Power distribution network fault positioning method and device based on zero sequence voltage distribution characteristics

Abstract

The invention relates to the technical field of power systems, and discloses a power distribution network fault positioning method and device based on zero-sequence voltage distribution characteristics, and the method comprises the steps: obtaining a power distribution network node graph and node zero-sequence voltage; constructing a matrix according to the node zero-sequence voltage to obtain a zero-sequence voltage matrix; performing clustering operation on the zero-sequence voltage matrix to obtain an initial class cluster set; performing iterative operation on each element in the zero-sequence voltage matrix according to a preset K-means algorithm to obtain an element mass center set; calculating the fault degree of the initial class cluster set, and screening to obtain a fault class cluster with the maximum fault degree; calculating the membership degree between the fault class cluster and each meta-quality centroid in the meta-quality centroid set, and screening to obtain a fault centroid with the maximum membership degree; and according to the fault centroid, positioning is carried out on the power distribution network node graph, and power distribution network fault positioning is realized. The method can solve the problem of inaccurate fault positioning of the power distribution network.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and particularly to a method and device for locating faults in a distribution network based on the characteristics of zero-sequence voltage distribution. Background Art

[0002] At present, the automation of distribution networks is the main trend in the development of global power systems. A key link in the automation of distribution networks is the location of distribution network faults.

[0003] In an existing technology, by collecting the zero-sequence voltages at each node of the fault section in real time, a zero-sequence voltage distribution matrix is obtained in the spatial dimension. Then, taking the zero-sequence voltage distribution matrix as the research object, clustering analysis is applied to obtain various clusters, and the fault degree of each cluster is located to achieve fault location. This method has certain limitations. For example, when performing clustering analysis on the zero-sequence voltage distribution matrix, the classification method directly uses the centroid distance as the classification basis, resulting in the situation that nodes with different voltage levels or different fault positions are classified into the same category, which affects the accuracy of distribution network fault location.

[0004] In summary, the fault location methods in the existing technology have the problem of inaccurate distribution network fault location. Summary of the Invention

[0005] The present invention provides a method and device for locating faults in a distribution network based on the characteristics of zero-sequence voltage distribution to solve the problem of inaccurate distribution network fault location.

[0006] In a first aspect, to solve the above technical problem, the present invention provides a method for locating faults in a distribution network based on the characteristics of zero-sequence voltage distribution, including:

[0007] Obtain a distribution network node map and the zero-sequence voltage of nodes;

[0008] Construct a matrix based on the zero-sequence voltage of the nodes to obtain a zero-sequence voltage matrix;

[0009] Perform a clustering operation on the zero-sequence voltage matrix to obtain an initial set of clusters;

[0010] Perform an iterative operation on each element in the zero-sequence voltage matrix according to the preset K-means algorithm to obtain a set of element centroids;

[0011] Calculate the fault degree of the initial set of clusters, and screen to obtain the fault cluster with the largest fault degree;

[0012] Calculate the membership degree of the fault cluster to each element centroid in the set of element centroids, and screen to obtain the fault centroid with the largest membership degree;

[0013] Locate on the distribution network node map according to the fault centroid to achieve the location of distribution network faults.

[0014] In an alternative embodiment, a matrix is constructed based on the zero-sequence voltage of the nodes to obtain a zero-sequence voltage matrix, including:

[0015] Number the fault nodes to obtain node numbers;

[0016] Record the sampling time of the fault nodes to obtain sampling time points;

[0017] Use the node numbers as the matrix rows, the sampling time points as the matrix columns, and the zero-sequence voltage of the nodes at the corresponding moments as the element values of the matrix to construct a matrix, obtaining an initial voltage matrix;

[0018] Normalize the initial voltage matrix to obtain a zero-sequence voltage matrix.

[0019] In an alternative embodiment, perform a clustering operation on the zero-sequence voltage matrix to obtain an initial cluster set, including:

[0020] Select multiple cluster centers from the elements of the zero-sequence voltage matrix through a clustering algorithm, where the number of cluster centers is the same as the number of nodes;

[0021] Calculate the distance between each element in the zero-sequence voltage matrix and the cluster centers, and assign the element to the cluster represented by the nearest cluster center;

[0022] Repeat the step of calculating the distance between each element in the zero-sequence voltage matrix and the cluster centers and assigning the element to the cluster represented by the nearest cluster center until all elements in the zero-sequence voltage matrix are assigned, obtaining an initial cluster set.

[0023] In an alternative embodiment, perform an iterative operation on the elements in the zero-sequence voltage matrix according to the preset K-means algorithm to obtain a set of element centroids, including:

[0024] According to the K-means algorithm, randomly select K elements in the zero-sequence voltage matrix as the initial element centroids, where K is equal to the number of nodes;

[0025] Calculate the membership degrees of each element in the zero-sequence voltage matrix to the initial element centroids;

[0026] If the membership degree is positive, it means that the initial element centroid can better represent the node. If the membership degree is negative, replace the current element with the initial element centroid;

[0027] Return the step of calculating the membership degree of each element in the calculated zero-sequence voltage matrix to the centroid of the initial elements, to implement the operation of updating the centroid of the elements. When the membership degree calculation operation of all elements in the node is completed, the iteration ends, and the output result is a centroid set of K element centroids.

[0028] In an alternative embodiment, calculating the membership degree of each element in the zero-sequence voltage matrix to the centroid of the initial elements includes:

[0029] Calculate the membership degree through the following formula:

[0030]

[0031] where, is the element x of the zero-sequence voltage distribution matrix j and the membership degree value to each initial centroid μ k , σ is the preset standard deviation, x j is the element of the zero-sequence voltage distribution matrix, μ k is the initial centroid, and K is the number of clusters.

[0032] In an alternative embodiment, calculating the fault degree of the initial cluster set and screening to obtain the fault cluster with the largest fault degree includes:

[0033] Calculate the fault degree through the following formula:

[0034]

[0035] where, Fault Degree k is the fault degree, R i is the fault degree of the i-th node, M is the total number of nodes contained in the same type of element cluster, α i is the weight coefficient, GNN i is the relationship feature of node i, r is the sum of the correlation coefficients of the zero-sequence voltage on each phase line of the fault-occurring node and the zero-sequence voltage on each phase line of the nodes in the same type of element cluster, r 0 is the average value of the correlation coefficients of the zero-sequence voltage on each phase line of the fault-occurring node and the zero-sequence voltage on each phase line of the nodes in the same type of element cluster.

[0036] In an alternative embodiment, calculating the membership degree of the fault cluster to each centroid in the centroid set of the elements and screening to obtain the fault centroid with the largest membership degree includes:

[0037] Calculate the membership degree through the following formula:

[0038]

[0039] where, c maxis the fault centroid with the largest membership degree, F is the fault cluster, n is the number of element centroids in the element centroid set, c is the element centroid set, D is the data point space dimension, and f is the coordinate of the element in the fault cluster.

[0040] In an optional implementation manner, based on the fault centroid, perform positioning on the distribution network node map to implement distribution network fault positioning, including:

[0041] The fault centroid is an element in the zero-sequence voltage matrix, and the fault node is obtained according to the node where the fault centroid is located;

[0042] Perform positioning on the distribution network node map according to the fault node to implement distribution network fault point positioning.

[0043] In a second aspect, the present invention provides a distribution network fault positioning device based on zero-sequence voltage distribution characteristics, including:

[0044] A data acquisition module for acquiring a distribution network node map and node zero-sequence voltage;

[0045] A matrix construction module for constructing a matrix according to the node zero-sequence voltage to obtain a zero-sequence voltage matrix;

[0046] A clustering operation module for performing a clustering operation on the zero-sequence voltage matrix to obtain an initial cluster set;

[0047] A centroid iteration module for performing an iterative operation on each element in the zero-sequence voltage matrix according to a preset K-means algorithm to obtain an element centroid set;

[0048] A fault degree calculation module for calculating the fault degree of the initial cluster set and screening to obtain the fault cluster with the largest fault degree;

[0049] A centroid selection module for calculating the membership degree of the fault cluster to each element centroid in the element centroid set and screening to obtain the fault centroid with the largest membership degree;

[0050] A fault positioning module for performing positioning on the distribution network node map according to the fault centroid to implement distribution network fault positioning.

[0051] In a third aspect, the present invention further provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a distribution network fault positioning method according to any one of the above.

[0052] In a fourth aspect, the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned distribution network fault location methods based on zero-sequence voltage distribution characteristics.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] The invention discloses a distribution network fault location method based on zero-sequence voltage distribution characteristics, comprising obtaining a distribution network node diagram and a node zero-sequence voltage; constructing a matrix according to the node zero-sequence voltage to obtain a zero-sequence voltage matrix; performing a clustering operation on the zero-sequence voltage matrix to obtain an initial cluster set; performing an iterative operation on each element in the zero-sequence voltage matrix according to a preset K-means algorithm to obtain an element centroid set; calculating the fault degree of the initial cluster set, and screening to obtain a fault cluster with the largest fault degree; calculating the membership degree of the fault cluster and the centroid of each element in the element centroid set, and screening to obtain the fault centroid with the largest membership degree; and performing positioning on the distribution network node diagram according to the fault centroid to realize distribution network fault location.

[0055] Compared with the inaccurate positioning in the fault location process of the prior art, the present invention utilizes the characteristic that the zero-sequence voltage of each node of the fault section presents a quasi-normal distribution when a distribution network fault occurs, abstracts the section into a zero-sequence voltage distribution matrix, and clusters the zero-sequence voltage distribution matrix using the membership function to obtain various clusters and the centroids of various clusters. Then, the optimal classification centroid is obtained through iterative processing, thereby achieving accurate classification of all nodes of the fault section. At the same time, the correlation coefficient between the nodes of each cluster and the node where the fault occurs is used to calculate the fault degree of each cluster, thereby quickly and accurately achieving fault location. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic flow chart of a method for locating a distribution network fault based on zero-sequence voltage distribution characteristics provided by the first embodiment of the present invention;

[0057] Figure 2 It is a structural schematic diagram of a distribution network fault location device based on zero-sequence voltage distribution characteristics provided by the second embodiment of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] Referring to Figure 1 , the first embodiment of the present invention provides a method for locating faults in a distribution network based on the characteristics of zero-sequence voltage distribution, including the following steps:

[0060] S11, Obtain the distribution network node diagram and the zero-sequence voltage of the nodes;

[0061] S12, Construct a matrix based on the zero-sequence voltage of the nodes to obtain a zero-sequence voltage matrix;

[0062] S13, Perform a clustering operation on the zero-sequence voltage matrix to obtain an initial set of clusters;

[0063] S14, Perform an iterative operation on each element in the zero-sequence voltage matrix according to the preset K-means algorithm to obtain a set of element centroids;

[0064] S15, Calculate the fault degree of the initial set of clusters, and screen to obtain the fault cluster with the largest fault degree;

[0065] S16, Calculate the membership degree of the fault cluster to each element centroid in the set of element centroids, and screen to obtain the fault centroid with the largest membership degree;

[0066] S17, Locate on the distribution network node diagram according to the fault centroid to achieve fault location in the distribution network.

[0067] In step S11, obtain the distribution network node diagram and the zero-sequence voltage of the nodes.

[0068] It should be noted that the distribution network node diagram is obtained by a fault recorder recording the change waveforms of various electrical quantities (such as voltage, current, frequency, etc.) during a power system fault. By analyzing the fault recording data and combining the topological structure and electrical parameters of the system, the time, type of the fault, and the lines and nodes involved can be determined. Using professional fault recording analysis software, the analysis results can be displayed in a graphical manner. When a fault occurs, it can be located on the distribution network node diagram.

[0069] Specifically, according to the electrical wiring diagram and equipment connection relationship of the power system, a topological model of the system is established. The nodes in the distribution network node diagram usually represent various components or positions in the system. For example, in the power system, they can represent devices such as transformers, substations, and generators, and in the network system, they can represent devices such as servers, routers, and switches. The connection methods and electrical relationships between each node (such as busbars, transformers, line endpoints, etc.) are clarified. Based on the collected fault information and combined with the system topological model, by analyzing the action logic of protection devices and the switch tripping situation, the area and specific nodes where the fault occurs are determined, and the fault nodes are clearly marked on the topological diagram, such as highlighting them with special colors, symbols, or marks.

[0070] Exemplarily, when a fault occurs in the system, the area where the fault may occur can be quickly located by analyzing the distribution network node diagram. For example, if the connection between certain nodes is interrupted or a node shows abnormalities, these nodes and their adjacent nodes can be observed in the diagram, thereby narrowing down the fault range. By observing the connection relationship between the nodes in the distribution network node diagram, the propagation path and range of the fault can be predicted. The connection relationship between nodes will cause the fault to spread from one node to another. By analyzing the edges in the diagram, the other nodes affected by the fault can be evaluated.

[0071] It should be noted that the zero-sequence voltage of the node needs to be obtained through a zero-sequence voltage transformer or a voltage sensor. Exemplarily, in the power system, the zero-sequence voltage can be measured at the open delta of the secondary side of the voltage transformer with a star connection grounded at the neutral point. During normal operation, the three-phase voltages are symmetrical, and the voltage at the open delta is zero. When an asymmetric fault such as a ground fault occurs in the system, a zero-sequence voltage will appear, and the value of the zero-sequence voltage can be directly read through a voltmeter or other measuring devices connected to the open delta.

[0072] Specifically, under normal circumstances, the amplitude of the zero-sequence voltage of the node is within a certain range. When a fault occurs, for example, a ground fault or a short-circuit fault, the amplitude of the zero-sequence voltage will change significantly and exceed the normal range. By monitoring the amplitude of the zero-sequence voltage, it can be used as an important indicator of the occurrence of a fault. Different positions or different types of faults will cause changes in the phase of the zero-sequence voltage of the node. By comparing the phase differences of the zero-sequence voltages of different nodes, the location and type of the fault can be further determined.

[0073] Specifically, clarify the current operation mode of the power system, including power source distribution, load conditions, and line switching status. Collect the electrical parameters of each component (such as generators, transformers, lines, etc.) in the system, such as resistance, reactance, transformation ratio, etc. According to the actual structure of the power system and component parameters, establish a zero-sequence equivalent network according to the flow path of zero-sequence current. In the zero-sequence network, only consider the action of zero-sequence components, and equivalently connect components such as power sources, neutral-point grounding devices, and lines in the system according to their zero-sequence impedances. According to the fault type and system boundary conditions, use fault analysis methods (such as the symmetrical component method) to calculate the zero-sequence current distribution in the system. According to the calculated zero-sequence current distribution, combined with the zero-sequence impedances of each component in the zero-sequence network, use Ohm's law and Kirchhoff's laws to solve the zero-sequence voltage of each node.

[0074] In step S12, construct a matrix based on the node zero-sequence voltage to obtain a zero-sequence voltage matrix.

[0075] Number the fault nodes to obtain node numbers;

[0076] Record the sampling time of the fault nodes to obtain sampling time points;

[0077] Use the node numbers as the rows of the matrix, the sampling time points as the columns of the matrix, and the node zero-sequence voltage at the corresponding moment as the element values of the matrix to construct a matrix, obtaining an initial voltage matrix;

[0078] Normalize the initial voltage matrix to obtain a zero-sequence voltage matrix.

[0079] It should be noted that by analyzing the topological structure of the system, count all bus nodes, transformer nodes, line connection nodes, etc. to obtain the number of nodes, and collect the node zero-sequence voltage values and corresponding sampling moments in real time.

[0080] Specifically, by observing the elements of a certain row, the change of the zero-sequence voltage of this node over time can be understood, so as to judge whether this node is operating normally. Observing a certain column can understand the zero-sequence voltage distribution of the entire system at this moment, which helps to judge whether the system has an overall abnormality at a certain moment. The magnitude of the element value can reflect the severity of the fault, and a larger zero-sequence voltage amplitude may indicate a more serious fault.

[0081] Exemplarily, assume that 3 fault nodes are detected in the power system, numbered 1, 2, and 3. During the fault, samples are taken every 1 second, and a total of 3 time points are recorded: t = 0s, t 2 = 1s, t 3 = 2s, for node 1: at t 1 = 0s it is 0.5 kV, at t 2 = 1s it is 0.6 kV, at t3 is 0.7 kV at t = 2 s. Node 2: t 1 is 0.4 kV at t = 0 s, t 2 is 0.45 kV at t = 1 s, t 3 is 0.5 kV at t = 2 s. Node 3: t 1 is 0.3 kV at t = 0 s, t 2 is 0.35 kV at t = 1 s, t 3 is 0.4 kV at t = 2 s.

[0082] Taking the node numbers as rows, the time points as columns, and filling the voltage values into the matrix, we get:

[0083]

[0084] It should be noted that the initial voltage matrix is normalized to obtain the zero-sequence voltage matrix. There are methods such as linear normalization and standard deviation normalization. The goal of normalization is to map the elements in the initial voltage matrix to a specific interval, such as the [0, 1] interval, which can facilitate subsequent analysis and processing.

[0085] Exemplarily, taking the linear normalization method as an example, find the maximum and minimum values in the matrix. By scanning the matrix row by row and column by column, compare each element with the currently recorded maximum and minimum values, and update the maximum and minimum values. Recombine all the elements obtained after the normalization calculation in the original matrix row and column order to form the normalized zero-sequence voltage matrix. The elements in this matrix are all within the [0, 1] interval, and the relative magnitude relationship between the elements in the original initial voltage matrix is retained, which is convenient for subsequent processing and use.

[0086] In step S13, perform a clustering operation on the zero-sequence voltage matrix to obtain an initial cluster set.

[0087] Through the clustering algorithm, select multiple cluster centers from the elements of the zero-sequence voltage matrix, where the number of cluster centers is the same as the number of nodes;

[0088] Calculate the distance between each element in the zero-sequence voltage matrix and the cluster center, and assign the element to the cluster represented by the cluster center with the closest distance;

[0089] Repeat the step of calculating the distance between each element in the zero-sequence voltage matrix and the cluster center and assigning the element to the cluster represented by the cluster center with the closest distance until all elements in the zero-sequence voltage matrix are assigned, obtaining an initial cluster set.

[0090] It should be noted that for the clustering operation of the zero-sequence voltage matrix, some clustering algorithms are used, such as K-means, hierarchical clustering, DBSCAN, etc. In the embodiment of the present invention, the K-means clustering algorithm is taken as an example to illustrate its process and related features.

[0091] It should be noted that multiple clustering centers are selected from the elements of the zero-sequence voltage matrix, where the number of clustering centers is the same as the number of nodes. In the zero-sequence voltage matrix, K different elements are randomly selected as the initial clustering centers. This is achieved by generating K groups of random row and column indices and then extracting the corresponding elements from the matrix. Exemplarily, elements with a relatively large distance are selected as the initial centroids to improve the clustering effect and stability. For example, first select an element in the matrix as the first centroid, and then select the element with the farthest distance from the first centroid as the second centroid, and so on until K centroids are selected.

[0092] It should be noted that the initial cluster set is a set formed by clustering elements with similar features through a clustering algorithm. They group the elements in the system, facilitating the analysis and management of the system. When a fault occurs, the search range can be narrowed, and the approximate location of the fault can be judged from the level of the element clusters. The centroids of different element clusters can be compared with each other to help distinguish different system states or fault states. The differences between different centroids can help judge different behavior patterns of the system.

[0093] It should be noted that for each element in the zero-sequence voltage matrix, the distance between it and the K centroids is calculated to represent the membership degree of each element to the initial classification centroids. The distance calculation can use distance measurement methods such as Euclidean distance and Manhattan distance.

[0094] Exemplarily, the calculation formula for Euclidean distance is:

[0095]

[0096] where x ij is an element in the zero-sequence voltage matrix, c k is the initial centroid, d(x ij , c k ) is the Euclidean distance, and l is the element dimension.

[0097] Specifically, each element is assigned to the cluster corresponding to the centroid closest to it. That is, for the element x ij , if the distance between it and the centroid c k is the smallest among the distances from it to all centroids, then x ij is assigned to the kth cluster.

[0098] It should be noted that by analyzing the obtained element clusters, the elements within each element cluster are somewhat similar to each other. The clustering results can be understood by analyzing the characteristics of the elements within the cluster (such as the average zero-sequence voltage, voltage change trend, etc.). For example, different element clusters may correspond to different fault types or power grid operating states.

[0099] Exemplarily, the nodes of the power system are clustered according to the zero-sequence voltage. The centroid of an element cluster can reflect the average zero-sequence voltage amplitude and phase change trend of the nodes belonging to this cluster at different times. By observing the characteristics of the centroid, the commonalities of the elements within this cluster can be understood, thereby judging the system state or fault state represented by this cluster.

[0100] In step S14, iterative operations are performed on each element in the zero-sequence voltage matrix according to the preset K-means algorithm to obtain a set of element centroids.

[0101] According to the K-means algorithm, randomly select K elements in the zero-sequence voltage matrix as the initial element centroids, where K is equal to the number of nodes;

[0102] Calculate the membership degrees of each element in the zero-sequence voltage matrix to the initial element centroids;

[0103] If the membership degree is positive, it indicates that the initial element centroid can better represent this node. If the membership degree is negative, replace the current element with the initial element centroid;

[0104] Return to the step of calculating the membership degrees of each element in the zero-sequence voltage matrix to the initial element centroids to implement the operation of updating the element centroids. When all the elements in the node have completed the calculation of the membership degrees, the iteration ends, and the output result is a set of element centroids composed of K element centroids.

[0105] It should be noted that the initial element centroid is the representative of the elements in the node and reflects the typical characteristics of the elements in this node. In fault diagnosis, it provides a reference central value. When it is used as the fault centroid, we can quickly judge the location of the fault by comparing it with the centroid under normal conditions. For example, the area where parameters such as voltage and current deviate significantly from the normal range.

[0106] It should be noted that when randomly selecting K different data points from the dataset as the initial element centroids, the K-means algorithm will try to select initial element centroids that are far apart to avoid the local optimum problem caused by improper selection of the initial element centroids. Exemplarily, when selecting several load points as the initial points in a distribution network, K-means will try to distribute them in different regions of the distribution network rather than concentrating in a small area. The update of the centroid is achieved by calculating the average value of the elements within the cluster. As the number of iterations increases, the centroid will gradually adjust its position to make it closer to the center position of the elements within the cluster.

[0107] Specifically, assign a weight to each initial centroid, which can be equal, for example, all being 1 / K. At the beginning, the importance of each centroid in subsequent calculations is the same. For each element in the zero-sequence voltage matrix, after calculating the membership degree to each initial centroid, cluster classification is performed.

[0108] It should be noted that when the number of calculated elements increases, the initial centroid and the initial weight need to be updated. Update the weight according to the membership degree, which can reflect the "voting" situation of the elements for different centroids. The higher the membership degree a centroid obtains, the greater its weight. At the same time, the centroid will move towards the direction where the elements are dense. Therefore, update the centroid and re-iterate until the preset number of iterations or the preset centroid threshold for element classification is reached, and then the set of element centroids can be output.

[0109] Exemplarily, for the clustering of the zero-sequence voltage matrix, the initial element centroids may be randomly selected, which may lead to misclassification of the zero-sequence voltage information of some nodes. By iteratively updating the centroid, those nodes that were initially misclassified can be re-assigned to more appropriate clusters, making the clustering result more in line with the actual situation and improving the classification accuracy of the node status.

[0110] Specifically, the set of finally output element centroids is the centroids that have stabilized after multiple iterations. They represent the final positions of different clusters. According to the final membership degree, the elements can be assigned to the most suitable clusters. By continuously iteratively updating the centroid and the membership degree, the centroid finally reaches a stable position, realizing the clustering of the data and dividing the data into different categories, which is convenient for subsequent analysis and processing.

[0111] The membership degree is calculated by the following formula:

[0112]

[0113] where is the element x of the zero-sequence voltage distribution matrix j and the membership degree value to each initial centroid μ k , σ is the preset standard deviation, x jis an element of the zero-sequence voltage distribution matrix, μ k is the initial centroid, and K is the number of clusters.

[0114] In step S15, calculate the fault degree of the initial cluster set and filter to obtain the fault cluster with the maximum fault degree.

[0115] Calculate the fault degree through the following formula:

[0116]

[0117] where Fault Degree k is the fault degree, R i is the fault degree of the i-th node, M is the total number of nodes in the cluster of similar elements, α i is the weight coefficient, GNN i is the relationship feature of node i, r is the sum of the correlation coefficients between the zero-sequence voltage of the fault-occurring node on each phase line and the zero-sequence voltage of the nodes in the cluster of similar elements on each phase line, r 0 is the average value of the correlation coefficients between the zero-sequence voltage of the fault-occurring node on each phase line and the zero-sequence voltage of the nodes in the cluster of similar elements on each phase line.

[0118] It should be noted that the fault degree is an index to measure the possibility of a cluster of elements having a fault. Its importance lies in being able to sort the clusters of elements and find the cluster of elements most likely to have a fault. Through the quantified fault degree, the clusters of elements with a high fault degree can be processed preferentially, improving the efficiency of fault diagnosis and handling.

[0119] It should be noted that the data of the cluster of similar elements includes the relevant information of each node in the cluster. This involves obtaining the relationship features of each node and collecting the zero-sequence voltage data of the fault-occurring node and the nodes in the cluster of similar elements on each phase line. For example, obtaining the zero-sequence voltage time series data of the nodes from the monitoring database of the power system.

[0120] It should be noted that calculate the correlation coefficients between the zero-sequence voltage of the fault-occurring node and each node in the cluster of similar elements on each phase line. The correlation coefficient reflects the degree of closeness between the fault-occurring node and each node in the cluster of similar elements. For example, using the Pearson correlation coefficient calculation method, which is an index to measure the linear correlation between two variables.

[0121] Specifically, after obtaining the correlation coefficients between all nodes and the fault-occurring node, calculate the average value of these correlation coefficients. An overall average correlation coefficient can be obtained, reflecting the average degree of association of the entire cluster of elements with respect to the fault-occurring node in terms of zero-sequence voltage.

[0122] Specifically, after calculating the fault degrees of each node within the element cluster, sum up the fault degrees to obtain the total fault degree of the element cluster.

[0123] Specifically, set the fault degree of the first element cluster as the current maximum fault degree, and then compare it with the fault degrees of other element clusters in turn. If a higher fault degree is found, update the current maximum fault degree and its corresponding element cluster.

[0124] In step S16, calculate the membership degrees of the fault cluster with each element centroid in the element centroid set, and screen out the fault centroid with the maximum membership degree.

[0125] Calculate the membership degree through the following formula:

[0126]

[0127] where c max is the fault centroid with the maximum membership degree, F is the fault cluster, n is the number of element centroids in the element centroid set, c is the element centroid set, D is the data point space dimension, and f is the element coordinates within the fault cluster.

[0128] It should be noted that the fault centroid is the optimal classification centroid of the element cluster that is most likely to have a fault finally selected, and it is the core reference for fault location and handling. Infer the location of the fault based on its characteristics. When a fault occurs, it can narrow down the search range and judge the approximate location of the fault from the node level.

[0129] Specifically, after calculating the fault degrees of all fault clusters, compare these fault degrees. The traversal method can be used. Set the fault degree of the first element cluster as the current maximum fault degree, and then compare it with the fault degrees of other element clusters in turn. If a higher fault degree is found, update the current maximum fault degree and its corresponding element cluster.

[0130] Specifically, for each fault cluster, the element centroid set has been calculated through the K-means algorithm or other clustering algorithms. The element centroid set is the representative of each element cluster. It summarizes the characteristics of the element cluster and is the central position or average state of the elements within the cluster.

[0131] Specifically, after finding the fault cluster with the maximum fault degree, calculate the membership degrees of the fault cluster with each element centroid, and incorporate the fault cluster under a certain element centroid. The element centroid represents the fault cluster and realizes more accurate node fault location. The fault centroid can be regarded as the representative of the area most likely to have a fault and is the key reference point for subsequent fault location, analysis, and handling.

[0132] Exemplarily, in a power system, if a certain element cluster is determined to have the maximum degree of fault, then its optimal classification centroid reflects the typical voltage and current states of the node group most likely to have a fault. Based on this fault centroid, the cause of the fault can be further analyzed, such as whether the voltage is too high or too low, or whether the current is abnormal, etc.

[0133] In step S17, based on the fault centroid, positioning is performed on the distribution network node diagram to achieve distribution network fault location.

[0134] The fault centroid is an element within the zero-sequence voltage matrix. The fault node is obtained based on the node where the fault centroid is located;

[0135] Based on the fault node, positioning is performed on the distribution network node diagram to achieve distribution network fault point location.

[0136] It should be noted that through the clustering operation, several clusters of similar elements are obtained. For each cluster, the degree of fault is calculated using the selected fault degree calculation method, and the calculated degrees of fault of each cluster are compared to find the cluster with the maximum degree of fault.

[0137] It is worth noting that the fault centroid is a vector representing the central characteristics of the elements within the cluster, and its elements include the comprehensive zero-sequence voltage information of the elements within the cluster and the average value of the zero-sequence voltage amplitudes of the elements within the cluster.

[0138] Specifically, a mapping relationship from the fault centroid information to the physical nodes of the distribution network is established: Since the elements were arranged in the order of nodes when constructing the matrix before, the corresponding physical node number or name can be found according to the position of the fault centroid element in the matrix, providing higher efficiency for fault location. Based on the established mapping relationship, the position of the fault centroid corresponding to the cluster with the maximum degree of fault in the matrix is determined, and thus the physical node number corresponding to this position is found.

[0139] Exemplarily, the positioning result is compared with other fault detection indicators such as overcurrent protection action information and differential protection information, or the historical fault data of this node is viewed to see if there are similar fault patterns or frequent fault records. If these indicators also indicate that this node or its adjacent nodes may have a fault, it means that this node has a fault, thus achieving more accurate fault location.

[0140] It is worth noting that when the verification result shows that the fault location is inaccurate, it is necessary to recheck the clustering algorithm, the fault degree calculation method, the mapping relationship from elements to physical nodes, etc., and it is necessary to readjust the number of clusters for clustering or modify the weight coefficients in the fault degree calculation, etc.

[0141] The working process of the present invention is described below by taking a relatively common scenario as an example. Please also refer to Figure 2 , which is Figure 1Schematic diagram of the working scenario of the method.

[0142] Step 1: The urban power system operates normally. A fault occurs at a certain point. The sensing zero-sequence voltage meter obtains the real-time node zero-sequence voltage generated by the fault and transmits the node zero-sequence voltage to the system. At the same time, the distribution network node diagram is obtained according to the relay protection action.

[0143] Step 2: Count all bus nodes, transformer nodes, line connection nodes, etc. to obtain the number of nodes. Real-time collect the node zero-sequence voltage values and the corresponding sampling times, construct a matrix and initialize to obtain the zero-sequence voltage matrix.

[0144] Step 3: Through the clustering algorithm, select multiple clustering centers from the elements of the zero-sequence voltage matrix, and calculate the distance between each element in the zero-sequence voltage matrix and the clustering centers. Assign the elements to the clusters represented by the nearest clustering centers to obtain the initial cluster set.

[0145] Step 4: According to the K-means algorithm, randomly select K elements in the zero-sequence voltage matrix as the initial element centroids, and then calculate the membership degrees of each element in the zero-sequence voltage matrix to the initial element centroids. When all the elements in the node have completed the operation of calculating the membership degrees, the iteration ends, and the output result is the element centroid set composed of K element centroids.

[0146] Step 5: Calculate the fault degree of the initial cluster set, and select the fault cluster with the largest fault degree.

[0147] Step 6: Calculate the membership degrees of the fault cluster to each element centroid in the element centroid set, and screen to obtain the fault centroid with the largest membership degree.

[0148] Step 7: Obtain the fault node according to the node where the fault centroid is located; locate according to the fault node on the distribution network node diagram to realize the location of the fault point in the distribution network.

[0149] In summary, the present invention discloses a distribution network fault location method based on the zero-sequence voltage distribution characteristics, including obtaining the distribution network node diagram and node zero-sequence voltage; constructing a matrix according to the node zero-sequence voltage to obtain the zero-sequence voltage matrix; performing clustering operation on the zero-sequence voltage matrix to obtain the initial cluster set; performing iterative operation on each element in the zero-sequence voltage matrix according to the preset K-means algorithm to obtain the element centroid set; calculating the fault degree of the initial cluster set and screening to obtain the fault cluster with the largest fault degree; calculating the membership degrees of the fault cluster to each element centroid in the element centroid set and screening to obtain the fault centroid with the largest membership degree; locating according to the fault centroid on the distribution network node diagram to realize the distribution network fault location.

[0150] In view of the inaccurate fault location in the prior art during the fault location process, the present invention utilizes the characteristic that the zero-sequence voltages of each node in the fault section present a quasi-normal distribution when a fault occurs in the distribution network. The section is abstracted into a zero-sequence voltage distribution matrix, and the membership function is used to cluster the zero-sequence voltage distribution matrix. After obtaining various clusters and the centroids of various clusters, the optimal classification centroid is obtained through iterative processing, realizing the accurate classification of all nodes in the fault section. At the same time, by using the correlation coefficient between the nodes of each cluster and the fault occurrence node, the fault degree of each cluster is calculated, and the fault location is quickly and accurately realized.

[0151] Referring to Figure 2 , the second embodiment of the present invention provides a distribution network fault location device based on the zero-sequence voltage distribution characteristics, including:

[0152] A data acquisition module, configured to acquire a distribution network node map and node zero-sequence voltages;

[0153] A matrix construction module, configured to construct a matrix according to the node zero-sequence voltages to obtain a zero-sequence voltage matrix;

[0154] A clustering operation module, configured to perform a clustering operation on the zero-sequence voltage matrix to obtain an initial cluster set;

[0155] A centroid iteration module, configured to perform an iterative operation on each element in the zero-sequence voltage matrix according to the preset K-means algorithm to obtain an element centroid set;

[0156] A fault degree calculation module, configured to calculate the fault degree of the initial cluster set and screen to obtain the fault cluster with the largest fault degree;

[0157] A centroid selection module, configured to calculate the membership degree of the fault cluster to each element centroid in the element centroid set and screen to obtain the fault centroid with the largest membership degree;

[0158] A fault location module, configured to perform location on the distribution network node map according to the fault centroid to realize the distribution network fault location.

[0159] It should be noted that a distribution network fault location device based on the zero-sequence voltage distribution characteristics provided in the embodiment of the present invention is used to execute all the process steps of a distribution network fault location method based on the zero-sequence voltage distribution characteristics in the above embodiment. Their working principles and beneficial effects correspond one by one, and thus will not be elaborated herein.

[0160] An embodiment of the present invention also provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a distribution network fault location program based on the zero-sequence voltage distribution characteristics. When the processor executes the computer program, the steps in the above embodiments of each distribution network fault location method based on the zero-sequence voltage distribution characteristics are implemented, such as Figure 1 step S11 shown. Alternatively, when the processor executes the computer program, the functions of each module / unit in the above device embodiments are implemented, such as the data acquisition module.

[0161] Exemplarily, the computer program may be divided into one or more modules / units, and the one or more modules / units are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the electronic device.

[0162] The electronic device may be a computing device such as a desktop computer, a notebook, a palm computer, and a smart tablet. The electronic device may include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the above components are only examples of the electronic device, and do not constitute a limitation on the electronic device. It may include more or fewer components than the above, or combine certain components, or different components. For example, the electronic device may further include input / output devices, network access devices, buses, etc.

[0163] The so-called processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor, or the processor may also be any conventional processor, etc. The processor is the control center of the electronic device, and connects various parts of the entire electronic device through various interfaces and lines.

[0164] The memory can be used to store the computer programs and / or modules. By running or executing the computer programs and / or modules stored in the memory, and by invoking the data stored in the memory, the processor implements various functions of the electronic device. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function (such as a sound playback function, an image playback function, etc.); the data storage area can store data created according to the use of the mobile phone (such as audio data, phone book, etc.). In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage devices.

[0165] Among them, if the modules / units integrated in the electronic device are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above-mentioned embodiment methods of the present invention, it can also be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0166] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the attached drawings of the device embodiments provided by the present invention, the connection relationship between modules indicates that they have a communication connection, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0167] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. In particular, for those skilled in the art, any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A distribution network fault location method based on zero-sequence voltage distribution characteristics, characterized in that: Executed by a computer, including: Obtain distribution network node diagram and node zero-sequence voltage; Constructing a matrix according to the node zero-sequence voltage to obtain a zero-sequence voltage matrix; Performing a clustering operation on the zero-sequence voltage matrix to obtain an initial cluster set; Performing an iterative operation on each element in the zero-sequence voltage matrix according to a preset K-means algorithm to obtain a set of element centroids; Calculating the fault degree of the initial cluster set, and screening out the fault cluster with the largest fault degree; Calculating the membership degree of the fault cluster and the centroid of each element in the element centroid set, and screening out the fault centroid with the largest membership degree; According to the fault centroid, positioning is performed on the distribution network node diagram to achieve distribution network fault positioning.

2. A distribution network fault location method based on zero-sequence voltage distribution characteristics according to claim 1, characterized in that: A matrix is ​​constructed according to the node zero-sequence voltage to obtain a zero-sequence voltage matrix, including: Number the faulty nodes to obtain node numbers; Record the sampling time of the fault node to obtain the sampling time point; The node numbers are used as matrix rows, the sampling time points are used as matrix columns, and the node zero-sequence voltages of the nodes at the corresponding moments are used as matrix element values ​​to construct a matrix, thereby obtaining an initial voltage matrix; The initial voltage matrix is ​​normalized to obtain a zero-sequence voltage matrix.

3. A distribution network fault location method based on zero-sequence voltage distribution characteristics according to claim 1, characterized in that: A clustering operation is performed on the zero-sequence voltage matrix to obtain an initial cluster set, including: A plurality of cluster centers are selected from the elements of the zero-sequence voltage matrix by a clustering algorithm, wherein the number of cluster centers is the same as the number of nodes; Calculating the distance between each element in the zero-sequence voltage matrix and the cluster center, and assigning the element to the cluster represented by the cluster center with the closest distance; Repeat the steps of calculating the distance between each element in the zero-sequence voltage matrix and the cluster center, and assigning the element to the cluster represented by the cluster center with the closest distance, until all elements in the zero-sequence voltage matrix are assigned, and obtain an initial cluster set.

4. A distribution network fault location method based on zero-sequence voltage distribution characteristics according to claim 1, characterized in that: An iterative operation is performed on each element in the zero-sequence voltage matrix according to a preset K-means algorithm to obtain a set of element centroids, including: According to the K-means algorithm, K elements are randomly selected in the zero-sequence voltage matrix as the initial element centroids, where K is equal to the number of nodes; Calculating the membership degree of each element in the zero-sequence voltage matrix to the centroid of the initial element; If the membership is positive, it means that the initial element centroid is more representative of the node. If the membership is negative, the current element is replaced with the initial element centroid. Return to the step of calculating the membership of each element in the zero-sequence voltage matrix to the initial element centroid, implement the operation of updating the element centroid, when all the membership calculation operations of the elements in the node are completed, the iteration ends, and the output result is an element centroid set consisting of K element centroids.

5. A distribution network fault location method based on zero-sequence voltage distribution characteristics according to claim 4, characterized in that: Calculating the membership of each element in the zero-sequence voltage matrix to the centroid of the initial element includes: The membership is calculated by the following formula: in, is the element x of the zero-sequence voltage distribution matrix j With each initial center of mass μ k The membership value of j is the element of the zero-sequence voltage distribution matrix, μ k is the initial centroid, and K is the number of clusters.

6. A distribution network fault location method based on zero-sequence voltage distribution characteristics according to claim 1, characterized in that: Calculating the fault degree of the initial cluster set and screening out the fault cluster with the largest fault degree, including: The fault degree is calculated by the following formula: Among them, Fault Degree k is the fault degree, R i is the fault degree of the i-th node, M is the total number of nodes in the same element cluster, α i is the weight coefficient, GNN i is the relationship characteristic of node i, r is the sum of the correlation coefficients between the zero-sequence voltage of the fault node on each phase line and the zero-sequence voltage of the same type of element cluster nodes on each phase line, and r0 is the average value of the correlation coefficients between the zero-sequence voltage of the fault node on each phase line and the zero-sequence voltage of the same type of element cluster nodes on each phase line.

7. A distribution network fault location method based on zero-sequence voltage distribution characteristics according to claim 1, characterized in that: Calculating the membership of the fault cluster and the centroid of each element in the element centroid set, and screening out the fault centroid with the largest membership, including: The membership is calculated by the following formula: Among them, c max is the fault centroid with the largest membership, F is the fault cluster, n is the number of element centroids in the element centroid set, c is the element centroid set, D is the spatial dimension of the data point, and f is the element coordinate in the fault cluster.

8. A distribution network fault location method based on zero-sequence voltage distribution characteristics according to claim 1, characterized in that: According to the fault centroid, positioning is performed on the distribution network node graph to achieve distribution network fault positioning, including: The fault centroid is an element in the zero-sequence voltage matrix, and the fault node is obtained according to the node where the fault centroid is located; The fault node is located on the distribution network node diagram to locate the distribution network fault point.

9. A distribution network fault location device based on zero-sequence voltage distribution characteristics, characterized in that: include: A data acquisition module is used to obtain a distribution network node diagram and node zero-sequence voltage; A matrix construction module, used for constructing a matrix according to the node zero-sequence voltage to obtain a zero-sequence voltage matrix; A clustering operation module, used for performing a clustering operation on the zero-sequence voltage matrix to obtain an initial cluster set; A centroid iteration module, used for iterating each element in the zero-sequence voltage matrix according to a preset K-means algorithm to obtain a centroid set of elements; A fault degree calculation module is used to calculate the fault degree of the initial cluster set and screen out the fault cluster with the largest fault degree; A centroid selection module, used to calculate the membership of the fault cluster and the centroids of each element in the element centroid set, and screen out the fault centroid with the largest membership; The fault location module is used to locate the fault on the distribution network node diagram according to the fault centroid, so as to realize the distribution network fault location.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute a distribution network fault location method based on zero-sequence voltage distribution characteristics as described in any one of claims 1 to 8.

Images