A power distribution network single-phase ground fault section intelligent threshold-free positioning method

Abstract

The application aims to provide a power distribution network single-phase grounding fault section intelligent threshold-free positioning method, when a single-phase grounding fault occurs in a power distribution network, zero sequence voltage derivative and zero sequence current waveform data of the first half wave after the fault are collected and extracted, transient zero sequence signals are normalized, a voltage-current characteristic scatter plot is drawn, an overall distribution of the scatter plot is represented to extract a voltage-current characteristic eigenvector, and a clustering center is calculated by using a K-means clustering algorithm; the transient zero sequence voltage derivative and the zero sequence current are multiplied and accumulated according to time to obtain a zero sequence power waveform, so as to extract a zero sequence power direction eigenvector; the two eigenvectors after the normalization are fused to construct a comprehensive eigenvector, and the fuzzy C-means clustering is performed on the comprehensive eigenvector, so as to distinguish the intact section and the fault section without threshold setting. The application does not need to calculate the threshold of the single-phase grounding fault section positioning criterion for different topological structures, and has strong universality.

Description

Technical Field

[0001] This invention belongs to the field of single-phase grounding fault location technology in power distribution networks, and specifically relates to an intelligent threshold-free location method for single-phase grounding fault sections in power distribution networks. Background Technology

[0002] With the development of power systems and the integration of a high proportion of distributed power sources, the complexity of distribution network structures is constantly increasing, making them more prone to faults. According to relevant statistics, the vast majority of power outages are related to distribution network faults, with single-phase grounding faults accounting for over 80%. Most existing distribution networks use non-effective grounding methods. The newly revised "Technical Guidelines for Distribution Networks" states that after a permanent single-phase grounding fault occurs on a medium-voltage line, it should be handled according to the principle of rapid, nearby fault isolation, i.e., rapid segment tripping. If the single-phase grounding fault section cannot be disconnected for an extended period, it can easily cause breakdown at weak points in the insulation of the non-faulty phases, forming a phase-to-phase short circuit and further amplifying the fault. Clearly, quickly and accurately locating and isolating single-phase grounding fault sections is of great significance for improving the operational reliability of distribution networks.

[0003] Although some progress has been made in the research on single-phase grounding fault location in distribution networks, the distribution network structure is complex, the operating environment is harsh, the communication burden is heavy, and the faults are highly random. Furthermore, traditional fault location methods require pre-setting a threshold for the fault location criterion, and different thresholds are set according to different power grid structures and operating states. However, such methods can introduce significant errors due to inaccurate threshold settings, resulting in poor performance of existing fault location methods in engineering applications. Therefore, threshold-free fault location methods have become one of the research focuses.

[0004] Existing methods for locating single-phase grounding fault sections suffer from the following main problems: First, existing methods have excessive communication overhead. When communication is delayed or fails, the location methods, which require transmitting large amounts of fault data, are significantly affected, easily leading to misjudgments. Second, influenced by prior knowledge, single-phase grounding fault location methods that only select a single feature quantity cannot fully characterize nonlinear fault features, cannot capture complete single-phase grounding fault information, and have many potential problems, resulting in low generalization ability. Furthermore, the setting of thresholds requires reference to network structure and operating parameters. However, the distribution network structure and parameters vary in different regions, and the operating mode of the same substation is not always constant. This makes the reliability and adaptability of manually set threshold single-phase grounding fault selection methods low. Finally, actual field data on single-phase grounding faults is difficult to obtain, and simulation models cannot simulate all fault conditions, resulting in poor universality of section location methods trained on large amounts of historical data. Summary of the Invention

[0005] Therefore, in order to fill the gaps and deficiencies in the existing technology, the present invention aims to provide an intelligent threshold-free location method for single-phase grounding fault sections in power distribution networks.

[0006] When a single-phase ground fault occurs in the distribution network, the zero-sequence voltage derivative and zero-sequence current waveform data of the first half-wave after the fault are actively collected and extracted. After normalization, a scatter plot of the volt-ampere characteristic is plotted. The cluster centers obtained by K-means clustering are used to characterize the overall distribution of the scatter plot, which serves as the volt-ampere characteristic feature vector. The transient zero-sequence voltage derivative and zero-sequence current are multiplied and accumulated over time to obtain the zero-sequence power accumulation waveform, from which the zero-sequence power direction feature vector is extracted. By fusing the normalized volt-ampere characteristic feature vector and the zero-sequence power direction feature vector to construct a comprehensive feature vector, fuzzy C-means clustering is performed on it, which can eliminate the need for threshold-based discrimination between healthy sections and faulty sections.

[0007] The intelligent threshold-free location method for single-phase grounding fault sections in distribution networks of this invention does not require calculating thresholds for single-phase grounding fault section location criteria for different network structures, nor does it require transmitting large amounts of fault data. It does not rely on a large amount of sample data or field fault data, requiring only a small amount of historical single-phase grounding fault data. It can accurately select single-phase grounding fault sections under various conditions such as line switching, high-resistance grounding, feeder end faults, and different compensation degrees. Furthermore, it avoids problems such as complex algorithms and large computational loads, requiring only a small amount of historical data, and has strong field universality and applicability. This provides a strong guarantee for the reliable and economical operation of distribution networks.

[0008] The specific technical solution adopted by this invention to solve its technical problem is as follows:

[0009] A smart threshold-free method for locating single-phase grounding fault sections in a distribution network eliminates the need for threshold settings. After a single-phase grounding fault occurs, the method collects and extracts the zero-sequence voltage derivative and zero-sequence current waveform data of the first half-wave following the fault, performs normalization processing, and plots a scatter plot of the volt-ampere characteristics. The cluster centers obtained through K-means clustering are used to characterize the overall distribution of the scatter plot, serving as the volt-ampere characteristic feature vector. The transient zero-sequence voltage derivative and zero-sequence current are multiplied and accumulated over time to obtain the zero-sequence power accumulation waveform, and the zero-sequence power direction feature vector is extracted. The normalized volt-ampere characteristic feature vector and the zero-sequence power direction feature vector are fused to construct a comprehensive feature vector, which is then subjected to fuzzy C-means clustering to distinguish between healthy and faulty sections without threshold settings.

[0010] Furthermore, it specifically includes the following steps:

[0011] Step S1: After a single-phase ground fault occurs, collect and extract the zero-sequence voltage derivative and zero-sequence current of the first half-wave after the fault, normalize the transient zero-sequence signal, and draw a scatter plot of the volt-ampere characteristics.

[0012] Step S2: Using the K-means clustering algorithm, find the cluster centers of the scatter plot representing the current-voltage characteristics. According to the first dimension data of the cluster centers in ascending order, take the second dimension data values ​​of the cluster centers in turn to obtain the feature vector representing the current-voltage characteristics.

[0013] Step S3: Accumulate the product of the transient zero-sequence voltage derivative and the zero-sequence current over time to obtain the zero-sequence power accumulation waveform. Select equally spaced points without a starting point from the waveform and take the ordinate values ​​corresponding to these equally spaced points to obtain the characteristic vector representing the direction of the zero-sequence power.

[0014] Step S4: After normalizing the current-voltage characteristic and the zero-sequence power direction feature vector respectively, the two are concatenated to obtain a comprehensive feature vector. Fuzzy C-means clustering is performed on the comprehensive feature vector to distinguish between healthy sections and faulty sections without setting a threshold.

[0015] Furthermore, regarding step S1: in the scatter plot of the current-voltage characteristic, for the healthy section, ignoring the line impedance, the derivative of its transient zero-sequence voltage is proportional to the zero-sequence current and has the same polarity, so the scatter plot distribution of the current-voltage characteristic is positively correlated; for the faulty section, ignoring the line impedance, the derivative of its transient zero-sequence voltage is inversely proportional to the zero-sequence current and has opposite polarities, so the scatter plot distribution of the current-voltage characteristic is negatively correlated.

[0016] Furthermore, regarding step S3:

[0017] The zero-sequence power of the healthy section, ignoring the line impedance, is only related to the total line capacitance, which is positive. The zero-sequence power is accumulated over time, and the accumulated value is positive.

[0018] The zero-sequence power of the fault section, ignoring the line impedance, is related to the inductance current of the arc suppression coil and the total line capacitance. The inductance current of the arc suppression coil is approximately zero in the first half-cycle after the fault occurs, the total line capacitance is positive, and the zero-sequence power is accumulated over time, with the accumulated value being negative.

[0019] Furthermore, the normalized volt-ampere characteristic feature vector and the zero-sequence power direction feature vector are concatenated, and the resulting feature quantities are used as the criteria for single-phase grounding fault segment selection, thereby improving the method's adaptability to various single-phase grounding fault conditions.

[0020] Furthermore, it does not require calculating the threshold for single-phase grounding fault location criteria for different topologies. It only requires a small amount of historical data on single-phase grounding faults and can accurately select single-phase grounding fault sections under conditions such as line switching, high-resistance grounding, feeder end faults, and different compensation degrees, making it highly universal.

[0021] Compared with the prior art, the present invention and its preferred embodiments have the following beneficial effects:

[0022] 1. The method utilizes two characteristic information, transient zero-sequence volt-ampere characteristics and transient zero-sequence power direction, as a criterion for locating single-phase ground fault sections. This enhances the fault tolerance of the single-phase ground fault section location method and improves its adaptability to various single-phase ground fault situations.

[0023] 2. The feature extraction method based on K-means clustering and the segment selection method based on fuzzy C-means clustering have the following advantages: First, since actual field fault data is difficult to obtain, this method relies on only a small amount of historical data. Most deep learning algorithms are supervised learning networks, which do not perform well with small data samples and have low engineering practicality. In comparison, this method is more advantageous in engineering terms. Second, compared with traditional segment location methods, the fuzzy C-means clustering algorithm uses the comprehensive feature vector as input and can distinguish between healthy segments and faulty segments without setting segment thresholds, thus having stronger universality. Third, even when the transition resistance changes randomly, it can still ensure the accuracy of locating single-phase grounding fault segments. Attached Figure Description

[0024] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0025] Figure 1 This is a schematic diagram illustrating the implementation process of the single-phase grounding fault location algorithm in an embodiment of the present invention.

[0026] Figure 2 This is a schematic diagram of the zero-sequence current distribution in an equivalent network according to an embodiment of the present invention;

[0027] Figure 3 This is a waveform diagram of the zero-sequence voltage derivative under a typical fault, according to an embodiment of the present invention.

[0028] Figure 4 This is a zero-sequence current waveform diagram under a typical fault according to an embodiment of the present invention;

[0029] Figure 5 This is a scatter plot of zero-sequence voltage derivative and zero-sequence current under typical fault conditions according to an embodiment of the present invention.

[0030] Figure 6 This is a waveform diagram of zero-sequence power accumulation under typical fault conditions according to an embodiment of the present invention;

[0031] Figure 7 This is a simulation model diagram of a 10kV resonant grounding system according to an embodiment of the present invention. Detailed Implementation

[0032] To make the features and advantages of this patent more apparent and understandable, specific embodiments are provided below for detailed explanation:

[0033] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0034] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0035] like Figure 7 As shown, this embodiment provides the topology of a 10kV distribution network single-phase ground fault system and verifies the effectiveness of the intelligent threshold-free location method for single-phase ground fault sections in the distribution network.

[0036] The algorithm flow for locating single-phase ground fault sections is as follows: Figure 1 As shown, after a single-phase ground fault occurs, the transient zero-sequence voltage derivative and zero-sequence current waveform data of the first half-wave after the fault are extracted. After normalizing the transient zero-sequence signal, a scatter plot of the volt-ampere characteristic is plotted. The cluster centers obtained by K-means clustering represent the distribution of the scatter plot and serve as the volt-ampere characteristic feature vector. At the same time, the zero-sequence voltage derivative and zero-sequence current are multiplied and accumulated over time to obtain the zero-sequence power accumulation waveform. The zero-sequence power direction feature vector is extracted, and the two are fused to obtain a comprehensive feature vector. This vector is then input into the fuzzy C-means clustering algorithm. Combined with a small amount of historical data, threshold-free fault segment location is achieved.

[0037] The technical solution of this embodiment will be described in detail from the perspective of principle as follows:

[0038] 1. Introduction to the principle of threshold-free section location for single-phase ground faults

[0039] 1.1 Transient I-V characteristic analysis of single-phase ground fault in distribution network

[0040] A schematic diagram of a single-phase grounding fault in a resonant grounding system is shown below. Figure 2 As shown, The electromotive forces of phases A, B, and C are respectively, and C is... 0i C represents the capacitance to ground of segment i. 01 R represents the capacitance to ground of the faulty section. L L represents the equivalent resistance and inductance of the arc suppression coil, and R... f Let be the fault grounding resistance. Ignoring line impedance, for a healthy section, the relationship between the derivative of the transient zero-sequence voltage and the transient zero-sequence current is:

[0041]

[0042] For the fault section, the relationship between the derivative of the transient zero-sequence voltage and the transient zero-sequence current is as follows:

[0043]

[0044] Where u 0j (t) represents the transient zero-sequence voltage at sampling point j, i 0j Let C be the transient zero-sequence current at sampling point j. 1j It is the equivalent zero-sequence capacitance to ground from sampling point j downstream of the healthy or faulty section to the end of the feeder, C. 2j It is the sum of the equivalent capacitance to ground from sampling point j upstream of the faulty section to the busbar and the zero-sequence capacitance to ground of all healthy lines, i L (t) represents the inductance current of the arc suppression coil.

[0045] In a resonant grounding system, after a single-phase ground fault occurs, the transient zero-sequence current of each non-faulty line flows from the busbar to the line, while the transient zero-sequence current of the faulty line flows from the line to the busbar. In other words, neglecting line impedance, the derivative of the zero-sequence voltage in the healthy section is proportional to the zero-sequence current and has the same polarity; in the faulty section, since the inductance current of the arc suppression coil remains essentially constant during the first half-wave after the fault, it can be approximated as zero, and the derivative of the zero-sequence voltage is inversely proportional to the zero-sequence current and has opposite polarities.

[0046] 1.2 Principle of Irradiance Characteristic Vector Extraction

[0047] like Figure 7 The image shows a simulation model of a 10kV resonant grounding system built using PSCAD / EMTDC software. In the figure, O... L For overhead lines; C L For cable lines; R f For grounding resistance; use F i This indicates the downstream segment immediately adjacent to sampling point i; i is the sampling node number, and the line parameters are shown in Table 1.

[0048] Table 1 Line Parameters

[0049]

[0050] Taking a single-phase ground fault of phase A in section F3 as an example, the fault phase angle is set to 90°, the grounding resistance to 10Ω, and the sampling frequency to 10kHz. Figure 7 Simulation data from sampling points 1, 2, 3, 4, 6, and 8, including the zero-sequence voltage derivative and zero-sequence current at each sampling point for the quarter-cycle before and the first half-cycle after the fault. Figure 3 , Figure 4 As shown.

[0051] After normalizing the zero-sequence signal, a scatter plot of the zero-sequence voltage derivative and zero-sequence current voltage-current characteristics is plotted. For healthy sections, the distribution of each scatter point will have a positive correlation; for faulty sections, the distribution of each scatter point will have a negative correlation.

[0052]

[0053] In the formula, X represents the sampled data of the zero-sequence voltage derivative or the zero-sequence current waveform. i Let X be the elements of vector X, i = 1, 2, ..., n, where n is the number of sampling points in X. `abs()` calculates the absolute value, and `max()` calculates the maximum value. For x i The normalized elements restrict the elements in vector X to [-1, 1]. It is worth noting that as the ground fault duration increases, the arc suppression coil transitions to an overcompensated state, and the zero-sequence voltage derivative and zero-sequence current in the fault section no longer have an inverse relationship. Therefore, it is necessary to select the first half-cycle after the fault for the transient zero-sequence signal as the characteristic criterion.

[0054] The distribution of the volt-ampere characteristic scatter plot needs to be characterized by cluster centers obtained using the K-means clustering algorithm. The volt-ampere characteristic feature vectors extracted from these cluster centers are used to distinguish between healthy and faulty sections. The basic idea of ​​the K-means clustering algorithm is to randomly select objects for partitioning, randomly set initial cluster centers, and iteratively adjust the cluster centers to minimize the sum of the distances of all objects to the cluster centers. Input m sample data X = {x1, x2, ..., x...} m Each sample data point consists of an n-dimensional vector, i.e., x. i ={x i1 ,x i2 ,...,x in Let i = 1, 2, ..., m. Assume there are M clusters. Randomly generate M cluster centers, where 1 < M < m. Calculate the Euclidean distance from each sample data point to each cluster, and assign each sample data point to the cluster with the smallest distance. The formula for the Euclidean distance is:

[0055]

[0056] In the formula, x it x represents i The t-th dimension of the data, c j Let c be the j-th cluster center, j = 1, 2, ..., M. jt Represented as c j Given the t-th dimension of the data, divide all sample data into M clusters and recalculate the cluster centers of these M clusters. The calculation method is the average of all sample data in each cluster, and the formula is:

[0057]

[0058] In the formula, a j S represents the number of samples in the j-th cluster. i S represents the sample data in the j-th cluster. it For S i The t-th dimension of the data. In this embodiment, n=2 is set to obtain 2-dimensional cluster centers. After 10 iterations, if the cluster centers remain unchanged, the clustering ends and the cluster centers are output. Then, the second dimension data values ​​g of the cluster centers are taken in ascending order of the first dimension data. 1i (i = 1, 2, ..., M), the eigenvectors of the current-voltage characteristic can be obtained:

[0059] G = [g 11 g 12 … g 1M (6)

[0060] The scatter plot of the calculated cluster center characterization of the volt-ampere properties is shown below. Figure 5 As shown in Table 2, after the normalization process described above, the volt-ampere characteristic feature vector extraction results for the downstream section of the sampling point are shown in Table 2.

[0061] Table 2 Irregular vectors of volt-ampere characteristics

[0062]

[0063] 1.3 Transient Zero-Sequence Power Direction Analysis in Single-Phase Ground Faults of Distribution Networks

[0064] The zero-sequence power is obtained by multiplying the zero-sequence voltage derivative by the zero-sequence current, for the healthy region as follows:

[0065]

[0066] For the faulty section:

[0067]

[0068] Ignoring line impedance, the sign of the zero-sequence power value depends only on the total line capacitance. Since the total line capacitance is not negative, the zero-sequence power value is positive in the healthy section. The zero-sequence power value is accumulated over time, and the accumulated value is positive. In the faulty section, the sign of the zero-sequence power value depends on the inductance current of the arc suppression coil and the total line capacitance. The inductance current of the arc suppression coil remains essentially unchanged during the first half-wave after the fault occurs, and can be approximated as zero. Since the total line capacitance is not negative, the zero-sequence power value is negative. The zero-sequence power value is accumulated over time, and the accumulated value is negative.

[0069] Therefore, the zero-sequence power waveform obtained by multiplying the zero-sequence voltage derivative and zero-sequence current of the first half-cycle after the fault is selected as the input quantity, and the cumulative sum waveform of the zero-sequence power value is obtained as a characteristic quantity to distinguish between the healthy section and the faulty section.

[0070] 1.4 Principle of Zero-Sequence Power Directional Feature Vector Extraction

[0071] Since the line impedance of the simulation model and the field system cannot be ignored, abnormal fluctuations in the zero-sequence power value increase. Selecting multiple points can preserve the waveform trend of the zero-sequence power accumulation sum, reducing the occurrence of misjudgments. For this embodiment, the zero-sequence power accumulation sum waveform is as follows: Figure 6 As shown, the cumulative zero-sequence power in the healthy section is positive, while the cumulative zero-sequence power in the faulty section is negative. Therefore, the cumulative zero-sequence power and waveform can be used to extract features to distinguish between healthy and faulty sections. Multiplying the zero-sequence voltage derivative data and zero-sequence current data at each sampling point yields the zero-sequence power value at each sampling point. Accumulating these values ​​over time yields the cumulative zero-sequence power value h at each sampling point. 1j (j = 1, 2, ..., n), from h 1j Select N equally spaced points (excluding the starting point) as the eigenvectors of the zero-sequence power, where N < n. The eigenvectors of the zero-sequence power direction can then be obtained:

[0072] H = [h] 11 h 12 … h 1N (9)

[0073] After normalization, the zero-sequence power direction feature vector extraction results for the downstream section of the sampling point are shown in Table 3.

[0074] Table 3 Zero-sequence power eigenvectors

[0075]

[0076] 2. Principle of Single-Phase Ground Fault Location Based on Clustering of Multiple Transient Feature Quantities

[0077] By using multiple feature information as criteria for locating single-phase ground fault sections, the method's adaptability to various single-phase ground fault situations is improved.

[0078] 2.1 Fusion of transient characteristic quantities of ground faults

[0079] The combined feature vector is obtained by concatenating the current-voltage characteristic feature vector and the zero-sequence power direction feature vector.

[0080]

[0081] 2.2 Clustering and localization of multiple transient features in single-phase ground fault sections

[0082] Fuzzy C-means clustering is a common unsupervised dynamic clustering algorithm. Its core idea is to assign multidimensional input data to cluster centers based on membership degrees, maximizing similarity between objects in the same cluster and minimizing similarity between objects in different clusters, thus achieving flexible fuzzy data partitioning. Let m samples form a dataset S = {S1, S2, ..., S...} m Each sample data point consists of an n-dimensional vector, i.e., S. i ={S i1 ,S i2 ,…,S im Let S be a collection of clusters, i = 1, 2, ..., m. Divide S into c classes, where 2 ≤ c ≤ m. Each sample is assigned to a different class within c based on its membership degree. The clustering iteration ends when the objective function is minimized.

[0083]

[0084] In the formula J fcm Let U be the objective function, and U be the membership matrix. ij};u ij The membership degree of the j-th sample in S belonging to the i-th class is represented by p; p > 1 is the fuzzy weighting index; ||s j -c i || represents sample s j To cluster center c i Euclidean distance; c i (i = 1, 2, ..., n) represents the cluster centers of the i-th class. The fuzzy C-means clustering algorithm iteratively obtains the optimal cluster centers c and the membership matrix U. Where u ij The following conditions must be met:

[0085]

[0086] By inputting the comprehensive feature vector of the segment data and the comprehensive feature vector of the historical data into the fuzzy C-means clustering algorithm, healthy segments and faulty segments can be distinguished without setting a threshold.

[0087] 2.2.1 Section location effect under typical fault conditions

[0088] In this embodiment, the preset number of fuzzy C-means clusters is 2, and the weighting index m = 2. According to the simulated fault conditions set in Table 4, 60 sets of healthy segment sample data and faulty segment sample data are selected to form 120 sets of historical sample data. The comprehensive feature vector of each segment and the comprehensive feature vector of the historical samples are used as inputs to the fuzzy C-means clustering algorithm to obtain the membership matrix. Underlined labels in the membership matrix indicate the segment sample classification. Healthy segments and faulty segments are divided according to the membership degree, and the results are shown in Table 5.

[0089] Table 4 Historical Simulation Sample Data Parameters

[0090] Fault Section <![CDATA[R f (Oh)]]> θ(°) b / ％ quantity F11-F18 0,100,500,1000 0,30,60,90 5,8,10 120

[0091] Table 5 Simulation section location results

[0092] Section number Membership matrix Judgment results Section number Membership matrix Judgment results F1 <![CDATA[[ 0.9165 0.0835]]]> Fault Section F4 <![CDATA[[0.0125 0.9875 ]]]> Sound section F2 <![CDATA[[ 0.8142 0.1858]]]> Fault Section F6 <![CDATA[[0.0351 0.9649 ]]]> Sound section F3 <![CDATA[[ 0.8800 0.1200]]]> Fault Section F8 <![CDATA[[0.0484 0.9516 ]]]> Sound section

[0093] like Figure 1 As shown in this embodiment, a threshold-free method for locating single-phase ground fault sections in a distribution network is provided. After a single-phase ground fault occurs, the zero-sequence voltage derivative and zero-sequence current waveform data of the first half-wave after the fault are collected and extracted. After normalization, a scatter plot of the volt-ampere characteristics is plotted. The cluster centers obtained by K-means clustering represent the distribution of the scatter plot and serve as the volt-ampere characteristic feature vector. The zero-sequence power waveform data is obtained by multiplying the zero-sequence voltage derivative and the zero-sequence current. The zero-sequence power direction feature vector is obtained by accumulating the data at time points and extracting it. The two are then concatenated to obtain a comprehensive feature vector, which is input into a fuzzy C-means clustering algorithm. Combined with a small amount of historical data, threshold-free location of single-phase ground fault sections is achieved. The intelligent threshold-free location method for single-phase grounding fault sections in distribution networks proposed in this invention uses a fuzzy C-means clustering algorithm to distinguish between intact and faulty sections without setting a threshold for single-phase grounding fault selection criteria. It has high applicability and utilizes multiple feature information as criteria for single-phase grounding fault selection, which enhances the fault tolerance of the selection method and improves its adaptability to various single-phase grounding fault situations. It can accurately locate sections under conditions such as line switching, high-resistance grounding, feeder end faults, and different compensation degrees, providing strong support for the economical and reliable operation of distribution networks and has strong universality.

[0094] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0095] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations 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, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0096] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0097] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0098] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

[0099] This patent is not limited to the above-described preferred embodiment. Anyone can derive other forms of intelligent threshold-free location method for single-phase ground fault sections in power distribution networks based on the inspiration of this patent. All equivalent changes and modifications made within the scope of this patent application shall fall within the scope of this patent.

Claims

1. A method for intelligent threshold-free location of single-phase ground fault sections in a distribution network, characterized in that, Single-phase grounding fault location can be achieved without setting a threshold. After a single-phase grounding fault occurs, the zero-sequence voltage derivative and zero-sequence current waveform data of the first half-wave after the fault are collected and extracted, normalized, and a scatter plot of the volt-ampere characteristics is drawn. The cluster centers obtained by K-means clustering are used to characterize the overall distribution of the scatter plot, which serves as the volt-ampere characteristic feature vector. The transient zero-sequence voltage derivative and zero-sequence current are multiplied and accumulated over time to obtain the zero-sequence power accumulation waveform, and the zero-sequence power direction feature vector is extracted. The normalized volt-ampere characteristic feature vector and the zero-sequence power direction feature vector are fused to construct a comprehensive feature vector, which is then subjected to fuzzy C-means clustering to distinguish between healthy sections and faulty sections without setting a threshold.

2. The intelligent threshold-free location method for single-phase grounding fault sections in a distribution network according to claim 1, characterized in that, Specifically, the following steps are included: Step S1: After a single-phase ground fault occurs, collect and extract the zero-sequence voltage derivative and zero-sequence current of the first half-wave after the fault, normalize the transient zero-sequence signal, and draw a scatter plot of the volt-ampere characteristics. Step S2: Using the K-means clustering algorithm, find the cluster centers of the scatter plot representing the current-voltage characteristics. According to the first dimension data of the cluster centers in ascending order, take the second dimension data values ​​of the cluster centers in turn to obtain the feature vector representing the current-voltage characteristics. Step S3: Accumulate the product of the transient zero-sequence voltage derivative and the zero-sequence current over time to obtain the zero-sequence power accumulation waveform. Select equally spaced points without a starting point from the waveform and take the ordinate values ​​corresponding to these equally spaced points to obtain the characteristic vector representing the direction of the zero-sequence power. Step S4: After normalizing the current-voltage characteristic and the zero-sequence power direction feature vector respectively, the two are concatenated to obtain a comprehensive feature vector. Fuzzy C-means clustering is performed on the comprehensive feature vector to distinguish between healthy sections and faulty sections without setting a threshold.

3. The intelligent threshold-free location method for single-phase ground fault sections in a distribution network according to claim 2, characterized in that: For step S1: In the scatter plot of the current-voltage characteristic, for the healthy section, ignoring the line impedance, the derivative of its transient zero-sequence voltage is proportional to the zero-sequence current and has the same polarity, and the distribution of the scatter plot of the current-voltage characteristic is positively correlated. For the fault section, ignoring the line impedance, the derivative of the transient zero-sequence voltage is inversely proportional to the zero-sequence current and has opposite polarities, and the scatter point distribution of the volt-ampere characteristic is negatively correlated.

4. The intelligent threshold-free location method for single-phase ground fault sections in a distribution network according to claim 3, characterized in that: For step S3: The zero-sequence power of the healthy section, ignoring the line impedance, is only related to the total line capacitance, which is positive. The zero-sequence power is accumulated over time, and the accumulated value is positive. The zero-sequence power of the fault section, ignoring the line impedance, is related to the inductance current of the arc suppression coil and the total line capacitance. The inductance current of the arc suppression coil is approximately zero in the first half-cycle after the fault occurs, the total line capacitance is positive, and the zero-sequence power is accumulated over time, with the accumulated value being negative.

Images