A high resistance grounding fault protection method based on general low resistance grounding system

Abstract

The application provides a high-resistance grounding fault protection method based on a general small-resistance grounding system in the technical field of power distribution network relay protection, comprising: collecting bus zero sequence voltage and zero sequence current at outgoing lines of a power distribution network after a fault; when the bus zero sequence voltage is greater than a zero sequence voltage starting value or the difference between any two adjacent zero sequence voltage sampling values is greater than a zero sequence voltage mutation value starting value: constructing an array according to the collected zero sequence current and calculating the similarity between zero sequence currents at any two outgoing lines to obtain a similarity matrix; presetting the value range of the cluster number according to the access condition of a distributed power supply, and determining the optimal cluster number by adopting the principle of minimizing the within-cluster sum of squares; based on the similarity matrix and an improved k-means algorithm, performing cluster analysis according to the optimal cluster number to obtain a cluster analysis result, judging the fault type and selecting a fault line. The application solves the problem of manually setting a setting value in traditional high-resistance protection.

Description

Technical Field

[0001] The invention relates to a high-resistance grounding fault protection method based on a universal low-resistance grounding system, and belongs to the technical field of distribution network relay protection. Background Art

[0002] With the modernization of urban distribution networks, the proportion of hybrid cable-overhead lines and pure cable lines has increased significantly, prompting the increasing use of low-resistance grounding systems in distribution networks. Low-resistance grounding systems can effectively limit overvoltage levels during single-phase ground faults, ensuring the safe and stable operation of the power grid. Currently, the mainstream protection strategy for low-resistance grounding systems in distribution networks is zero-sequence current protection. This method uses the zero-sequence current generated during a fault to detect and locate faults, and its effectiveness is particularly evident in situations where the grounding resistance is low and the fault zero-sequence current is large. To overcome the limitations of zero-sequence overcurrent protection in certain situations, such as when large grounding resistance makes the fault current less apparent, researchers have recently proposed a new protection method based on the difference in zero-sequence electrical quantities, including phase and amplitude, between the fault line and the non-fault line. By analyzing the phase relationship of steady-state electrical quantities, this method can avoid the transient process of the zero-sequence component during data acquisition and set appropriate protection settings accordingly, significantly improving the accuracy of single-phase ground fault detection and its tolerance to transient resistance.

[0003] Although a novel protection method based on the phase-amplitude relationship between zero-sequence electrical quantities on faulted and non-faulted lines has achieved significant progress in detecting single-phase ground faults in low-resistance grounded systems, this technology still has some limitations. First, this method primarily focuses on the characteristics of traditional distribution networks and does not fully consider the new challenges posed by the integration of large numbers of renewable energy sources into distribution networks. The integration of renewable energy changes the power flow distribution and fault characteristics of distribution networks, potentially leading to more complex variations in zero-sequence electrical quantities, which in turn affects the accuracy and reliability of protection. Second, while this method performs well under steady-state conditions, it still faces certain difficulties in handling transient fault processes. The rapid changes in zero-sequence components during transient processes can make it difficult to determine the phase-amplitude relationship, thus affecting the rapid response capability of protection. Furthermore, with the increasing complexity and intelligence of distribution network structures, the requirements for adaptability and intelligence in protection methods are also increasing, and existing methods still need further improvement and refinement in these aspects.

[0004] The information disclosed in this background section is only intended to enhance understanding of the overall background of the invention and should not be considered as an admission or any form of suggestion that the information constitutes the prior art already known to a person of ordinary skill in the art. Summary of the Invention

[0005] The technical problem to be solved by the present invention is: how to solve the problem that the setting value needs to be manually set in traditional high-resistance protection.

[0006] In order to solve the above technical problems, the present invention is implemented by adopting the following technical solutions.

[0007] This paper proposes a universal high-resistance grounding fault protection method for low-resistance grounding systems based on an improved clustering algorithm and zero-sequence current similarity. It does not require manual setting of the protection setting value, and does not need to avoid the zero-sequence component transient process in the fault acquisition link. The data acquisition time window is short and it can be applied to distribution networks with multiple outgoing renewable energy access and without renewable energy access.

[0008] The present invention provides a high-resistance ground fault protection method based on a universal low-resistance grounding system, comprising:

[0009] Collect the busbar zero-sequence voltage and the zero-sequence current at each outgoing line of the distribution network after a fault occurs;

[0010] When the bus zero-sequence voltage is greater than the zero-sequence voltage starting value or the difference between any two adjacent zero-sequence voltage sampling values ​​is greater than the zero-sequence voltage mutation starting value, the following steps are performed:

[0011] An array is constructed based on the collected zero-sequence currents, and the similarity between the zero-sequence currents at any two outgoing lines is calculated to obtain a similarity matrix containing the characteristic information of the fault line and the characteristic information of the non-fault line;

[0012] The range of the number of clusters is preset according to the access situation of distributed power sources, and the optimal number of clusters is determined by minimizing the sum of squares within the cluster;

[0013] Based on the similarity matrix and the improved k-means algorithm, performing cluster analysis according to the optimal number of clusters to obtain a cluster analysis result;

[0014] The fault type is determined based on the cluster analysis results and the faulty line is selected.

[0015] Furthermore, judging the fault type according to the cluster analysis result includes:

[0016] When a distributed power source is connected to a low-resistance grounded distribution network system with multiple outgoing lines:

[0017] When the number of clusters k=2, a busbar single-phase grounding fault occurs;

[0018] When the number of clusters k = 3, the smallest cluster has a single-phase grounding fault;

[0019] In a low-resistance grounded distribution network system without distributed generation:

[0020] When the number of clusters k=1, a busbar single-phase grounding fault occurs;

[0021] When the number of clusters k=2, a single-phase grounding fault occurs in the smallest cluster.

[0022] Furthermore, the method for calculating the similarity between the zero-sequence currents at any two outgoing lines includes:

[0023] The cosine distance is used to measure the similarity between the zero-sequence currents at any two outgoing lines.

[0024] Furthermore, the cosine distance is calculated based on the cosine value of the angle between the zero-sequence current vectors of any two outgoing lines in the vector space. The calculation expression of the cosine value of the angle is:

[0025] (1);

[0026] Where, is the cosine value of the angle between the zero-sequence current vectors of any two outgoing lines, Indicates the outlet Zero sequence current vector With the output line Zero sequence current vector The cosine calculation formula between Indicates the total number of outgoing lines in the distribution network, Indicates the outlet The zero sequence current vector samples, Indicates the outlet The zero sequence current vector samples.

[0027] Furthermore, the cosine distance is calculated based on the cosine value of the angle between the zero-sequence current vectors at any two outgoing lines in the vector space, wherein the closer the cosine value of the angle is to 1, the more similar the waveforms of the zero-sequence current at any two outgoing lines are; the closer the cosine value of the angle is to -1, the more opposite the waveforms of the zero-sequence current at any two outgoing lines are.

[0028] Furthermore, the mathematical expression of the similarity matrix containing the fault line characteristic information and the non-fault line characteristic information is:

[0029] (2);

[0030] Where r is the similarity matrix containing the characteristic information of the faulty line and the characteristic information of the non-faulty line, Indicates the cosine similarity of the zero-sequence current of line 1 itself, represents the cosine similarity between line 1 and line 2, represents the cosine similarity between line 1 and line n, = represents the cosine similarity between line 2 and line 1, Indicates the cosine similarity of the zero-sequence current of line 2 itself, represents the cosine similarity between line 2 and line n, = represents the cosine similarity between line n and line 1, = represents the cosine similarity between line n and line 2, represents the cosine similarity of the zero-sequence current of line n itself, where Indicates the total number of outgoing lines in the distribution network.

[0031] Furthermore, the range of values ​​of the number of clusters preset according to the access situation of distributed power sources includes: when there is no distributed power source connected to the distribution network outgoing line, the number of clusters k=1 or 2; when there is a distributed power source connected to the distribution network outgoing line, the number of clusters k=2 or 3.

[0032] Furthermore, the improved method of the k-means algorithm includes:

[0033] When the number of clusters k is equal to 1, a vector is randomly generated as the cluster center;

[0034] When the number of clusters k is equal to 2 or 3, perform the following steps to select cluster centers:

[0035] Randomly select a row vector in the similarity matrix as the initial cluster center, and the remaining vectors are candidate row vectors;

[0036] Repeat the following steps until k-1 cluster centers are selected;

[0037] sequentially calculating the shortest distances between unselected candidate row vectors in the similarity matrix and the selected cluster centers;

[0038] Based on the weighted probability distribution of the shortest distance, a candidate row vector is randomly selected from the currently unselected candidate vectors as a new cluster center.

[0039] Furthermore, the candidate row vectors that are not currently selected are calculated and the selected kth cluster center The mathematical expression of the shortest distance between is:

[0040] (3);

[0041] Where, Represents the candidate row vector that is not currently selected With the selected cluster center The shortest distance between.

[0042] Furthermore, based on the weighted probability distribution of the shortest distance, a candidate row vector is randomly selected from the currently unselected candidate row vectors as a new cluster center, wherein the probability of the currently unselected candidate row vector being selected as the new cluster center is:

[0043] (4);

[0044] Where, Indicates the probability that the currently unselected candidate row vector is selected as the new cluster center, Represents any unselected candidate row vector The shortest distance between the kth cluster center and the selected cluster center.

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

[0046] (1) The present invention can realize fault identification and fault line selection for active distribution networks and traditional distribution networks by changing the value range of cluster centers;

[0047] (2) The present invention uses cosine distance to characterize the similarity of zero-sequence current at each outgoing line of the distribution network. By taking advantage of the fact that cosine distance is insensitive to vector size, the present invention can effectively identify faults when the zero-sequence current of a high-resistance grounding line is small.

[0048] (3) The improved k-means method adopted by the present invention can effectively mine the information in the similarity matrix representing the fault characteristics by improving the selection of cluster centers. Therefore, in the data acquisition link, there is no need to avoid the transient process of zero-sequence electrical quantities, and the data acquisition time window is short. In addition, the method provided by the present invention does not require setting a threshold, which solves the problem of how to set the setting value in traditional high-resistance protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 The present invention provides a flowchart of a high-resistance grounding fault protection method based on a universal low-resistance grounding system. DETAILED DESCRIPTION

[0050] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations on the technical solution of the present invention. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0051] The term "and / or" simply describes a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists alone, A and B exist simultaneously, or B exists alone. Additionally, the character " / " generally indicates an "or" relationship between the related objects.

[0052] Example 1

[0053] like Figure 1 As shown, this embodiment introduces a high-resistance ground fault protection method based on a universal low-resistance grounding system, including:

[0054] Step 1: Collect the busbar zero-sequence voltage and the zero-sequence current at each outgoing line of the distribution network after the fault occurs.

[0055] The present invention acquires electrical quantity information after a fault occurs by collecting the zero-sequence voltage of the busbar and the zero-sequence current at each outgoing line of the distribution network, and performs fault analysis and location.

[0056] Step 2: When the bus zero-sequence voltage is greater than the zero-sequence voltage starting value or the difference between any two adjacent zero-sequence voltage sampling values ​​is greater than the zero-sequence voltage mutation starting value, perform the following steps:

[0057] The present invention uses the bus zero-sequence voltage being greater than the zero-sequence voltage starting value or the difference between any two adjacent zero-sequence voltage sampling values ​​being greater than the zero-sequence voltage mutation starting value as a starting criterion for triggering a subsequent fault diagnosis process. By setting the zero-sequence voltage starting value and the zero-sequence voltage mutation starting value, it is ensured that the fault diagnosis process is started in time when a fault occurs, avoiding unnecessary calculations and analyses in a non-fault state, and improving the efficiency and accuracy of fault diagnosis.

[0058] During normal power system operation, the three-phase system may be unbalanced, resulting in a certain amount of unbalanced zero-sequence voltage on the busbar. However, this voltage is typically no greater than 15% of the busbar phase voltage. When the busbar zero-sequence voltage exceeds the preset zero-sequence voltage pickup value, it is considered a ground fault. In certain special circumstances, such as a sudden ground fault, the zero-sequence voltage may experience a sudden change. By detecting whether the difference between two consecutive zero-sequence voltage samples is greater than the zero-sequence voltage mutation pickup value, the initial moment of the fault can be accurately captured, helping to more accurately locate the fault point.

[0059] Step 2.1: Construct an array based on the collected zero-sequence currents and calculate the similarity between the zero-sequence currents at any two outgoing lines to obtain a similarity matrix containing the characteristic information of the fault line and the characteristic information of the non-fault line.

[0060] The present invention calculates the similarity between the zero-sequence currents at any two outgoing lines and constructs a similarity matrix reflecting the electrical connections between them. This similarity matrix incorporates characteristic information about faulty and non-faulty lines, enabling cluster analysis to more accurately reflect the electrical connections and fault characteristics between the outgoing lines.

[0061] Step 2.2: Preset the range of the number of clusters according to the access situation of distributed power sources, and determine the optimal number of clusters by adopting the principle of minimizing the sum of squares within the cluster.

[0062] The present invention presets the value range of the number of clusters by considering the access situation of distributed power sources, and uses the principle of minimizing the sum of squares within the cluster to determine the optimal number of clusters, so as to ensure that the data of the similarity matrix is ​​accurately and reliably divided into a reasonable number of categories in the cluster analysis, thereby avoiding analysis errors.

[0063] Step 2.3: Based on the similarity matrix and the improved k-means algorithm, perform cluster analysis according to the optimal number of clusters to obtain cluster analysis results.

[0064] The present invention uses an improved k-means algorithm to perform cluster analysis on the similarity matrix to obtain cluster analysis results. The cluster analysis results divide each outgoing line into different clusters, each cluster representing a specific electrical connection or fault characteristic, making fault diagnosis more accurate and efficient.

[0065] Step 2.4: Determine the fault type and select the faulty line based on the cluster analysis results.

[0066] The present invention judges the fault type and selects the faulty line based on the cluster analysis results. By comparing the characteristic information of different types, the type and location of the fault can be determined, and finally the fault type judgment and faulty line selection are achieved.

[0067] Example 2

[0068] Based on the same inventive concept as Example 1, this embodiment introduces a high-resistance ground fault protection method based on a universal low-resistance grounding system, which specifically includes the following steps:

[0069] Step 1: Collect the bus zero-sequence voltage after the fault occurs and the zero-sequence current at each outgoing line of the distribution network;

[0070] Step 2: When the bus zero sequence voltage Greater than the zero-sequence voltage starting value Or the difference between any two adjacent zero-sequence voltage sampling values Greater than the zero-sequence voltage sudden change starting value , perform the following steps, where Indicates time The zero-sequence voltage sampling value, Indicates time Zero-sequence voltage sampling value:

[0071] Step 2.1: Construct an array based on the collected zero-sequence currents and calculate the similarity between the zero-sequence currents at any two outgoing lines to obtain a similarity matrix containing the characteristic information of the fault line and the characteristic information of the non-fault line.

[0072] The method for calculating the similarity between the zero-sequence currents at any two outgoing lines includes:

[0073] The cosine distance is used to measure the similarity between the zero-sequence currents at any two outgoing lines.

[0074] The cosine distance is calculated based on the cosine value of the angle between the zero-sequence current vectors of any two outgoing lines in the vector space. The calculation expression of the cosine value of the angle is:

[0075] (1);

[0076] Where, is the cosine value of the angle between the zero-sequence current vectors of any two outgoing lines, Indicates the outlet Zero sequence current vector With the output line Zero sequence current vector The cosine calculation formula between Indicates the total number of outgoing lines in the distribution network, Indicates the outlet The zero sequence current vector samples, Indicates the outlet The zero sequence current vector samples.

[0077] Among them, the cosine value of the angle between the zero-sequence current vectors at any two outgoing lines is The value range of is [-1, 1]. The cosine distance is calculated based on the cosine value of the angle between the zero-sequence current vectors at any two outgoing lines in the vector space. The closer the cosine value of the angle is to 1, the more similar the waveforms of the zero-sequence currents at any two outgoing lines are; the closer the cosine value of the angle is to -1, the more opposite the waveforms of the zero-sequence currents at any two outgoing lines are.

[0078] The mathematical expression of the similarity matrix containing the fault line characteristic information and the non-fault line characteristic information is:

[0079] (2);

[0080] Where r is the similarity matrix containing the characteristic information of the faulty line and the characteristic information of the non-faulty line, Indicates the cosine similarity of the zero-sequence current of line 1 itself, represents the cosine similarity between line 1 and line 2, represents the cosine similarity between line 1 and line n, = represents the cosine similarity between line 2 and line 1, Indicates the cosine similarity of the zero-sequence current of line 2 itself, represents the cosine similarity between line 2 and line n, = represents the cosine similarity between line n and line 1, = represents the cosine similarity between line n and line 2, represents the cosine similarity of the zero-sequence current of line n itself, where Indicates the total number of outgoing lines in the distribution network.

[0081] Since the cosine distance normalizes the lengths of two high-dimensional zero-sequence current vectors, the normalized calculation result only reflects the similarity in the vector direction and is independent of the length of the vector itself. Therefore, the similarity algorithm can still maintain its effectiveness when the fault current is weak due to a high-resistance grounding fault.

[0082] Step 2.2: Preset the range of the number of clusters according to the access situation of distributed power sources, and determine the optimal number of clusters by adopting the principle of minimizing the sum of squares within the cluster.

[0083] This embodiment sets the value range of the number of clusters based on the access status of distributed power sources in the distribution network system. Specifically:

[0084] In traditional distribution networks, since there are only two situations: busbar single-phase grounding and outgoing line single-phase grounding, the number of clusters may be set to 1 or 2 to accommodate these two different fault modes.

[0085] The situation becomes more complex when multiple outgoing lines in the distribution network are connected to distributed generation (DGs). In this case, the non-faulty lines are further subdivided into those with and without DGs. Therefore, the number of clusters may be set to 2 or 3 to more accurately reflect the distinction between faulty and non-faulty lines (including those with and without DGs).

[0086] This setting is intended to ensure that cluster analysis can accurately reflect the actual operating status of the distribution network, thereby improving the accuracy and reliability of fault identification.

[0087] In summary, the present invention sets up two scenarios: one with and one with distributed power supply access. Therefore, the value range of the number of clusters is preset according to the access situation of the distributed power supply, including: when there is no distributed power supply access in the distribution network outgoing line, the number of clusters k=1 or 2; when there is distributed power supply access in the distribution network outgoing line, the number of clusters k=2 or 3.

[0088] Step 2.3: Based on the similarity matrix and the improved k-means algorithm, perform cluster analysis according to the optimal number of clusters to obtain cluster analysis results.

[0089] Since the zero-sequence currents of fault lines and normal lines with or without distributed generation are significantly different, the cluster centers of each type of line should be as far apart as possible. Based on the above reasons, the original k-means algorithm is improved.

[0090] The improved method of the k-means algorithm includes:

[0091] When the number of clusters k is equal to 1, a vector is randomly generated as the cluster center;

[0092] When the number of clusters k is equal to 2 or 3, perform the following steps to select cluster centers:

[0093] Randomly select a row vector in the similarity matrix as the initial cluster center, and the remaining vectors are candidate row vectors;

[0094] Repeat the following steps until k-1 cluster centers are selected;

[0095] The shortest distances between the currently unselected candidate row vectors in the similarity matrix and the selected cluster centers are calculated in sequence.

[0096] The calculation of the candidate row vector that is not currently selected and the selected kth cluster center The mathematical expression of the shortest distance between is:

[0097] (3);

[0098] Where, Represents the candidate row vector that is not currently selected With the selected cluster center The shortest distance between.

[0099] Based on the weighted probability distribution of the shortest distance, a candidate row vector is randomly selected from the currently unselected candidate row vectors as a new cluster center; wherein the probability of the currently unselected candidate row vector being selected as the new cluster center is:

[0100] (4);

[0101] Where, Indicates the probability that the currently unselected candidate row vector is selected as the new cluster center, Represents any unselected candidate row vector The shortest distance between the kth cluster center and the selected cluster center.

[0102] Step 2.4: Determine the fault type and select the faulty line based on the cluster analysis results.

[0103] The method of determining the fault type according to the cluster analysis result includes:

[0104] When a distributed power source is connected to a low-resistance grounded distribution network system with multiple outgoing lines:

[0105] When the number of clusters k=2, a busbar single-phase grounding fault occurs;

[0106] When the number of clusters k = 3, the smallest cluster has a single-phase grounding fault;

[0107] In a low-resistance grounded distribution network system without distributed generation:

[0108] When the number of clusters k=1, a busbar single-phase grounding fault occurs;

[0109] When the number of clusters k=2, a single-phase grounding fault occurs in the smallest cluster.

[0110] In the actual application of the present invention, corresponding alarms or removal instructions can be given based on the fault judgment results.

[0111] Step 3: Simulation verification.

[0112] This embodiment uses MATLAB / Simulink to simulate a single-phase grounding fault in a typical 10kV low-resistance grounding distribution network.

[0113] In this example, five outgoing lines are cable lines with lengths of 6 km, 10 km, 12 km, 13 km, and 20 km, respectively. The neutral grounding resistance Rn is 10 Ω. The fault trigger time is set to 0.1 seconds, the initial fault phase angle is set to 0°, and the sampling frequency is set to 6.4 kHz. The fault type is a single-phase ground fault, and the fault point is located on phase A of outgoing line 1, 3 km from the busbar. The line parameters used in the simulation model are shown in Table 1.

[0114]

[0115] In this embodiment, the grounding transition resistance is set to 100Ω, 1000Ω, 2000Ω, and 2500Ω, respectively. The unfiltered half cycle after the fault occurs is used as the data acquisition window. The length of the half-cycle data acquisition window is 10ms. When the transition resistance is 100Ω, the similarity matrix between each outgoing line, the clustering labels under different transition resistance conditions, and the fault line selection results are shown in Tables 2 and 3.

[0116]

[0117]

[0118] It can be seen from Table 2 and Table 3 that when the grounding resistance in the low-resistance grounding system ranges from 100 to 2500Ω, the method provided in this embodiment can achieve reliable line selection of the fault line.

[0119] When a distributed power source is connected, the amplitude and phase relationship of the zero-sequence current on each outgoing line will change significantly. When outgoing lines 2 and 4 are connected to the distributed power source, and assuming that outgoing line 1 still has a fault and the transition resistance is 1000Ω, the method provided in this embodiment produces a cluster label of "1 2 3 2 3," indicating that the fault occurred on outgoing line 1. The outgoing lines are then divided into a faulty outgoing line, a non-faulty outgoing line connected to the distributed power source, and a non-faulty outgoing line not connected to the distributed power source.

[0120] Example 3

[0121] Based on the same inventive concept as other embodiments, this embodiment introduces a computer-readable storage medium on which computer instructions are stored, characterized in that when the computer instructions are executed by a processor, the steps of the method of the above-mentioned embodiment 1 or embodiment 2 are implemented.

[0122] In summary, the present invention demonstrates a high degree of flexibility and adaptability in fault identification and fault line selection. By dynamically adjusting the value range of the cluster center, the present invention can not only be effectively applied to fault diagnosis of traditional distribution networks, but can also adapt to the complex environment of active distribution networks. This feature enables the present invention to achieve accurate fault identification and line selection when facing different types of distribution networks without the need for additional configuration or adjustment. In addition, by constructing a similarity matrix containing characteristic information of fault lines and non-fault lines, and combining it with an optimized cluster analysis algorithm, the present invention can quickly and accurately lock the fault line from a large number of lines, thereby greatly improving the efficiency and accuracy of fault handling.

[0123] This invention significantly improves the effectiveness of fault identification. It uses the cosine distance to characterize the similarity of zero-sequence currents at any outgoing line in the distribution network. This method leverages the cosine distance's insensitivity to vector magnitude, effectively identifying faults even in high-resistance ground faults with small zero-sequence currents. This innovation not only broadens the scope of fault identification but also improves its accuracy.

[0124] At the same time, the improved k-means method employed in the present invention further optimizes the cluster center selection process, enabling full mining and utilization of fault feature information in the similarity matrix. Therefore, during data acquisition, the present invention can perform fault identification without waiting for the transient process of the zero-sequence electrical quantity to end, significantly shortening the time window for data acquisition. Furthermore, the method provided by the present invention eliminates the need for complex threshold settings, resolving the difficulty of setting setpoints in traditional high-resistance protection and further enhancing the convenience and practicality of fault identification.

[0125] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0126] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0127] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0128] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0129] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which are all protected by the present invention.

Claims

1. A high-resistance ground fault protection method based on a universal low-resistance grounding system, characterized in that: include: Collect the busbar zero-sequence voltage and the zero-sequence current at each outgoing line of the distribution network after a fault occurs; When the bus zero-sequence voltage is greater than the zero-sequence voltage starting value or the difference between any two adjacent zero-sequence voltage sampling values ​​is greater than the zero-sequence voltage mutation starting value, the following steps are performed: An array is constructed based on the collected zero-sequence currents, and the similarity between the zero-sequence currents at any two outgoing lines is calculated to obtain a similarity matrix containing the characteristic information of the fault line and the characteristic information of the non-fault line; The range of the number of clusters is preset according to the access situation of distributed power sources, and the optimal number of clusters is determined by minimizing the sum of squares within the cluster; Based on the similarity matrix and the improved k-means algorithm, performing cluster analysis according to the optimal number of clusters to obtain a cluster analysis result; Determine the fault type and select the faulty line according to the cluster analysis result; Determining the fault type according to the cluster analysis result includes: When a distributed power source is connected to a low-resistance grounded distribution network system with multiple outgoing lines: When the number of clusters k=2, a busbar single-phase grounding fault occurs; When the number of clusters k = 3, the smallest cluster has a single-phase grounding fault; In a low-resistance grounded distribution network system without distributed generation: When the number of clusters k=1, a busbar single-phase grounding fault occurs; When the number of clusters k = 2, the smallest cluster has a single-phase grounding fault; Methods for calculating the similarity between zero-sequence currents at any two outgoing lines include: The cosine distance is used to measure the similarity between the zero-sequence currents at any two outgoing lines; The improved method of the k-means algorithm includes: When the number of clusters k is equal to 1, a vector is randomly generated as the cluster center; When the number of clusters k is equal to 2 or 3, perform the following steps to select cluster centers: Randomly select a row vector in the similarity matrix as the initial cluster center, and the remaining vectors are candidate row vectors; Repeat the following steps until k-1 cluster centers are selected; sequentially calculating the shortest distances between unselected candidate row vectors in the similarity matrix and the selected cluster centers; Based on the weighted probability distribution of the shortest distance, a candidate vector is randomly selected from the currently unselected candidate row vectors as a new cluster center.

2. The high-resistance grounding fault protection method based on a universal low-resistance grounding system according to claim 1, characterized in that: The cosine distance is calculated based on the cosine value of the angle between the zero-sequence current vectors of any two outgoing lines in the vector space. The calculation expression of the cosine value of the angle is: （1）； Where, is the cosine value of the angle between the zero-sequence current vectors of any two outgoing lines, Indicates the outlet Zero sequence current vector With the output line Zero sequence current vector The cosine calculation formula between Indicates the total number of outgoing lines in the distribution network, Indicates the outlet The zero sequence current vector samples, Indicates the outlet The zero sequence current vector samples.

3. The high-resistance grounding fault protection method based on a universal low-resistance grounding system according to claim 2, characterized in that: The cosine distance is calculated based on the cosine value of the angle between the zero-sequence current vectors at any two outgoing lines in the vector space, wherein the closer the cosine value of the angle is to 1, the more similar the waveforms of the zero-sequence currents at any two outgoing lines are; and the closer the cosine value of the angle is to -1, the more opposite the waveforms of the zero-sequence currents at any two outgoing lines are.

4. The high-resistance grounding fault protection method based on a universal low-resistance grounding system according to claim 3, characterized in that: The mathematical expression of the similarity matrix containing the fault line characteristic information and the non-fault line characteristic information is: （2）； Where r is the similarity matrix containing the characteristic information of the faulty line and the characteristic information of the non-faulty line, Indicates the cosine similarity of the zero-sequence current of line 1 itself, represents the cosine similarity between line 1 and line 2, represents the cosine similarity between line 1 and line n, = represents the cosine similarity between line 2 and line 1, Indicates the cosine similarity of the zero-sequence current of line 2 itself, represents the cosine similarity between line 2 and line n, = represents the cosine similarity between line n and line 1, = represents the cosine similarity between line n and line 2, represents the cosine similarity of the zero-sequence current of line n itself, where Indicates the total number of outgoing lines in the distribution network.

5. The high-resistance grounding fault protection method based on a universal low-resistance grounding system according to claim 4, characterized in that: Calculate the candidate row vector that is not currently selected and the selected kth cluster center The mathematical expression of the shortest distance between is: （3）； Where, Represents the candidate row vector that is not currently selected With the selected cluster center The shortest distance between.

6. The high-resistance grounding fault protection method based on a universal low-resistance grounding system according to claim 5, characterized in that: Based on the weighted probability distribution of the shortest distance, a candidate vector is randomly selected from the currently unselected candidate row vectors as a new cluster center, wherein the probability of the currently unselected candidate row vector being selected as the new cluster center is: （4）； Where, Indicates the probability that the currently unselected candidate row vector is selected as the new cluster center, Represents any unselected candidate row vector The shortest distance between the kth cluster center and the selected cluster center.

Images