Power line fault point positioning method and device applied to power engineering construction

Abstract

The invention relates to the field of electrical variable fault detection, in particular to an electric power line fault point positioning method and device applied to electric power engineering construction, and the method comprises the steps: recording a node with the maximum traveling wave signal intensity as a reference node, and recording nodes except the reference node as to-be-corrected nodes; correcting the traveling wave signal of the to-be-corrected node by using the traveling wave signal of each reference node to obtain an initial corrected traveling wave signal of the to-be-corrected node; extracting a stable frequency combination according to the frequency distribution difference of the initial correction traveling wave signals of all reference nodes to the same node to be corrected; and recombining all the stable frequency combinations according to the overall distribution difference of the stable frequency combinations of all the to-be-corrected nodes to obtain a new stable frequency combination, re-extracting the traveling wave signals of the nodes by using the new stable frequency combination, and carrying out fault point positioning. According to the invention, the accuracy of the fault point positioning result is improved.

Description

Technical Field

[0001] This invention relates to the field of electrical variable fault detection, and more specifically to a method and apparatus for locating power line fault points in power engineering construction. Background Technology

[0002] Fault location in power lines is a crucial preliminary step in power engineering construction. A common method for fault location is the two-end traveling wave method. This method involves installing traveling wave acquisition devices (usually voltage transformers) at both ends of the line to capture the traveling wave generated by the fault. The time difference between the received traveling waves at both ends is used to locate the fault point. However, distribution network topologies are often complex, with numerous power line branches. The two-end traveling wave method yields significant location errors; for example, if the fault point is not located between the two ends of the line, the location result will be incorrect. Furthermore, as the traveling wave signal propagates along multiple power lines, it undergoes multiple reflections and long-distance attenuation. Simultaneously, different noise interferences (such as grid noise caused by the connection of power equipment like frequency converters and inverters) are superimposed at different locations on the power lines. These noises can submerge and interfere with the traveling wave signal. Ultimately, this not only results in multiple power line points receiving the traveling wave signal in the distribution network, making the two-end traveling wave method unsuitable for multi-end traveling wave detection scenarios, but also means that the traveling wave signals detected at different power line points may have significant interference, further reducing the accuracy of fault location. Summary of the Invention

[0003] To address the aforementioned problems, this invention provides a method and apparatus for locating power line fault points in power engineering construction.

[0004] The power line fault location method and device of the present invention, applied to power engineering construction, adopts the following technical solution: One embodiment of the present invention provides a method for locating power line fault points applied in power engineering construction, the method comprising the following steps: Traveling wave signals are detected at multiple nodes of the power line. The nodes with the strongest traveling wave signal intensity are designated as reference nodes, and the nodes other than the reference nodes are designated as nodes to be corrected. The traveling wave signal of any node to be corrected is corrected using the traveling wave signal of each reference node to obtain the initial corrected traveling wave signal of any node to be corrected. Based on the frequency distribution differences of the initial correction traveling wave signals of all reference nodes for the same node to be corrected, a stable frequency combination is extracted from the initial correction traveling wave signal of the node to be corrected. The stable frequency combination contains several frequency signals. For each stable frequency combination of a node to be corrected, the difference between the traveling wave signals of all reference nodes and the initial corrected traveling wave signals of all nodes to be corrected under the stable frequency combination is obtained, which is simply referred to as the overall distribution difference of the stable frequency combination. For all stable frequency combinations of nodes to be corrected, all stable frequency combinations are reorganized according to the overall distribution difference of all stable frequency combinations to obtain a new stable frequency combination, such that the overall distribution difference of the new stable frequency combination is minimized. The traveling wave signal of the node is re-extracted using the new stable frequency combination, and the fault point is located.

[0005] Preferably, the specific steps of correcting the traveling wave signal of any node to be corrected using the traveling wave signal of each reference node to obtain the initial corrected traveling wave signal of any node to be corrected are as follows: For any voltage sequence acquired at a node to be corrected, initialize a filter and use the filter to filter the voltage sequence to obtain the first filtered sequence. Then, linearly normalize the first filtered sequence and the traveling wave signal of the reference node. The DTW distance between the normalized first filtered sequence and the normalized traveling wave signal of the reference node is denoted as the fluctuation trend difference. The parameters of the filter are optimized using the simulated annealing algorithm to minimize the difference in fluctuation trends. The first filtered sequence with the smallest difference in fluctuation trends is denoted as the initial corrected traveling wave signal of any reference node to any node to be corrected.

[0006] Preferably, the specific steps for extracting stable frequency combinations from the initial corrected traveling wave signal of the node to be corrected based on the frequency distribution differences of the initial corrected traveling wave signal of all reference nodes for the same node to be corrected include the following: For any node to be corrected, and for each reference node, the initial corrected traveling wave signal of the node to be corrected is obtained by using the Fourier transform algorithm. The spectral distribution of the initial corrected traveling wave signal includes the amplitude of each frequency in the initial corrected traveling wave signal. Any frequency in the frequency distribution corresponding to all reference nodes is denoted as the target frequency. The response and stability of the target frequency are obtained based on the amplitude of the target frequency in the frequency distribution corresponding to all reference nodes. The product of the response and stability of the target frequency is denoted as the stability index of the target frequency. Among all frequencies in the spectral distribution corresponding to all reference nodes, the N1 frequencies with the largest stability index are selected as the stable frequency combination. N1 is the stable frequency extraction parameter of the node to be corrected; the initial value of the stable frequency extraction parameter is a preset value.

[0007] Preferably, for each stable frequency combination of the node to be corrected, the difference between the traveling wave signals of all reference nodes and the initial corrected traveling wave signals of all nodes to be corrected under the stable frequency combination, referred to as the overall distribution difference of the stable frequency combination, includes the following specific steps: For any stable frequency combination among all the stable frequency combinations of the correction nodes; use this stable frequency combination to re-extract the traveling wave signal at any node, including: The traveling wave signal of any reference node or the initial corrected traveling wave signal of any node to be corrected is denoted as the basic signal of any node. The spectral distribution F1 of the basic signal is obtained by using the Fourier transform algorithm. The amplitude of all frequencies other than those included in the stable frequency combination in F1 is set to 0, which is denoted as the filtered spectral distribution of any node. The filtered spectral distribution is inversely transformed by the inverse Fourier transform algorithm to obtain the final signal of any node. The final signal represents the re-extracted traveling wave signal. The final signals of all nodes are linearly normalized, and the DTW distance between any two normalized final signals of any two nodes is denoted as the trend difference between any two nodes; the mean of the trend differences between all nodes is denoted as the overall distribution difference of the stable frequency combination.

[0008] Preferably, for all stable frequency combinations of nodes to be corrected, the stable frequency combinations are reorganized according to the overall distribution differences of all stable frequency combinations to obtain a new stable frequency combination, such that the overall distribution differences of the new stable frequency combination are minimized. The specific steps include the following: D1: The set S1 represents the combination of stable frequencies of all nodes to be corrected; D2: For all stable frequency combinations in set S1, obtain the stable frequency combination Q1 with the smallest overall distribution difference; in set S1, stable frequency combinations other than stable frequency combination Q1 are denoted as reference combinations. Based on the overall distribution difference of the reference combinations and the overlap between the frequencies contained in each reference combination and stable frequency combination Q1, obtain the merging index of each reference combination; merge the reference combination with the largest merging index into stable frequency combination Q1, and denote the resulting stable frequency combination Q1 as the merged combination, and obtain the overall distribution difference of the merged combination; D3: Modify the stable frequency extraction parameters of each node to be corrected by utilizing the overall distribution difference of the stable frequency combination of each node to be corrected. The modified stable frequency extraction parameters are positively correlated with the overall distribution difference. Using the modified stable frequency extraction parameters of each node to be corrected, the stable frequency combination of each node to be corrected is obtained again, and the overall distribution difference of each stable frequency combination is obtained again; the stable frequency combination of all nodes to be corrected and the merged combination are re-represented as set S1; D4: Repeat D2 and D3 several times; after the repeated execution is completed, obtain the stable frequency combination or merged combination with the smallest overall distribution difference from all the obtained sets S1, and use it as the new stable frequency combination.

[0009] Preferably, the specific steps for re-extracting the traveling wave signal of the node using the new stable frequency combination and locating the fault point are as follows: The traveling wave signal at any node is re-extracted using the new stable frequency combination, denoted as the traveling wave filtered signal, and the arrival time of the traveling wave filtered signal is obtained. Several candidate points are marked in the distribution network. The time of the traveling wave signal generated by any candidate point to each node is obtained and recorded as the theoretical time of any candidate point to each node. The theoretical time of any candidate point to all nodes constitutes the theoretical time series, and the arrival time of all nodes constitutes the arrival time series. The Pierce correlation coefficient of the theoretical time series and the arrival time series is obtained and recorded as the candidate index of any candidate point. The candidate point with the largest candidate index is selected as the fault location point.

[0010] Preferably, the specific steps for obtaining the response and stability of the target frequency based on the amplitude of the target frequency in the frequency distribution corresponding to all reference nodes are as follows: For the frequency distribution corresponding to all reference nodes, obtain the mean value of the target frequency amplitude in all frequency distributions, and use it as the response degree of the target frequency; The amplitude of the target frequency in all frequency distributions is linearly normalized, and the standard deviation of all linearly normalized amplitudes at the target frequency is denoted as x. exp(-x) is used as the stability of the target frequency; where exp() represents an exponential function with the natural constant as the base.

[0011] Preferably, the specific steps for obtaining the merging index of each reference combination based on the overall distribution differences of the reference combinations and the degree of overlap between the frequencies contained in each reference combination and the stable frequency combination Q1 are as follows: The overall distributional differences of all reference combinations are normalized and used as the attention coefficient for each reference combination. The first, second, and third intervals of the frequencies contained in each reference combination are denoted as distribution vector G1; the first, second, and third intervals of the frequencies contained in the stable frequency combination Q1 are denoted as distribution vector G2; the cosine similarity between distribution vector G1 and distribution vector G2 is denoted as the overlap degree; the product of the overlap degree and the attention coefficient of each reference combination is denoted as the merging index of each reference combination.

[0012] Preferably, the specific steps for modifying the stable frequency extraction parameters of each node to be corrected by utilizing the overall distribution differences of the stable frequency combinations of each node to be corrected are as follows: The overall distribution difference of the stable frequency combinations of all nodes to be corrected is normalized and used as the update scale k for each node to be corrected; the modified stable frequency extraction parameters for each node to be corrected are denoted as N2. Where N1 represents the initial value of the stable frequency extraction parameters. This indicates the parameters extracted from the stable frequency before modification. This indicates rounding down to the nearest integer.

[0013] Another embodiment of the present invention provides a power line fault location device for power engineering construction. The device includes traveling wave detectors installed at all nodes. The device also includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor reads the voltage signal collected by the traveling wave detector and runs the computer program, it executes all the steps of the power line fault location method described above for power engineering construction.

[0014] The beneficial effects of the technical solution of the present invention are: This invention uses the traveling wave signal of each reference node to correct the traveling wave signal of any node to be corrected, thus obtaining the initial corrected traveling wave signal of any node to be corrected. During this process, the traveling wave signal of the reference node has a high strength and is subject to relatively less interference. The traveling wave signal of the node to be corrected (i.e., the initial corrected traveling wave signal) is re-extracted based on the traveling wave signal of the reference node, ensuring that the fluctuation trend of the initial corrected traveling wave signal can match the traveling wave signal received by the reference node from the same fault point, thereby initially obtaining a more accurate traveling wave signal.

[0015] Furthermore, based on the frequency distribution differences of the initial corrected traveling wave signals of all reference nodes for the same node to be corrected, a stable frequency combination is extracted from the initial corrected traveling wave signal of the node to be corrected. The stable frequency combination contains several frequency signals. In this process, the several frequency signals contained in the stable frequency combination are the frequency distributions of the traveling wave signal that are relatively less affected by noise interference when it is transmitted to each node to be corrected. These frequency distributions help to further identify the traveling wave signal of the node to be corrected.

[0016] Furthermore, for all stable frequency combinations of the nodes to be corrected, the stable frequency combinations are recombined based on the overall distribution differences of all stable frequency combinations to obtain a new stable frequency combination, which minimizes the overall distribution differences of the new stable frequency combination. In this process, the new stable frequency combination not only helps to further identify the traveling wave signal of the node to be corrected, but also minimizes the problem of time delay error in the arrival of the traveling wave when directly using the stable frequency combinations of all nodes to be corrected for fault location.

[0017] Finally, the traveling wave signal of the node was re-extracted using a new stable frequency combination to locate the fault point, which improved the accuracy of the location results and avoided the influence of noise interference caused by the wave signal being reflected multiple times and attenuated over long distances, and by different noises being superimposed at different locations on the power line. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the steps of a power line fault location method applied to power engineering construction, as provided in an embodiment of the present invention. Detailed Implementation

[0020] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the power line fault location method and apparatus applied to power engineering construction according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0022] The following description, in conjunction with the accompanying drawings, details the specific scheme of the power line fault location method and device provided by the present invention, which is applied to power engineering construction.

[0023] Example 1: Please see Figure 1The diagram illustrates a flowchart of a method for locating power line faults in power engineering construction, provided by an embodiment of the present invention. The method includes the following steps: Step S101: Detect traveling wave signals from multiple nodes of the power line. The nodes with the strongest traveling wave signal are recorded as reference nodes, and the nodes other than the reference nodes are recorded as nodes to be corrected.

[0024] Multiple traveling wave detectors are installed on the power lines of the distribution network, with each detector's installation location considered a node. In this embodiment, the traveling wave detector is a voltage transformer used to acquire the voltage sequence at each node in real time, filter the voltage sequence, and extract the traveling wave signal at each node. The principle of the traveling wave signal generation is as follows: when a fault occurs at a certain point on the power line, the fault point will generate an electromagnetic transient pulse signal due to the sudden change in voltage, which is denoted as the traveling wave signal. In this example, the traveling wave signal is represented as a voltage change sequence over time, and the average of the absolute values ​​of all voltages in the traveling wave signal is denoted as the traveling wave signal strength.

[0025] For all nodes, the traveling wave signals and their strengths are analyzed. The N0 nodes with the strongest traveling wave signal strength are identified and designated as reference nodes. Nodes other than the reference nodes are designated as nodes to be corrected. Compared to the reference nodes, the traveling wave signal strength of the nodes to be corrected is weaker. In this embodiment, N0 is equal to half the number of all nodes (rounded down).

[0026] Step S102: Use the traveling wave signal of each reference node to correct the traveling wave signal of any node to be corrected, so as to obtain the initial corrected traveling wave signal of any node to be corrected.

[0027] Because the power distribution network has many branches and complex lines, when a fault occurs at a certain point on the power line, the traveling wave signal will undergo multiple reflections and long-distance attenuation as it propagates along the power line. At the same time, different noise interferences (such as grid noise caused by the connection of power equipment such as frequency converters and inverters) will be superimposed at different locations on the power line. These noises may submerge and interfere with the traveling wave signal. Therefore, compared with the reference node, the traveling wave signal extracted at the node to be corrected may be inaccurate, have large errors, or even cause misidentification of the traveling wave signal, making it impossible to locate the fault point based on the traveling wave signal of the node to be corrected.

[0028] In this embodiment, the traveling wave signal of each reference node is used to correct the traveling wave signal of any node to be corrected, thereby further ensuring the accuracy of the traveling wave signal extracted at the corrected node.

[0029] As an example, the traveling wave signal of any node to be corrected is corrected using the traveling wave signal of each reference node. The methods include: For any voltage sequence acquired at a node to be corrected, initialize a filter and use the filter to filter the voltage sequence to obtain the first filtered sequence. Linearly normalize the first filtered sequence and the traveling wave signal of the reference node respectively, and obtain the DTW distance between the normalized first filtered sequence and the normalized traveling wave signal of the reference node, which is denoted as the filtering difference.

[0030] The parameters of the filter are optimized using the simulated annealing algorithm to minimize the filtering difference. The first filtering sequence with the minimum filtering difference is denoted as the initial corrected traveling wave signal of any reference node to any node to be corrected.

[0031] The DTW distance is obtained by the DTW algorithm, which is a well-known technique and is not specifically limited in this embodiment. The purpose of linear normalization is to eliminate the difference in signal strength between the first filtered sequence and the traveling wave signal (i.e., to remove the difference in magnitude), so that the filtering difference can be used to describe the difference in the fluctuation trend of the first filtered sequence and the traveling wave signal. Therefore, the filtering difference can also be recorded as the fluctuation trend difference.

[0032] During this process, the traveling wave signal of the reference node has a high strength and is relatively less affected by interference. The traveling wave signal of the node to be corrected (i.e., the initial corrected traveling wave signal) is extracted again based on the traveling wave signal of the reference node to ensure that the fluctuation trend of the initial corrected traveling wave signal can match the traveling wave signal received by the reference node from the same fault point, thus obtaining a more accurate traveling wave signal in the early stage.

[0033] Furthermore, for all reference nodes, corrections are made according to the above method, and each reference node obtains an initial corrected traveling wave signal for any node to be corrected.

[0034] Step S103: Based on the frequency distribution differences of the initial corrected traveling wave signals of all reference nodes for the same node to be corrected, extract stable frequency combinations from the initial corrected traveling wave signals of the node to be corrected.

[0035] When the traveling wave signal generated at the fault point is transmitted to different nodes in the distribution network, it will have different noise interference. This noise interference refers to the noise interference formed after the traveling wave signal undergoes multiple reflections and long-distance attenuation, and different noises are superimposed at different locations on the power line. The traveling wave signals extracted from all reference nodes also contain noise. The noise distribution between reference nodes and between reference nodes and the node to be corrected may also be different. This causes the traveling wave signal of each reference node to still interfere with the initial corrected traveling wave signal of each node to be corrected. Moreover, the traveling wave signals of different reference nodes have different interferences on the initial corrected traveling wave signal of the same node to be corrected. When using the initial corrected traveling wave signal of the node to be corrected to locate the fault point, the location result still has a large error, which is not conducive to the rapid location of the fault point. It is necessary to further extract the accurate traveling wave signal of the node to be corrected based on the initial corrected traveling wave signal.

[0036] This embodiment extracts a stable frequency combination from the initial corrected traveling wave signal of the node to be corrected based on the frequency distribution differences of the initial corrected traveling wave signal of all reference nodes for the same node to be corrected.

[0037] The stable frequency combination of each node to be corrected contains several frequency signals, which represent the frequency distribution of the traveling wave signal when it is transmitted to each node to be corrected, where the noise interference is relatively insignificant. These frequency distributions help to further identify the traveling wave signal of the node to be corrected.

[0038] As an example, based on the frequency distribution differences of the initial corrected traveling wave signals of all reference nodes for the same node to be corrected, the stable frequency combination is extracted from the initial corrected traveling wave signal of the node to be corrected. The steps include: For any node to be corrected, and for each reference node, the initial corrected traveling wave signal of that node is obtained by using the Fourier transform algorithm. The spectral distribution of the initial corrected traveling wave signal includes the amplitude of different frequencies in the initial corrected traveling wave signal.

[0039] For any node to be corrected, each reference node obtains a spectral distribution using the method described above; any frequency in the frequency distribution corresponding to all reference nodes is denoted as the target frequency, and the response and stability of the target frequency are obtained based on the amplitude of the target frequency in all frequency distributions.

[0040] The greater the response of the target frequency, the more significant the response of the target frequency. This indicates that the initial corrected traveling wave signals obtained after different reference nodes correct the traveling wave signal of the node to be corrected all have a significant response at the target frequency. The greater the stability of the target frequency, the more significant the response of the target frequency. Therefore, when both the response and stability of the target frequency are large, it indicates that the signal component at the target frequency in the traveling wave signal is not significantly interfered with when it propagates to the node to be corrected, and this can be used as a basis for identifying the traveling wave signal.

[0041] In this embodiment, the product of the target frequency's responsiveness and stability is denoted as the target frequency's stability index. For all frequencies in the spectral distribution, the N1 frequencies with the highest stability index are selected as a stable frequency combination. Here, P1 represents the target frequency's responsiveness; P2 represents the target frequency's stability; and N1 represents the stable frequency extraction parameter for the node to be corrected. In this embodiment, the initial value of N1 is set to 5.

[0042] Thus, for any node to be corrected, a stable frequency combination is obtained by extracting parameters from the stable frequency of the node. For all nodes to be corrected, the corresponding stable frequency combinations are obtained using the above method.

[0043] As an example, methods for obtaining the response and stability of a target frequency based on the amplitude of the target frequency across all frequency distributions include: Obtain the mean value of the target frequency's amplitude across all frequency distributions, and use it as the response level of the target frequency.

[0044] The amplitude of the target frequency in all frequency distributions is linearly normalized, and the standard deviation of all linearly normalized amplitudes at the target frequency is denoted as x. exp(-x) is used as the stability of the target frequency. Here, exp() represents an exponential function with the natural constant as the base. The smaller x is, the more similar the initial corrected traveling wave signal's response is at the target frequency, and the greater the stability of the target frequency.

[0045] Step S104: For each stable frequency combination of a node to be corrected, obtain the difference between the traveling wave signals of all reference nodes and the initial corrected traveling wave signals of all nodes to be corrected under the stable frequency combination, which is abbreviated as the overall distribution difference of the stable frequency combination.

[0046] As an optional example, for any stable frequency combination of a node to be corrected, the amplitude of all frequencies other than those included in the stable frequency combination in the spectrum distribution of the node to be corrected is set to 0 to obtain the filtered spectrum distribution of any node to be corrected. The filtered spectrum distribution is then inversely transformed using the inverse Fourier transform algorithm to obtain the final traveling wave signal of any node to be corrected. Finally, the traveling wave signals of all reference nodes and the final traveling wave signals of all nodes to be corrected are used to locate the fault point.

[0047] In this optional example, the final traveling wave signal can relatively accurately represent the signal when the traveling wave generated at the fault point is transmitted to the node to be corrected.

[0048] However, this optional example has the following problems: the final traveling wave signal of the node to be corrected only includes a part of the signal components of the complete traveling wave signal, or a part of the signal components of the traveling wave signal of the reference node, and the final traveling wave signal of different nodes to be corrected may represent different parts of the traveling wave signal.

[0049] However, when using the traveling wave signals detected and identified by all nodes to locate fault points, it is necessary to capture the arrival time of the traveling wave. When the final traveling wave signal of the node to be corrected has the above-mentioned problem, there will be a delay error in capturing the arrival time of the traveling wave, resulting in errors in the fault point location results.

[0050] Based on this, in this embodiment, for each stable frequency combination of correction nodes, the difference between the traveling wave signals of all reference nodes and the initial corrected traveling wave signals of all nodes to be corrected under the stable frequency combination is obtained, which is simply referred to as the overall distribution difference of the stable frequency combination.

[0051] The overall distribution difference of the stable frequency combination represents the difference between the extracted traveling wave signals when each stable frequency combination is used to re-extract the traveling wave signals of all nodes. The greater the difference, the more difficult it is for each stable frequency combination to extract traveling wave signals with the same or similar components from all nodes, thus making it impossible to avoid the problem of significant errors in the arrival time of the traveling wave.

[0052] As an example, for each stable frequency combination of correction nodes, the differences between the traveling wave signals of all reference nodes and the initial corrected traveling wave signals of all nodes to be corrected under the stable frequency combination are obtained, which is abbreviated as the overall distribution difference of the stable frequency combination. The steps include the following: For any stable frequency combination among all the stable frequency combinations of all correction nodes.

[0053] (1) Re-extract the traveling wave signal at any node using this stable frequency combination. This includes: The traveling wave signal of any reference node or the initial corrected traveling wave signal of any node to be corrected is denoted as the fundamental signal of any node; the spectral distribution F1 of the fundamental signal is obtained using the Fourier transform algorithm.

[0054] In F1, the amplitude of all frequencies other than those included in the stable frequency combination is set to 0, denoted as the filter spectrum distribution of any node. The filter spectrum distribution is then inversely transformed using the inverse Fourier transform algorithm to obtain the final signal of any node. This final signal represents the traveling wave signal re-extracted using the stable frequency combination at any node.

[0055] (2) Following the method in (1), the final signal at each node is re-extracted using the stable frequency combination.

[0056] The final signals of all nodes are linearly normalized. The DTW distance between any two normalized final signals is denoted as the trend difference between the two nodes. The smaller the trend difference, the more similar or identical the fluctuation patterns of the final signals of the two nodes. The mean of the trend differences among all nodes is denoted as the overall distributional difference of this stable frequency combination.

[0057] Thus, an overall distributional difference was obtained for each arbitrary combination of stable frequencies.

[0058] Step S105: For all stable frequency combinations of nodes to be corrected, reorganize all stable frequency combinations according to the overall distribution difference of all stable frequency combinations to obtain a new stable frequency combination, so that the overall distribution difference of the new stable frequency combination is minimized.

[0059] In this embodiment, to avoid the problem of delayed arrival time errors in the stable frequency combinations obtained when correcting the traveling wave signal of each node to be corrected, all stable frequency combinations are recombined to obtain new stable frequency combinations. These new stable frequency combinations not only help to further identify the traveling wave signal of the node to be corrected, but also minimize the problem of delayed arrival time errors in the traveling wave.

[0060] As an example, all stable frequency combinations are recombined to obtain new stable frequency combinations, specifically including: The stable frequency combination of all nodes to be corrected is represented as set S1, and each stable frequency combination in set S1 corresponds to an overall distribution difference.

[0061] (1) Frequency recombination based on set S1, including: For all stable frequency combinations in set S1, the stable frequency combination with the smallest overall distribution difference is selected and denoted as Q1. Since the stable frequency combination Q1 has the smallest overall distribution difference, it is more helpful to avoid the problem of delay error in the arrival time of the traveling wave. Therefore, the stable frequency combination Q1 is used as the benchmark for frequency reorganization.

[0062] Specifically, in set S1, the stable frequency combinations other than stable frequency combination Q1 are denoted as reference combinations. The overall distribution differences of all reference combinations are normalized using the softmax formula, and the normalized overall distribution differences are denoted as the attention coefficient of each reference combination.

[0063] Obtain the overlap between the frequencies contained in each reference combination and the stable frequency combination Q1. The product of this overlap and the attention coefficient of each reference combination is recorded as the merging index of each reference combination. Merge all the frequencies contained in the reference combination with the largest merging index into the stable frequency combination Q1. The resulting merged stable frequency combination Q1 is simply referred to as the merged combination.

[0064] Then, follow step S104 to obtain the overall distribution difference of the merged combination.

[0065] (2) Update the stable frequency combination for each node to be corrected: The overall distribution difference of the stable frequency combinations of all nodes to be corrected is normalized using the softmax formula, and the normalized overall distribution difference is recorded as the update scale of each node to be corrected.

[0066] The stable frequency extraction parameters of each node to be corrected are modified by using the update scale. The modified stable frequency extraction parameters are positively correlated with the update scale. That is, the larger the update scale (i.e. the greater the overall distribution difference), the less favorable it is for the stable frequency combination to avoid the problem of delayed arrival time error of traveling waves. In this case, it is more necessary to significantly modify the stable frequency extraction parameters and obtain a new stable frequency combination.

[0067] Specifically, the stable frequency combination of each node to be corrected is obtained again by extracting the parameters of the stable frequency after the correction of each node (see step S103), and the overall distribution difference of each stable frequency combination is obtained again.

[0068] Represent the stable frequency combinations obtained by all nodes to be corrected and the merged combination obtained in (1) as set S1. Note that the stable frequency combinations in set S1 at this time include the stable frequency combinations obtained by all nodes to be corrected and the merged stable frequency combination Q1 (i.e., the merged combination).

[0069] (3) Repeat steps (1) and (2) several times (e.g., 5 times). After each execution of steps (1) and (2), a set S1 is obtained. After the repeated execution ends, for all sets S1 obtained during the repeated execution, each element in all sets S1 is a stable frequency combination or a merged combination, and each element corresponds to an overall distribution difference. The element with the smallest overall distribution difference is selected as the new stable frequency combination.

[0070] Step S106: Use the new stable frequency combination to re-extract the traveling wave signal of the node and locate the fault point.

[0071] Following the method in step S104 (1), the traveling wave signal at any node is re-extracted using the new stable frequency combination, and denoted as the traveling wave filtered signal. The specific process includes: The traveling wave signal of any reference node or the initial corrected traveling wave signal of any node to be corrected is denoted as the fundamental signal of any node; the spectral distribution F2 of the fundamental signal is obtained using the Fourier transform algorithm.

[0072] In F2, the amplitude of all frequencies other than those included in the new stable frequency combination is set to 0, denoted as the filter spectrum distribution of any node. The filter spectrum distribution is then inversely transformed using the inverse Fourier transform algorithm to obtain the traveling wave filter signal of any node. This traveling wave filter signal represents the traveling wave signal re-extracted at any node using the new stable frequency combination.

[0073] Fault location is achieved by using traveling wave filtered signals at all nodes.

[0074] This concludes the example.

[0075] Example 2: In Example 1, step S105 includes: obtaining the overlap between the frequencies contained in each reference combination and the stable frequency combination Q1; as an example, this process includes: The frequencies contained in each reference combination are denoted as the first frequency set. The first, second, and third intervals of all frequencies in the first frequency set are obtained, and the three-dimensional vector formed by the first, second, and third intervals is denoted as the distribution vector G1 of the first frequency set. Obtain the first, second, and third intervals of all frequencies in the second frequency set, and denote the three-dimensional vector formed by the first, second, and third intervals as the distribution vector G2 of the second frequency set. The cosine similarity between the distribution vector G1 of the first frequency set and the distribution vector G2 of the second frequency set is denoted as the overlap degree. The greater the overlap degree, the more likely that the frequencies in each reference combination and the stable frequency combination Q1 have the same or similar distribution patterns.

[0076] In Example 1, step S105 includes: modifying the stable frequency extraction parameters of each node to be corrected using the update scale; as an example, this process includes: The modified stable frequency extraction parameters for each node to be corrected are denoted as N2. .

[0077] Where k represents the update scale, and N1 represents the initial values ​​of the stable frequency extraction parameters. This indicates the parameters extracted from the stable frequency before modification. This indicates rounding down to the nearest integer.

[0078] It should be noted that when the stable frequency extraction parameters of each node to be corrected are modified for the first time, the stable frequency extraction parameter N0 before modification is equal to N1; when the stable frequency extraction parameters are modified for the second time, the stable frequency extraction parameter N0 before modification is equal to the stable frequency extraction parameter after the first modification; when the stable frequency extraction parameters are modified for the third time, the stable frequency extraction parameter N0 before modification is equal to the stable frequency extraction parameter after the second modification; and so on.

[0079] Specifically, when the update scale k is less than 0.2, this embodiment sets the update scale k to 0, indicating that no modification is made.

[0080] Example 3: Step S101 in Example 1 includes: filtering the voltage sequence and extracting the traveling wave signal at each node. As an example, this process specifically includes: First, it should be noted that the power line in this embodiment is a line in a 220V low-voltage distribution network. The traveling wave detector at each node uses Beidou time synchronization, and all traveling wave detectors collect data synchronously. Each traveling wave detector collects the voltage magnitude once every 1 millisecond.

[0081] In this embodiment, each second is considered a moment. The time sequence consisting of the voltages collected at several moments before each moment (e.g., 10 moments before each moment, including the current moment) is denoted as the voltage sequence collected at each moment.

[0082] At any node, for a preset time period prior to the current time (excluding the current time) (e.g., within 2 hours prior to the current time), for each moment within the preset time period, the spectral distribution of the voltage sequence at each moment is obtained using the Fourier transform algorithm. For the spectral distributions at all moments within the preset time period, these spectral distributions are averaged (i.e., the average of all amplitudes corresponding to the same frequency) to obtain the average spectral distribution F0. This average spectral distribution F0 represents the frequency distribution of the distribution network during normal operation in its historical working process at any node.

[0083] The spectral distribution of the voltage sequence at the current moment is obtained using the Fourier transform algorithm. The difference between this frequency distribution and the average spectral distribution F0 is denoted as the traveling wave spectral distribution. Specifically, the difference between this frequency distribution and the average spectral distribution F0 refers to the difference between the amplitude of the former and the amplitude of the latter at the same frequency (when the difference is less than 0, the difference is set to 0).

[0084] The inverse Fourier transform algorithm is used to transform the spectral distribution of the traveling wave to obtain the traveling wave signal at each node.

[0085] This process is equivalent to removing the frequency distribution of the power grid during normal operation from the frequency distribution of the voltage sequence at the current moment, that is, filtering the voltage sequence at the current moment to obtain a traveling wave signal.

[0086] Specifically, when the maximum value of the traveling wave signal approaches 0 (in this embodiment, a signal value less than 5% of the 220V voltage is considered to be approaching 0; otherwise, it is considered to be approaching 0), each node is marked as a node where the traveling wave is not detected. If all nodes are marked as nodes where the traveling wave is not detected at the current moment, it indicates that no fault point has occurred in the distribution network at the current moment, and no traveling wave signal has been detected at any node. The subsequent steps of Embodiment 1 are then not executed. If the maximum value of the traveling wave signal of one or more nodes does not approach 0 at the current moment, it is determined that all nodes have detected the traveling wave at the current moment, and the subsequent steps of Embodiment 1 are executed to locate the fault point.

[0087] Step S102 of Example 1 includes: initializing a filter, using the filter to filter the voltage sequence to obtain a first filtered sequence; it also includes: optimizing the filter parameters using a simulated annealing algorithm to minimize the filtering difference; as an example, the specific steps included in this process are: For each frequency in the average spectral distribution F0, a scaling factor is assigned to each frequency. The scaling factor is randomly initialized, and the value of each scaling factor ranges from [0, 1]. The scaling factor of each frequency is multiplied by the amplitude of each frequency in the average spectral distribution F0, and the result is used as the adjusted spectral distribution. This adjusted spectral distribution is essentially an initialized filter, where the scaling factor of each frequency is regarded as the filter parameter.

[0088] The spectral distribution of the voltage sequence is obtained using the Fourier transform algorithm. The difference between this frequency distribution and the adjusted spectral distribution is denoted as the difference spectral distribution. The difference spectral distribution is then transformed using the inverse Fourier transform algorithm to obtain the first filtered sequence.

[0089] In other embodiments, the method for obtaining the first filtered sequence further includes: A Gaussian filter with a kernel length of 11 can be used as the initial filter, where the kernel length is a parameter of the filter. This filter is then used to filter the voltage sequence, resulting in a low-pass filtered sequence. The difference between the voltage sequence and the low-pass filtered sequence is recorded as the first filtered sequence. It should be noted that in this embodiment, the traveling wave signal is a high-frequency signal; however, the Gaussian filter is a low-pass filter. Therefore, the difference between the voltage sequence and the low-pass filtered sequence needs to be recorded as the first filtered sequence to achieve high-pass filtering.

[0090] Furthermore, the parameters of the filter are optimized using the simulated annealing algorithm (i.e., the scaling factor or Gaussian kernel length for each frequency) to minimize the filtering differences.

[0091] Example 4: Step S106 of Example 1 includes: locating the fault point using the traveling wave filtered signals at all nodes; as an example, the specific process includes: For the traveling wave filtered signal at each node, obtain the arrival time of the traveling wave.

[0092] As an example, the time of the first maximum point of the traveling wave filtered signal is denoted as the arrival time of the traveling wave.

[0093] Further, the arrival times of traveling waves at all nodes are used to locate the fault point. Specifically, this includes: In the power distribution network, some candidate points are marked, for example, one candidate point is marked every 50 meters on each power line.

[0094] For any candidate point, the ratio of the line length from the candidate point to each node to the propagation speed of the traveling wave signal is taken as the time for the traveling wave signal generated by the candidate point to reach each node, and is denoted as the theoretical time from any candidate point to each node.

[0095] The traveling wave signal propagates at 98% of the speed of light.

[0096] For any candidate point, the theoretical times of all nodes are arranged into a theoretical time series. Additionally, the arrival times of the traveling waves of all nodes are also arranged into an arrival time series. The nodes are arranged in ascending order of their ID numbers; however, other orders may be used in other embodiments, as long as the sorting method for the theoretical time series and the arrival time series is the same.

[0097] Obtain the Pierre correlation coefficient between the theoretical time series and the arrival time series, and denote it as the candidate index of any candidate point. Obtain the candidate point with the largest candidate index as the fault location point.

[0098] Example 5: This embodiment provides a power line fault location device for power engineering construction. The device includes traveling wave detectors installed at all nodes. The device also includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor reads the voltage sequence collected by the traveling wave detector and runs the computer program, it executes all the steps of all the above embodiments.

[0099] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for locating power line fault points in power engineering construction, characterized in that, The method includes the following steps: Traveling wave signals are detected at multiple nodes of the power line. The nodes with the strongest traveling wave signal intensity are designated as reference nodes, and the nodes other than the reference nodes are designated as nodes to be corrected. The traveling wave signal of any node to be corrected is corrected using the traveling wave signal of each reference node to obtain the initial corrected traveling wave signal of any node to be corrected. Based on the frequency distribution differences of the initial correction traveling wave signals of all reference nodes for the same node to be corrected, a stable frequency combination is extracted from the initial correction traveling wave signal of the node to be corrected. The stable frequency combination contains several frequency signals. For each stable frequency combination of a node to be corrected, the difference between the traveling wave signals of all reference nodes and the initial corrected traveling wave signals of all nodes to be corrected under the stable frequency combination is obtained, which is simply referred to as the overall distribution difference of the stable frequency combination. For all stable frequency combinations of nodes to be corrected, all stable frequency combinations are reorganized according to the overall distribution difference of all stable frequency combinations to obtain a new stable frequency combination, such that the overall distribution difference of the new stable frequency combination is minimized. The traveling wave signal of the node is re-extracted using the new stable frequency combination, and the fault point is located.

2. The method for locating power line fault points in power engineering construction according to claim 1, characterized in that, The specific steps involved in correcting the traveling wave signal of any node to be corrected using the traveling wave signal of each reference node to obtain the initial corrected traveling wave signal of any node to be corrected are as follows: For any voltage sequence acquired at a node to be corrected, initialize a filter and use the filter to filter the voltage sequence to obtain the first filtered sequence. Then, linearly normalize the first filtered sequence and the traveling wave signal of the reference node. The DTW distance between the normalized first filtered sequence and the normalized traveling wave signal of the reference node is denoted as the fluctuation trend difference. The parameters of the filter are optimized using the simulated annealing algorithm to minimize the difference in fluctuation trends. The first filtered sequence with the smallest difference in fluctuation trends is denoted as the initial corrected traveling wave signal of any reference node to any node to be corrected.

3. The method for locating power line fault points in power engineering construction according to claim 1, characterized in that, The specific steps for extracting stable frequency combinations from the initial corrected traveling wave signal of the node to be corrected based on the frequency distribution differences of the initial corrected traveling wave signal of all reference nodes for the same node to be corrected include the following: For any node to be corrected, and for each reference node, the initial corrected traveling wave signal of the node to be corrected is obtained by using the Fourier transform algorithm. The spectral distribution of the initial corrected traveling wave signal includes the amplitude of each frequency in the initial corrected traveling wave signal. Any frequency in the frequency distribution corresponding to all reference nodes is denoted as the target frequency. The response and stability of the target frequency are obtained based on the amplitude of the target frequency in the frequency distribution corresponding to all reference nodes. The product of the response and stability of the target frequency is denoted as the stability index of the target frequency. Among all frequencies in the spectral distribution corresponding to all reference nodes, the N1 frequencies with the largest stability index are selected as the stable frequency combination. N1 is the stable frequency extraction parameter of the node to be corrected; the initial value of the stable frequency extraction parameter is a preset value.

4. The method for locating power line fault points in power engineering construction according to claim 1, characterized in that, For each stable frequency combination of the node to be corrected, the difference between the traveling wave signals of all reference nodes and the initial corrected traveling wave signals of all nodes to be corrected under the stable frequency combination is obtained, which is referred to as the overall distribution difference of the stable frequency combination. The specific steps include the following: For any stable frequency combination among all the stable frequency combinations of the correction nodes; use this stable frequency combination to re-extract the traveling wave signal at any node, including: The traveling wave signal of any reference node or the initial corrected traveling wave signal of any node to be corrected is denoted as the basic signal of any node. The spectral distribution F1 of the basic signal is obtained by using the Fourier transform algorithm. The amplitude of all frequencies other than those included in the stable frequency combination in F1 is set to 0, which is denoted as the filtered spectral distribution of any node. The filtered spectral distribution is inversely transformed by the inverse Fourier transform algorithm to obtain the final signal of any node. The final signal represents the re-extracted traveling wave signal. The final signals of all nodes are linearly normalized, and the DTW distance between any two normalized final signals of any two nodes is denoted as the trend difference between any two nodes; the mean of the trend differences between all nodes is denoted as the overall distribution difference of the stable frequency combination.

5. The method for locating power line fault points in power engineering construction according to claim 3, characterized in that, For all stable frequency combinations of nodes to be corrected, the stable frequency combinations are reorganized according to the overall distribution differences of all stable frequency combinations to obtain a new stable frequency combination, which minimizes the overall distribution differences of the new stable frequency combination. The specific steps include the following: D1: The set S1 represents the combination of stable frequencies of all nodes to be corrected; D2: For all stable frequency combinations in set S1, obtain the stable frequency combination Q1 with the smallest overall distribution difference; in set S1, stable frequency combinations other than stable frequency combination Q1 are denoted as reference combinations. Based on the overall distribution difference of the reference combinations and the overlap between the frequencies contained in each reference combination and stable frequency combination Q1, obtain the merging index of each reference combination; merge the reference combination with the largest merging index into stable frequency combination Q1, and denote the resulting stable frequency combination Q1 as the merged combination, and obtain the overall distribution difference of the merged combination; D3: Modify the stable frequency extraction parameters of each node to be corrected by utilizing the overall distribution difference of the stable frequency combination of each node to be corrected. The modified stable frequency extraction parameters are positively correlated with the overall distribution difference. Using the modified stable frequency extraction parameters of each node to be corrected, the stable frequency combination of each node to be corrected is obtained again, and the overall distribution difference of each stable frequency combination is obtained again; the stable frequency combination of all nodes to be corrected and the merged combination are re-represented as set S1; D4: Repeat D2 and D3 several times; after the repeated execution is completed, obtain the stable frequency combination or merged combination with the smallest overall distribution difference from all the obtained sets S1, and use it as the new stable frequency combination.

6. The method for locating power line fault points in power engineering construction according to claim 4, characterized in that, The specific steps involved in re-extracting the traveling wave signal of the node using a new stable frequency combination and locating the fault point are as follows: The traveling wave signal at any node is re-extracted using the new stable frequency combination, denoted as the traveling wave filtered signal, and the arrival time of the traveling wave filtered signal is obtained. In the distribution network, mark several candidate points, obtain the time of the traveling wave signal generated by any candidate point to each node, and record it as the theoretical time of any candidate point to each node; The theoretical time series is formed by any candidate point with respect to the theoretical time of all nodes, and the arrival time of all nodes is formed with respect to the arrival time series. The Pierre correlation coefficient between the theoretical time series and the arrival time series is obtained and denoted as the candidate index of any candidate point. The candidate point with the largest candidate index is selected as the fault location point.

7. The method for locating power line fault points in power engineering construction according to claim 3, characterized in that, The specific steps for obtaining the response and stability of the target frequency based on the amplitude of the target frequency in the frequency distribution corresponding to all reference nodes are as follows: For the frequency distribution corresponding to all reference nodes, obtain the mean value of the target frequency amplitude in all frequency distributions, and use it as the response degree of the target frequency; The amplitude of the target frequency in all frequency distributions is linearly normalized, and the standard deviation of all linearly normalized amplitudes at the target frequency is denoted as x. exp(-x) is used as the stability of the target frequency. Where exp() represents an exponential function with the natural constant as the base.

8. The method for locating power line fault points in power engineering construction according to claim 5, characterized in that, The specific steps for obtaining the merging index of each reference combination based on the overall distribution differences of the reference combinations and the degree of overlap between the frequencies contained in each reference combination and the stable frequency combination Q1 are as follows: The overall distributional differences of all reference combinations are normalized and used as the attention coefficient for each reference combination. The first, second, and third intervals of the frequencies contained in each reference combination are denoted as distribution vector G1; the first, second, and third intervals of the frequencies contained in the stable frequency combination Q1 are denoted as distribution vector G2; the cosine similarity between distribution vector G1 and distribution vector G2 is denoted as the overlap degree; the product of the overlap degree and the attention coefficient of each reference combination is denoted as the merging index of each reference combination.

9. The method for locating power line fault points in power engineering construction according to claim 5, characterized in that, The specific steps involved in modifying the stable frequency extraction parameters of each node to be corrected by utilizing the overall distribution differences of the stable frequency combinations of each node to be corrected are as follows: The overall distribution difference of the stable frequency combinations of all nodes to be corrected is normalized and used as the update scale k for each node to be corrected; the modified stable frequency extraction parameters for each node to be corrected are denoted as N2. Where N1 represents the initial value of the stable frequency extraction parameters. This indicates the parameters extracted from the stable frequency before modification. This indicates rounding down to the nearest integer.

10. A power line fault location device applied in power engineering construction, the device comprising traveling wave detectors installed at all nodes, the device further comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor reads the voltage signal collected by the traveling wave detector and runs the computer program, it executes all the steps of the power line fault location method applied to power engineering construction as described in any one of claims 1 to 9.

Images