor 610, a memory 620 and a bus 630;

[0214] The memory 620 is used to store computer-executable instructions of the processor 610;

[0215] The processor 610 is configured to execute the technical solution of any of the aforementioned method embodiments by executing computer execution instructions.

[0216] Optionally, the memory 620 may be independent or integrated with the processor 610 .

[0217] Optionally, the memory 620 may include a random access memory (RAM), and may also include a non-volatile memory (NVM), such as at least one disk memory.

[0218] Bus 630 may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Buses can be classified as address buses, data buses, control buses, etc. For ease of illustration, the drawings of this application use only one thick line, but this does not mean that there is only one bus or only one type of bus.

[0219] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, and discrete hardware components.

[0220] The electronic device is used to execute the technical solution of any of the aforementioned method embodiments, and its implementation principle and technical effects are similar and will not be repeated here.

[0221] An embodiment of the present application also provides a computer-readable storage medium on which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the technical solution provided by any of the above method embodiments.

[0222] An embodiment of the present application also provides a computer program product, including a computer program, which includes computer instructions stored in a computer-readable storage medium. When the computer program is executed by a processor, it is used to implement the technical solution provided by any of the above method embodiments.

[0223] It should be noted that for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all optional embodiments, and the actions and modules involved are not necessarily required by this application.

[0224] It should be further noted that, although the various steps in the flowchart are shown in sequence as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps may be performed in other orders. Moreover, at least a portion of the steps in the flowchart may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily performed at the same time, but may be performed at different times. The execution order of these sub-steps or stages is not necessarily to be performed in sequence, but may be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0225] It should be understood that the above-described device embodiments are merely illustrative, and the device of the present application may also be implemented in other ways. For example, the division of units / modules in the above-described embodiments is merely a logical functional division, and actual implementations may employ other division methods. For example, multiple units, modules, or components may be combined or integrated into another system, or some features may be omitted or not implemented.

[0226] In addition, unless otherwise specified, the functional units / modules in the various embodiments of the present application may be integrated into a single unit / module, each unit / module may exist physically separately, or two or more units / modules may be integrated together. The aforementioned integrated units / modules may be implemented in the form of hardware or software program modules.

[0227] If the integrated unit / module is implemented in hardware, the hardware may be digital circuits, analog circuits, etc. The physical implementation of the hardware structure includes, but is not limited to, transistors, memristors, etc. Unless otherwise specified, the processor may be any appropriate hardware processor, such as a CPU, GPU, FPGA, DSP, and ASIC. Unless otherwise specified, the storage unit may be any appropriate magnetic storage medium or magneto-optical storage medium, such as resistive random access memory (RRAM), dynamic random access memory (DRAM), static random access memory (SRAM), enhanced dynamic random access memory (EDRAM), high-bandwidth memory (HBM), hybrid memory cube (HMC), etc.

[0228] If the integrated unit / module is implemented in the form of a software program module and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a memory and includes a number of instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory includes various media that can store program codes, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0229] In the above embodiments, the description of each embodiment has its own emphasis. For parts not described in detail in a particular embodiment, please refer to the relevant description of other embodiments. The technical features of the above embodiments can be combined in any way. To keep the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0230] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of the present application and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0231] It should be understood that the present application is not limited to the exact structure described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A method for determining a distribution network fault, characterized in that: include: Determining first normalized data of valid data in the operating status data corresponding to at least one node in the distribution network to be processed; Inputting the first normalized data of the at least one node into a preset detection model to obtain a target fault probability corresponding to a target data feature of the at least one node, wherein the detection model is determined based on node information of the at least one node in the distribution network to be processed and a connection relationship between each node; The target location information corresponding to the node with fault in the distribution network to be processed is determined according to the target failure probability of the at least one node and the location information of each node.

2. The method according to claim 1, characterized in that Before inputting the first normalized data of the at least one node into a preset detection model to obtain a target failure probability corresponding to the target data feature of the at least one node, the method further includes: Determining a degree matrix and an adjacency matrix of the distribution network to be processed according to node information of at least one node in the distribution network to be processed and a connection relationship between each node; Determining a Laplace matrix of the power distribution network to be processed and second normalized data corresponding to the Laplace matrix according to the degree matrix and the adjacency matrix; The detection model is determined based on the second normalized data corresponding to the Laplacian matrix and the initial convolution kernel in the preset graph convolutional neural network model.

3. The method according to claim 2, characterized in that The determining of the detection model according to the second normalized data corresponding to the Laplacian matrix and the initial convolution kernel in the preset graph convolutional neural network model includes: Determine a first characteristic matrix corresponding to the second normalized data corresponding to the Laplace matrix; Determining a spectral convolution kernel according to the first feature matrix and an initial convolution kernel in the graph convolutional neural network model; Performing Chebyshev processing on the spectral convolution kernel to obtain a Chebyshev convolution kernel; The detection model is determined according to the Chebyshev convolution kernel and the second normalized data.

4. The method according to claim 3, characterized in that Determining the detection model according to the Chebyshev convolution kernel and the second normalized data includes: Determining model convolution information of the graph convolutional neural network model according to the Chebyshev convolution kernel and the second normalized data; According to at least one historical power grid topology map and the labeled data features corresponding to the at least one historical power grid topology map, the graph convolutional neural network model corresponding to the model convolution information is trained to obtain the detection model.

5. The method according to claim 4, characterized in that The step of training a graph convolutional neural network model corresponding to the model convolution information based on at least one historical power grid topology map and the annotated data features corresponding to the at least one historical power grid topology map to obtain the detection model includes: For each historical power grid topology graph, converting the labeled data features of at least one node in the historical power grid topology graph into one-hot encoded data; According to each one-hot encoded data and the at least one historical power grid topology map, a graph convolutional neural network model corresponding to the model convolution information is trained to obtain the detection model.

6. The method according to any one of claims 1 to 5, characterized in that The determining, based on the target failure probability of the at least one node and the location information of each node, target location information corresponding to the node with a fault in the distribution network to be processed includes: The node with the highest probability of target failure is regarded as the node with failure; The location information of the node having the fault is determined as the target location information.

7. The method according to any one of claims 1 to 5, characterized in that The determining first normalized data of valid data in the operating status data corresponding to at least one node in the distribution network to be processed includes: Obtaining operating status data corresponding to at least one node in the distribution network to be processed; For each node, determining valid data from the operating status data corresponding to the node; The valid data corresponding to each node is normalized to obtain the first normalized data of each node.

8. A device for determining a distribution network fault, characterized in that: include: A first determining module is used to determine first normalized data of valid data in the operating status data corresponding to at least one node in the distribution network to be processed; a processing module, configured to input the first normalized data of the at least one node into a preset detection model to obtain a target failure probability corresponding to a target data feature of the at least one node, wherein the detection model is determined based on node information of the at least one node in the distribution network to be processed and a connection relationship between each node; The second determining module is configured to determine target location information corresponding to a node with a fault in the distribution network to be processed according to the target failure probability of the at least one node and the location information of each node.

9. An electronic device, characterized in that: include: Memory, processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor performs the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

Images