Power distribution network fault positioning method, device and equipment based on double-end positioning and medium
By using a dual-end positioning method and utilizing waveform recording files and neural network models, the problems of large errors and high costs in traditional power distribution network fault location are solved, and high-precision fault location in complex networks is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- GUANGZHOU POWER SUPPLY BUREAU GUANGDONG POWER GRID CO LTD
- Filing Date
- 2025-06-11
- Publication Date
- 2026-06-09
AI Technical Summary
Traditional fault location methods for power distribution networks suffer from large errors, high hardware costs, and difficulty in large-scale deployment. In particular, severe signal attenuation occurs in mixed cable-overhead lines or complex branch networks, leading to inaccurate location accuracy.
A dual-end positioning method is adopted. By acquiring waveform files from both ends of the faulty line, extracting the relative zero time, performing wavelet transform to obtain the voltage traveling wave mode component, calculating the relational feature value of the overlapping wavefront, and using a neural network model to locate the fault point, the problems of time synchronization and wave velocity normalization are avoided.
It improves the accuracy of fault location in distribution networks, reduces hardware costs, is suitable for fault location in complex networks, and reduces errors.
Smart Images
Figure CN120669050B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power distribution network fault monitoring technology, and in particular to a power distribution network fault location method, device, equipment and medium based on dual-end positioning. Background Technology
[0002] Traditional power distribution networks typically rely on manual troubleshooting, which is time-consuming and labor-intensive. Therefore, accurate fault diagnosis of power distribution networks is of great significance in order to significantly reduce power outage time caused by fault repair.
[0003] Currently, the main method for fault location in distribution networks is the traveling wave method, which is further divided into single-ended and double-ended traveling wave methods. The former utilizes the time difference of the neutral mode wavefront for location and often uses the theoretical wave speed in practical applications. However, the propagation speed of the traveling wave is affected by line parameters (such as conductor type, insulation medium, and aging degree), resulting in wave speed instability and large error fluctuations. On the other hand, distribution networks typically contain a large number of branch lines and short-distance feeders. The traveling wave undergoes reflection and refraction at multiple nodes, leading to severe waveform distortion, and the overlapping of the neutral and neutral mode wavefronts makes it difficult to distinguish. The double-ended method only requires the extraction of the first wavefront, but the data from both ends need to be strictly synchronized, and the error needs to be controlled within the microsecond level, or even within 50 nanoseconds. Both traveling wave location methods require accurate extraction of the wavefront, but the high-frequency components of the traveling wave signal are easily attenuated during propagation, especially in cable-overhead mixed lines or complex branch networks. Due to high-frequency attenuation and noise interference, the signal becomes weak, easily leading to misjudgment, affecting the accuracy of time difference calculation, and resulting in large location accuracy errors. Traveling wave positioning requires high-frequency sampling devices (MHz level) and needs to be widely installed at key nodes, resulting in high hardware costs and making it difficult to promote on a large scale in the power distribution network. Summary of the Invention
[0004] This invention provides a method, device, equipment, and medium for fault location in power distribution networks based on dual-end positioning, which can improve the accuracy of fault location in power distribution networks.
[0005] In a first aspect, embodiments of the present invention provide a method for fault location in a distribution network based on dual-end positioning, comprising:
[0006] Obtain the waveform recording files of the dual-end positioning devices at both ends of the faulty line, and extract the start timestamps of the two waveform recording files respectively to obtain the relative zero time when fault waveform recording is started at both ends of the faulty line.
[0007] Align the relative zero time and extract the first voltage traveling wave data of both ends of the fault line within a preset time period before and after the relative zero time, so as to obtain the first voltage traveling wave line mode components of both ends of the fault line based on the first voltage traveling wave data.
[0008] Wavelet transform is performed on the first voltage traveling wave line mode component to obtain the first time domain waveform diagrams of different layers, and the first wavefront time series of both ends of the fault line are obtained according to the first time domain waveform diagrams.
[0009] Based on the time series of the first wavefronts at both ends of the faulty line, the first overlapping wavefront and the second overlapping wavefront are obtained, and the relationship feature value between the first overlapping wavefront and the second overlapping wavefront is calculated.
[0010] The relationship feature value and the upstream and downstream time difference of the fault point are input into a preset neural network model, the fault point distance is output, and the fault location of the distribution network is performed based on the fault point distance; wherein, the upstream and downstream time difference of the fault point is obtained based on the waveform file.
[0011] This invention aligns fault waveform data at both ends of the faulty line relative to time zero, providing data preparation for subsequent acquisition of the first voltage traveling wave mode components at both ends, and avoiding the problem of strict time synchronization at both ends required in the traditional double-ended traveling wave method. By obtaining the first wavefront time series at both ends of the faulty line based on the first voltage traveling wave mode components, it provides data preparation for subsequent acquisition of the first and second overlapping wavefronts. By calculating the relationship feature values between the first and second overlapping wavefronts, it provides data support for subsequent fault location using a preset neural network model, and by using these relationship feature values as input values to the neural network model, it avoids the problem of traditional... The double-ended traveling wave method requires wave velocity normalization. The neural network model obtains the distance to the fault point based on the relationship feature values and the time difference between the upstream and downstream sides of the fault point. Traditional double-ended traveling wave methods require strict time synchronization and wave velocity normalization. However, the traveling wave propagation speed is affected by the actual physical parameters of the line (such as conductor type, insulation medium, and aging degree), resulting in wave velocity instability. This makes it difficult to fully achieve time synchronization and wave velocity normalization, leading to a large error in the final fault location result. This invention avoids the problem of requiring time synchronization and wave velocity normalization by using the sum of the relationship features and the time difference between the upstream and downstream sides of the fault point as the input value of the neural network model. Therefore, compared with existing technologies, this invention can improve the accuracy of fault location in distribution networks.
[0012] Furthermore, the step of obtaining the first voltage traveling wave line mode components at both ends of the faulty line based on the first voltage traveling wave data specifically involves:
[0013] The first voltage traveling wave data is subjected to Karlenbauer transform to obtain the first voltage traveling wave line mode components corresponding to both ends of the faulty line.
[0014] In this embodiment of the invention, the first voltage traveling wave line mode components at both ends of the faulty line are obtained based on the first voltage traveling wave data, providing data support for the subsequent acquisition of the first wavefront time series.
[0015] Furthermore, the step of performing wavelet transform on the first voltage traveling wave mode components to obtain time-domain waveforms at different levels is specifically as follows:
[0016] Wavelet transform is performed on the first voltage traveling wave line mode components to obtain the high-frequency coefficients of their respective frequency bands. The corresponding first voltage traveling wave line mode components are then reconstructed based on the high-frequency coefficients to obtain time-domain waveform diagrams of different layers at both ends of the faulted line.
[0017] In this embodiment of the invention, wavelet transform is performed on the first voltage traveling wave line mode component to obtain time-domain waveforms at different levels, providing data preparation for subsequently obtaining the first wavefront time series at both ends of the faulty line.
[0018] Furthermore, the step of obtaining the first wavefront time series at both ends of the faulty line based on the time-domain waveform diagram specifically involves:
[0019] Take the maximum modulus value of each layer in the time-domain waveform diagram, and set the modulus values that are less than the maximum modulus value by a preset percentage to 0;
[0020] Extract points with non-zero modulus values and their corresponding time coordinates to construct the first wavefront time series at both ends of the faulty line.
[0021] In this embodiment of the invention, by setting the modulus value of a preset percentage less than the maximum modulus value to 0, the first wavefront time series at both ends of the faulty line are constructed respectively, which can avoid noise interference and thus improve the accuracy of subsequent fault location.
[0022] Furthermore, the step of obtaining the first overlapping wavefront and the second overlapping wavefront based on the time sequence of the first wavefront at both ends of the faulty line specifically involves:
[0023] The first wavefront in the time series of the first wavefront at both ends of the faulty line is extracted respectively, and the first wavefront is aligned by time shifting to obtain the first overlapping wavefront;
[0024] Continuing to search along the time coordinate, extract the wavefronts that are aligned again to obtain the second overlapping wavefront.
[0025] This invention provides data preparation for subsequent calculation of relational feature values by acquiring the first overlapping wavefront and the second overlapping wavefront.
[0026] Furthermore, the relational characteristic values between the first overlapping wavefront and the second overlapping wavefront include the modulus ratio between the first wavefronts at both ends of the faulty line, the modulus ratio between the realigned wavefronts, and the time difference between the time coordinates corresponding to the first overlapping wavefront and the time coordinates corresponding to the second overlapping wavefront.
[0027] This invention uses rich and multi-dimensional relational feature values as input values for a neural network model to improve the accuracy of fault location results.
[0028] Furthermore, obtaining the upstream and downstream time difference of the fault point based on the recorded waveform file specifically involves:
[0029] The initial recording time of each recording file is obtained, and the recording data in each recording file is aligned to absolute zero time according to the initial recording time.
[0030] Extract the second voltage traveling wave data of both ends of the fault line within a preset time period before and after the absolute zero moment, so as to obtain the second voltage traveling wave line mode components of both ends of the fault line based on the second voltage traveling wave data.
[0031] Wavelet transform is performed on the second voltage traveling wave line mode component to obtain the second time domain waveform diagrams at different levels, and the second wavefront time series at both ends of the fault line are obtained based on the second time domain waveform diagrams.
[0032] Find the initial wavefront of the second wavefront time series respectively, and calculate the time from the time corresponding to the initial wavefront to the respective initial recording time to obtain the upstream initial wavefront time and the downstream initial wavefront time;
[0033] Take the absolute value of the difference between the upstream initial wavefront time and the downstream initial wavefront time, and use the absolute value as the upstream and downstream time difference of the fault point.
[0034] This invention improves the accuracy of the fault distance output by the neural network model by calculating the time difference between the upstream and downstream of the fault point and inputting the time difference into the neural network model.
[0035] Secondly, embodiments of the present invention provide a distribution network fault location device based on dual-end positioning, including a relative zero-time acquisition module, a first voltage traveling wave line mode component module, a first wavefront time series module, a relationship feature value module, an upstream and downstream time difference module, and a fault point distance acquisition module, wherein...
[0036] The relative zero-time acquisition module is used to acquire the waveform recording files of the dual-end positioning devices at both ends of the faulty line, and extract the start timestamps of the two waveform recording files respectively to obtain the relative zero-time when fault waveform recording is started at both ends of the faulty line.
[0037] The first voltage traveling wave line mode component module is used to align the relative zero time and extract the first voltage traveling wave data at both ends of the fault line within a preset time period before and after the relative zero time, so as to obtain the first voltage traveling wave line mode components at both ends of the fault line according to the first voltage traveling wave data.
[0038] The first wavefront time series module is used to perform wavelet transform on the first voltage traveling wave line mode component to obtain the first time domain waveform diagrams of different layers, and to obtain the first wavefront time series at both ends of the fault line based on the first time domain waveform diagrams.
[0039] The relational feature value module is used to obtain the first overlapping wavefront and the second overlapping wavefront based on the first wavefront time series at both ends of the faulty line, and to calculate the relational feature value between the first overlapping wavefront and the second overlapping wavefront.
[0040] The fault point distance acquisition module is used to input the relationship feature value and the upstream and downstream time difference of the fault point into a preset neural network model, output the fault point distance, and perform distribution network fault location based on the fault point distance; wherein, the upstream and downstream time difference of the fault point is obtained based on the waveform file.
[0041] This invention employs a relative zero-time acquisition module to acquire the relative zero-time of fault waveform data in the waveform recording file, providing data support for subsequent relative zero-time alignment; a first voltage traveling wave mode component module to acquire the first voltage traveling wave mode components at both ends of the fault line based on the first voltage traveling wave data, providing data support for subsequent finding of overlapping wavefronts; a first wavefront time series module to acquire the first wavefront time series at both ends of the fault line, providing data preparation for subsequent calculation of the relationship feature values; and a relationship feature value module to calculate the relationship feature values between the first overlapping wavefront and the second overlapping wavefront, providing rich and comprehensive input values for the subsequent neural network model to improve the model output results. The accuracy of fault location is improved by using a fault point distance acquisition module to input the relationship feature value and the upstream and downstream time difference of the fault point into a preset neural network model. In the traditional double-end traveling wave method, strict time synchronization and wave velocity normalization are required to obtain wave velocity parameters and calculate time differences. However, due to the influence of uncontrollable external factors such as line aging in actual transmission lines, it is difficult to accurately obtain wave velocity parameters, thus affecting the accuracy of the final fault location result. This invention uses the relationship feature value obtained by relative zero-time alignment and the upstream and downstream time difference of the fault point as model input data, thereby avoiding the problem of difficulty in accurately obtaining wave velocity parameters in the prior art and improving the accuracy of distribution network fault location.
[0042] Thirdly, embodiments of the present invention provide a terminal device, including: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus;
[0043] The memory is used to store at least one executable instruction that causes the processor to perform the operation of the power distribution network fault location method based on dual-end positioning as described above.
[0044] Fourthly, embodiments of the present invention provide a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device / apparatus containing the computer-readable storage medium to perform the power distribution network fault location method based on dual-end positioning as described above.
[0045] The above description is merely an overview of the technical solutions of the embodiments of the present invention. In order to better understand the technical means of the embodiments of the present invention and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description
[0046] Figure 1 A schematic diagram of a power distribution network fault location method based on dual-end positioning provided in an embodiment of the present invention;
[0047] Figure 2 A 10kV single-ended radial distribution network simulation model was developed to verify embodiments of the present invention.
[0048] Figure 3 This is a structural diagram of a power distribution network fault location device based on dual-end positioning, provided in an embodiment of the present invention. Detailed Implementation
[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0050] Example 1:
[0051] like Figure 1 As shown in the figure, a method for fault location in a distribution network based on dual-end positioning provided by an embodiment of the present invention includes the following steps:
[0052] S11, obtain the waveform recording files of the dual-end positioning devices at both ends of the faulty line, and extract the start timestamps of the two waveform recording files respectively to obtain the relative zero time when fault waveform recording is started at both ends of the faulty line.
[0053] S12, align the relative zero time, and extract the first voltage traveling wave data of both ends of the fault line within a preset time period before and after the relative zero time, so as to obtain the first voltage traveling wave line mode components of both ends of the fault line according to the first voltage traveling wave data.
[0054] S13, perform wavelet transform on the first voltage traveling wave line mode component respectively to obtain the first time domain waveform diagram of different layers, and obtain the first wavefront time sequence of both ends of the fault line according to the first time domain waveform diagram.
[0055] S14. Based on the time series of the first wavefronts at both ends of the faulty line, obtain the first overlapping wavefront and the second overlapping wavefront, and calculate the relationship feature value between the first overlapping wavefront and the second overlapping wavefront.
[0056] S15, input the relationship feature value and the upstream and downstream time difference of the fault point into the preset neural network model, output the distance of the fault point, and perform distribution network fault location based on the distance of the fault point; wherein, the upstream and downstream time difference of the fault point is obtained based on the waveform file.
[0057] In the actual implementation, the time when the two ends of the traveling wave emitted from the fault point arrive at the monitoring point is defined as time 0. However, after the actual fault occurs, the fault recording time of the dual-end positioning device is also inconsistent because the fault current at both ends of the fault is inconsistent. It is necessary to combine the start time stamp of the recording file to align the relative zero time at both ends.
[0058] Preferably, the step of extracting the first voltage traveling wave data at both ends of the fault line within a preset time period before and after the relative zero time is, in this embodiment, specifically: extracting a voltage traveling wave data window within a 5ms time period near the fault zero point.
[0059] In this embodiment, obtaining the first voltage traveling wave line mode components at both ends of the fault line based on the first voltage traveling wave data specifically involves performing a Karenbauer transform on the first voltage traveling wave data to obtain the first voltage traveling wave line mode components corresponding to each end of the fault line.
[0060] In this embodiment, the step of performing wavelet transform on the first voltage traveling wave line mode components to obtain time-domain waveforms at different levels specifically involves performing wavelet transform on the first voltage traveling wave line mode components to obtain high-frequency coefficients for their respective different frequency bands, and reconstructing the corresponding first voltage traveling wave line mode components based on the high-frequency coefficients to obtain time-domain waveforms at different levels at both ends of the fault line.
[0061] In the specific implementation, the collected voltage traveling wave line mode components are decomposed into high-frequency coefficients of different frequency bands through Haar wavelet transform in 8 layers, and then reconstructed to obtain time-domain waveforms of different layers.
[0062] In this embodiment, the step of obtaining the first wavefront time series of both ends of the faulty line based on the time-domain waveform diagram is as follows: take the maximum modulus value of each layer in the time-domain waveform diagram, and set the modulus values less than the maximum modulus value by a preset percentage to 0; extract the points with non-zero modulus values and their corresponding time coordinates to construct the first wavefront time series of both ends of the faulty line.
[0063] In a specific implementation, in order to reduce the communication volume of the dual-end positioning device, the maximum value of the waveform of each layer is calculated, and the data below 10% of the maximum value is set to 0. Then, the non-zero points and their corresponding time coordinates are transmitted to the other end as a two-dimensional matrix sequence; wherein, the two-dimensional matrix sequence is the first wave head time sequence.
[0064] It should be noted that the core of fault location lies in finding the fault wavefront. The magnitude of the wavefront amplitude at both ends of the fault is affected by the fault angle, fault resistance, and branch nodes. Sometimes it is impossible to find an obvious wavefront in the first layer of wavelet transform, so it is necessary to continue searching in the lower frequency band. Secondly, the noise of the waveform recording device is relatively large. In order to prevent wavefront misjudgment, anything below 10% of the modulus maxima in each layer is considered noise. These two processing measures can ensure accurate location of the fault wavefront.
[0065] In this embodiment, obtaining the first overlapping wavefront and the second overlapping wavefront based on the time sequence of the first wavefront at both ends of the faulty line specifically involves: extracting the first wavefront from the time sequence of the first wavefront at both ends of the faulty line, aligning the first wavefront by time shifting to obtain the first overlapping wavefront; and continuing to search forward according to the time coordinate to extract the wavefront that is aligned again to obtain the second overlapping wavefront.
[0066] In the specific implementation, the first wavefront time series of the line-mode voltages at both ends of the fault section, after being reconstructed by high-frequency coefficients, has an amplitude near 0 before the fault occurs. After the fault occurs, an amplitude greater than 0 appears, and these are called wavefronts. The first wavefronts that appear in the sequences at both ends (i.e., the initial arrival time of the fault wavefront) are aligned by time shifting, and this is called the first overlapping wavefront. After the time series at both ends are aligned to obtain the first overlapping wavefront, the subsequent wavefronts will automatically appear at the time of realignment. The aligned wavefront that appears after the first overlapping wavefront is called the second overlapping wavefront.
[0067] In this embodiment, the relational feature values between the first overlapping wavefront and the second overlapping wavefront include the modulus ratio between the first wavefronts at both ends of the faulty line, the modulus ratio between the realigned wavefronts, and the time difference between the time coordinates corresponding to the first overlapping wavefront and the time coordinates corresponding to the second overlapping wavefront.
[0068] It should be noted that the modulus ratio between the first wavefronts at both ends of the faulty line and the modulus ratio between the re-aligned wavefronts are related to the distance from the fault point to both ends, reflecting the attenuation relationship between the wavefront amplitude and the distance; the time difference between the time coordinates corresponding to the first overlapping wavefront and the time coordinates corresponding to the second overlapping wavefront is related to the distance from the fault point to the near measurement end and the wave velocity. The above-mentioned characteristic values can reflect the fault distance information from different perspectives.
[0069] In this embodiment, obtaining the upstream and downstream time difference of the fault point based on the waveform recording file specifically involves: acquiring the initial waveform recording time of each waveform recording file, and aligning the waveform data in each waveform recording file to absolute zero time based on the initial waveform recording time; extracting the second voltage traveling wave data at both ends of the fault line within a preset time period before and after the absolute zero time, so as to obtain the second voltage traveling wave line mode components at both ends of the fault line based on the second voltage traveling wave data; performing wavelet transform on the second voltage traveling wave line mode components to obtain second time domain waveform diagrams at different levels, and obtaining the second wavefront time series at both ends of the fault line based on the second time domain waveform diagrams; finding the initial wavefront of the second wavefront time series, and calculating the time from the time corresponding to the initial wavefront to the respective initial waveform recording time to obtain the upstream initial wavefront time and the downstream initial wavefront time; taking the absolute value of the difference between the upstream initial wavefront time and the downstream initial wavefront time, and using the absolute value as the upstream and downstream time difference of the fault point.
[0070] It should be noted that the upstream and downstream time difference of the fault point is related to the distance from the fault point to both ends and the wave speed. By replacing the wave speed parameter with the upstream and downstream time difference of the fault point, the problem of wave speed in the traditional two-end ranging method, which is affected by line parameters (such as conductor type, insulation medium, and aging degree), is avoided, which leads to wave speed instability and ultimately makes the fault location result calculated based on wave speed inaccurate.
[0071] In the specific implementation, the relationship feature value and the time difference between the upstream and downstream of the fault point are finally input into the preset neural network model to obtain the distance to the fault point.
[0072] Preferably, in this embodiment of the invention, a PSO-BP neural network model (a backpropagation neural network model optimized by particle swarm optimization) is selected as the neural network model for outputting fault distance.
[0073] Preferably, the construction process of the PSO-BP neural network model is as follows: A localization model of the PSO-BP neural network is constructed; the weights and bias parameters of the particle swarm optimization algorithm in the PSO-BP model are initialized; the maximum number of iterations is set to 50, and the learning rate is set to 0.99; the prepared training set is input into the PSO-BP neural network model; the localization effect of the proposed method is verified using a test set of the PSO-BP neural network fault localization model. If the localization accuracy does not reach the expected threshold, the parameter settings are modified and the training is repeated until the localization accuracy meets the requirements; wherein the preset threshold is 100m.
[0074] Preferably, the process of obtaining the prepared training and test sets is as follows: A fault location model is built on simulation software based on the actual power distribution network system; fault distance, fault type, fault location, and different faulty lines are used as variables for simulation to obtain fault waveform data; the calculated data is labeled with the actual fault distance as the corresponding sample; the fault voltage waveform within 5ms before and after the fault occurs is extracted using a sampling frequency of 1MHz, and the extracted voltage traveling wave data is subjected to Karlenbauer transform to obtain its line-mode voltage traveling wave component, forming a sample set of easily updated and reused relational feature values and the upstream and downstream time difference of the fault point. Finally, the sample set is divided into a training set and a test set in an 8:3 ratio.
[0075] This invention aligns fault waveform data at both ends of the faulty line relative to time zero, providing data preparation for subsequent acquisition of the first voltage traveling wave mode components at both ends, and avoiding the problem of strict time synchronization at both ends required in the traditional double-ended traveling wave method. By obtaining the first wavefront time series at both ends of the faulty line based on the first voltage traveling wave mode components, it provides data preparation for subsequent acquisition of the first and second overlapping wavefronts. By calculating the relationship feature values between the first and second overlapping wavefronts, it provides data support for subsequent fault location using a preset neural network model, and by using these relationship feature values as input values to the neural network model, it avoids the problem of traditional... The double-ended traveling wave method requires wave velocity normalization. The neural network model obtains the distance to the fault point based on the relationship feature values and the time difference between the upstream and downstream sides of the fault point. Traditional double-ended traveling wave methods require strict time synchronization and wave velocity normalization. However, the traveling wave propagation speed is affected by the actual physical parameters of the line (such as conductor type, insulation medium, and aging degree), resulting in wave velocity instability. This makes it difficult to fully achieve time synchronization and wave velocity normalization, leading to a large error in the final fault location result. This invention avoids the problem of requiring time synchronization and wave velocity normalization by using the sum of the relationship features and the time difference between the upstream and downstream sides of the fault point as the input value of the neural network model. Therefore, compared with existing technologies, this invention can improve the accuracy of fault location in distribution networks.
[0076] It should be noted that, in order to demonstrate the beneficial effects of the embodiments of the present invention, some simulation experimental data will be provided as a reference. The specific simulation verification process is as follows:
[0077] like Figure 2 The image shows a simulation model of a 10kV single-ended radial distribution network used to verify embodiments of the present invention.
[0078] In the specific implementation, the physical parameters set in the power distribution network simulation model include: bus voltage level of 10kV, fault occurrence time of 0.098s; sampling frequency of 1MHz; line parameters of positive sequence impedance: 0.482 (Ω / km); positive sequence reactance: 0.846 (Ω / km); positive sequence ground admittance: 1.24e-5 (S / km); zero sequence impedance: 0.923 (Ω / km); zero sequence reactance: 4.74 (Ω / km); zero sequence ground admittance: 5.03e-6 (S / km); and system simulation duration of 0.1s.
[0079] Furthermore, a PSO-BP artificial neural network was constructed. Wavelet transform was used to extract the modulus maxima ratios of the first overlapping wavefront (FCW) and the second overlapping wavefront (SCW) of the voltage (i.e., the modulus ratio between the first wavefronts at both ends of the faulty line and the modulus ratio between the re-aligned wavefronts), k1 and k2, respectively. The time difference t1 (the time difference between the corresponding time coordinates of the first and second overlapping wavefronts) and the time difference t2 (the time difference between the upstream and downstream ends of the fault point) of the first coincident wavefronts at both ends of the fault point were also extracted. These constituted the input feature quantities [k1, k2, t1, t2] of the PSO artificial neural network. Eleven fault points were randomly set from 1 km away from the fault location device on the bus side to the end of the line area. The fault type was a phase-to-phase grounding fault; the fault resistance was 10Ω.
[0080] To demonstrate the applicability of the constructed artificial neural network fault location model based on particle swarm optimization algorithm, training and testing were conducted under conditions of different fault initial phase angles, different transition resistances, and different fault locations. Figure 2 A single-phase ground fault was set on line L6 as shown, with a transition resistance of 10Ω and initial phase angles of 10°, 30°, 45°, 70°, and 90°. The location results are shown in Table 1. The proposed method can achieve accurate location under different initial phase angles of the fault, and the impact of different initial phase angles on fault location is small.
[0081] Table 1. Initial phase angle location results for phase-to-phase faults.
[0082]
[0083] exist Figure 2A single-phase ground fault occurred on feeder L6 at a distance of 4.6 km from point A. The initial phase angle of the fault was 30°, and the fault transition resistances were 0.01Ω, 1Ω, 10Ω, 100Ω, 300Ω, and 1000Ω, respectively. Simulation results show that the proposed method has high positioning accuracy under different transition resistances and is not affected by the transition resistance, as shown in Table 2 below.
[0084] Table 2. Location results of different fault resistors during single-phase faults.
[0085]
[0086] exist Figure 1 The training tests were conducted for different fault types, different fault sections, different fault feeder lengths, and different fault lines. The fault resistance was 10Ω, the initial phase angle of the fault was 30°, and the fault location results for single-phase grounding faults and phase-to-phase faults at different fault locations are shown in Tables 3 and 4, respectively.
[0087] Table 3. Location results of different fault locations during single-phase ground faults.
[0088]
[0089] Table 3 shows that the fault location error of this method is within 150m under different fault distances, different feeder lengths, and different fault lines. Simulation results indicate that for single-phase grounding faults, different fault locations have little impact on the proposed fault location method.
[0090] Table 4. Location results of different fault locations during phase-to-phase faults
[0091]
[0092] According to the data in Table 4, for phase-to-phase faults, the location error is mostly below 150m under different fault lines, different fault sections and different fault distances, and the location results are quite satisfactory.
[0093] Example 2:
[0094] like Figure 3 As shown, this embodiment provides a power distribution network fault location device based on dual-end positioning, including a relative zero-time acquisition module 001, a first voltage traveling wave line mode component module 002, a first wavefront time series module 003, a relational feature value module 004, and a fault point distance acquisition module 005, wherein,
[0095] The relative zero time acquisition module 001 is used to acquire the waveform recording files of the dual-end positioning devices at both ends of the faulty line, and extract the start timestamps of the two waveform recording files respectively, so as to obtain the relative zero time when fault waveform recording is started at both ends of the faulty line.
[0096] The first voltage traveling wave line mode component module 002 is used to align the relative zero time and extract the first voltage traveling wave data at both ends of the fault line within a preset time period before and after the relative zero time, so as to obtain the first voltage traveling wave line mode components at both ends of the fault line according to the first voltage traveling wave data.
[0097] The first wavefront time sequence module 003 is used to perform wavelet transform on the first voltage traveling wave line mode component to obtain the first time domain waveform diagrams of different layers, and to obtain the first wavefront time sequence of both ends of the fault line according to the first time domain waveform diagrams.
[0098] The relational feature value module 004 is used to obtain the first overlapping wavefront and the second overlapping wavefront based on the first wavefront time series at both ends of the faulty line, and to calculate the relational feature value between the first overlapping wavefront and the second overlapping wavefront.
[0099] The fault point distance acquisition module 005 is used to input the relationship feature value and the upstream and downstream time difference of the fault point into a preset neural network model, output the fault point distance, and perform distribution network fault location based on the fault point distance; wherein, the upstream and downstream time difference of the fault point is obtained based on the waveform file.
[0100] In this embodiment, the first voltage traveling wave line mode component module 002 obtains the first voltage traveling wave line mode components at both ends of the fault line according to the first voltage traveling wave data. Specifically, the first voltage traveling wave line mode component module 002 performs Karenbauer transform on the first voltage traveling wave data to obtain the first voltage traveling wave line mode components corresponding to each end of the fault line.
[0101] In this embodiment, the first wavefront time series module 003 performs wavelet transform on the first voltage traveling wave line mode component to obtain time-domain waveforms at different levels. Specifically, the first wavefront time series module 003 performs wavelet transform on the first voltage traveling wave line mode component to obtain high-frequency coefficients of different frequency bands, and reconstructs the corresponding first voltage traveling wave line mode component based on the high-frequency coefficients to obtain time-domain waveforms at different levels at both ends of the fault line.
[0102] Furthermore, the first wavefront time series module 003 obtains the first wavefront time series of both ends of the faulty line according to the time domain waveform diagram. Specifically, the first wavefront time series module 003 takes the maximum modulus value of each layer in the time domain waveform diagram and sets the modulus values less than the maximum modulus value to 0 by a preset percentage. It extracts the points with non-zero modulus values and their corresponding time coordinates to construct the first wavefront time series of both ends of the faulty line.
[0103] In this embodiment, the relation feature value module 004 obtains the first overlapping wavefront and the second overlapping wavefront based on the time sequence of the first wavefront at both ends of the faulty line. Specifically, the relation feature value module 004 extracts the first wavefront from the time sequence of the first wavefront at both ends of the faulty line, aligns the first wavefront by time shifting, and obtains the first overlapping wavefront; it continues to search forward according to the time coordinate and extracts the wavefront that is aligned again to obtain the second overlapping wavefront.
[0104] For a more detailed explanation of the working principle and procedures of this embodiment, please refer to the relevant description in Embodiment 1.
[0105] This invention employs a relative zero-time acquisition module 001 to acquire the relative zero-time of fault waveform data in the waveform recording file, providing data support for subsequent relative zero-time alignment; a first voltage traveling wave mode component module 002 to acquire the first voltage traveling wave mode components at both ends of the fault line based on the first voltage traveling wave data, providing data support for subsequent finding of overlapping wavefronts; a first wavefront time series module 003 to acquire the first wavefront time series at both ends of the fault line, providing data preparation for subsequent calculation of the relationship feature values; and a relationship feature value module 004 to calculate the relationship feature values between the first overlapping wavefront and the second overlapping wavefront, providing rich and comprehensive input values for the subsequent neural network model, thereby improving the model's performance. The accuracy of the output results is improved. The fault point distance acquisition module 005 inputs the relationship feature value and the upstream and downstream time difference of the fault point into a preset neural network model for distribution network fault location. In the traditional double-ended traveling wave method, strict time synchronization and wave velocity normalization are required to obtain wave velocity parameters and calculate the time difference. However, due to the influence of uncontrollable external factors such as line aging in actual transmission lines, it is difficult to accurately obtain wave velocity parameters, thus affecting the accuracy of the final fault location result. This invention uses the relationship feature value obtained through relative zero-time alignment and the upstream and downstream time difference of the fault point as model input data, thereby avoiding the problem of accurately obtaining wave velocity parameters in existing technologies and improving the accuracy of distribution network fault location.
[0106] Example 3:
[0107] This embodiment provides a terminal device, including: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus;
[0108] The memory is used to store at least one executable instruction that causes the processor to perform the operation of the power distribution network fault location method based on dual-end positioning as described above.
[0109] Example 4:
[0110] This invention provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device / apparatus containing the computer-readable storage medium to perform the power distribution network fault location method based on dual-end positioning as described above.
[0111] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.
[0112] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.
Claims
1. A method for fault location in a distribution network based on dual-end positioning, characterized in that, include: Obtain the waveform recording files of the dual-end positioning devices at both ends of the faulty line, and extract the start timestamps of the two waveform recording files respectively to obtain the relative zero time when fault waveform recording is started at both ends of the faulty line. Align the relative zero time and extract the first voltage traveling wave data of both ends of the fault line within a preset time period before and after the relative zero time, so as to obtain the first voltage traveling wave line mode components of both ends of the fault line based on the first voltage traveling wave data. Wavelet transform is performed on the first voltage traveling wave line mode component to obtain the first time domain waveform diagrams of different layers, and the first wavefront time series of both ends of the fault line are obtained according to the first time domain waveform diagrams. Specifically, obtaining the first wavefront time series of both ends of the faulty line based on the time-domain waveform diagram involves: taking the maximum modulus value of each layer in the time-domain waveform diagram and setting the modulus values less than the maximum modulus value to 0 by a preset percentage; extracting the points with non-zero modulus values and their corresponding time coordinates to construct the first wavefront time series of both ends of the faulty line. Based on the time series of the first wavefronts at both ends of the faulty line, the first overlapping wavefront and the second overlapping wavefront are obtained, and the relationship feature value between the first overlapping wavefront and the second overlapping wavefront is calculated. The relationship feature values between the first overlapping wavefront and the second overlapping wavefront include the modulus ratio between the first wavefronts at both ends of the faulty line, the modulus ratio between the wavefronts that are aligned again, and the time difference between the time coordinates corresponding to the first overlapping wavefront and the time coordinates corresponding to the second overlapping wavefront. The relationship feature value and the upstream and downstream time difference of the fault point are input into a preset neural network model, the fault point distance is output, and the fault location of the distribution network is performed based on the fault point distance; wherein, the upstream and downstream time difference of the fault point is obtained based on the waveform file.
2. The distribution network fault location method based on dual-end positioning as described in claim 1, characterized in that, The step of obtaining the first voltage traveling wave line mode components at both ends of the faulty line based on the first voltage traveling wave data is specifically as follows: The first voltage traveling wave data is subjected to Karlenbauer transform to obtain the first voltage traveling wave line mode components corresponding to both ends of the faulty line.
3. The distribution network fault location method based on dual-end positioning as described in claim 1, characterized in that, The step of performing wavelet transform on the traveling wave mode components of the first voltage to obtain time-domain waveforms at different levels is specifically as follows: Wavelet transform is performed on the first voltage traveling wave line mode components to obtain the high-frequency coefficients of their respective frequency bands. The corresponding first voltage traveling wave line mode components are then reconstructed based on the high-frequency coefficients to obtain time-domain waveform diagrams of different layers at both ends of the faulted line.
4. The distribution network fault location method based on dual-end positioning as described in claim 1, characterized in that, The step of obtaining the first overlapping wavefront and the second overlapping wavefront based on the time sequence of the first wavefront at both ends of the faulty line specifically involves: The first wavefront in the time series of the first wavefront at both ends of the faulty line is extracted respectively, and the first wavefront is aligned by time shifting to obtain the first overlapping wavefront; Continuing to search along the time coordinate, extract the wavefronts that are aligned again to obtain the second overlapping wavefront.
5. The distribution network fault location method based on dual-end positioning as described in claim 1, characterized in that, The step of obtaining the upstream and downstream time difference of the fault point based on the recorded waveform file is specifically as follows: The initial recording time of each recording file is obtained, and the recording data in each recording file is aligned to absolute zero time according to the initial recording time. Extract the second voltage traveling wave data of both ends of the fault line within a preset time period before and after the absolute zero moment, so as to obtain the second voltage traveling wave line mode components of both ends of the fault line based on the second voltage traveling wave data. Wavelet transform is performed on the second voltage traveling wave line mode component to obtain the second time domain waveform diagrams at different levels, and the second wavefront time series at both ends of the fault line are obtained based on the second time domain waveform diagrams. Find the initial wavefront of the second wavefront time series respectively, and calculate the time from the time corresponding to the initial wavefront to the respective initial recording time to obtain the upstream initial wavefront time and the downstream initial wavefront time; Take the absolute value of the difference between the upstream initial wavefront time and the downstream initial wavefront time, and use the absolute value as the upstream and downstream time difference of the fault point.
6. A distribution network fault location device based on dual-end positioning, characterized in that, It includes a relative zero-time acquisition module, a first voltage traveling wave mode component module, a first wavefront time series module, a relational feature value module, and a fault point distance acquisition module, among which, The relative zero-time acquisition module is used to acquire the waveform recording files of the dual-end positioning devices at both ends of the faulty line, and extract the start timestamps of the two waveform recording files respectively to obtain the relative zero-time when fault waveform recording is started at both ends of the faulty line. The first voltage traveling wave line mode component module is used to align the relative zero time and extract the first voltage traveling wave data at both ends of the fault line within a preset time period before and after the relative zero time, so as to obtain the first voltage traveling wave line mode components at both ends of the fault line according to the first voltage traveling wave data. The first wavefront time series module is used to perform wavelet transform on the first voltage traveling wave line mode component to obtain the first time domain waveform diagrams of different layers, and to obtain the first wavefront time series at both ends of the fault line based on the first time domain waveform diagrams. Specifically, obtaining the first wavefront time series of both ends of the faulty line based on the time-domain waveform diagram involves: taking the maximum modulus value of each layer in the time-domain waveform diagram and setting the modulus values less than the maximum modulus value to 0 by a preset percentage; extracting the points with non-zero modulus values and their corresponding time coordinates to construct the first wavefront time series of both ends of the faulty line. The relational feature value module is used to obtain the first overlapping wavefront and the second overlapping wavefront based on the first wavefront time series at both ends of the faulty line, and to calculate the relational feature value between the first overlapping wavefront and the second overlapping wavefront. The relationship feature values between the first overlapping wavefront and the second overlapping wavefront include the modulus ratio between the first wavefronts at both ends of the faulty line, the modulus ratio between the wavefronts that are aligned again, and the time difference between the time coordinates corresponding to the first overlapping wavefront and the time coordinates corresponding to the second overlapping wavefront. The fault point distance acquisition module is used to input the relationship feature value and the upstream and downstream time difference of the fault point into a preset neural network model, output the fault point distance, and perform distribution network fault location based on the fault point distance; wherein, the upstream and downstream time difference of the fault point is obtained based on the waveform file.
7. A terminal device, characterized in that, include: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction that causes the processor to perform the operation of the power distribution network fault location method based on dual-end positioning as described in any one of claims 1 to 5.
8. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device / apparatus containing the computer-readable storage medium to perform the power distribution network fault location method based on dual-end positioning as described in any one of claims 1 to 5.