An overhead line single-end fault traveling wave positioning method and device

Abstract

The application relates to the technical field of line fault traveling wave positioning, in particular to an overhead line single-end fault traveling wave positioning method and equipment, which comprises the following steps: reconstructing a pure traveling wave signal phase space, constructing a Hankel matrix, performing singular value decomposition on the matrix, selecting effective singular values for signal reconstruction, and adopting a Teager energy operator to perform nonlinear enhancement on the denoised signal; adopting an improved AIC criterion to perform wave head detection, calculating a waveform correlation coefficient to determine whether it is a fault point reflection wave; adaptively estimating a traveling wave velocity, combining a time difference between an initial traveling wave and a fault point reflection wave to obtain a distance from the fault point to a measurement end, so as to obtain a fault positioning result. The application can improve line fault detection and positioning precision.

Description

Technical Field

[0001] This application relates to the field of traveling wave location technology for line faults, specifically to a method and device for locating traveling wave faults at one end of an overhead line. Background Technology

[0002] Overhead transmission lines are a crucial component of the power system, undertaking the vital task of transmitting electrical energy. Because overhead lines are exposed to the elements for extended periods, operating in complex environments, they are susceptible to faults caused by lightning strikes, pollution, icing, and external damage. Quickly and accurately locating fault points is of paramount importance for minimizing power outage time, improving power supply reliability, and reducing economic losses.

[0003] Traveling wave (TW) method is one of the most widely used fault location methods. Depending on the number of measuring ends, TW location is divided into two-end and single-end methods. The two-end method requires simultaneous measurements at both ends of the line, synchronized with GPS, and offers high positioning accuracy. However, it requires measuring equipment at both ends, resulting in high cost, and it relies on communication synchronization. The single-end method only requires measuring equipment to be installed at one end of the line, offering lower cost and easier deployment, thus gaining wider attention and application.

[0004] However, single-ended traveling wave positioning has several drawbacks in practical applications, including low signal-to-noise ratio of traveling wave signals, susceptibility to noise interference, low wavefront detection accuracy, difficulty in identifying reflected waves from fault points, uncertainty in wave velocity affecting positioning accuracy, and insufficient reliability of positioning results. Summary of the Invention

[0005] To address the aforementioned technical problems, the purpose of this application is to provide a method and device for locating traveling waves in single-end faults of overhead lines. The specific technical solution adopted is as follows: This application provides a method for locating traveling wave faults at one end of an overhead line, including the following steps: The transient voltage and transient current traveling wave signals after the fault are preprocessed by outlier removal, baseline correction and power frequency component filtering to obtain a clean traveling wave signal; A Hankel matrix is ​​constructed by reconstructing the phase space of a clean traveling wave signal. The matrix is ​​then subjected to singular value decomposition. Based on the distribution characteristics of the singular values, effective singular values ​​are selected for signal reconstruction. Finally, the Teager energy operator is used to nonlinearly enhance the denoised signal. The enhanced traveling wave signal is subjected to wavefront detection using an improved AIC criterion, wherein a sliding window variable step size search strategy is introduced into the improved AIC criterion. Based on the waveform characteristics of the initial traveling wave and using the waveform correlation coefficient, it is determined whether it is a fault point reflected wave, so as to determine the arrival time of the fault point reflected wave. Based on the actual structural and operational parameters of the overhead line, the traveling wave velocity is adaptively estimated. Then, by combining the time difference between the initial traveling wave and the reflected wave at the fault point, the distance from the fault point to the measurement end is obtained to obtain the fault location result.

[0006] Preferably, the preprocessing specifically includes: The Dixon test is used to identify outliers in the sampling. The data are sorted by size and the test statistic is calculated. The statistic is compared with the critical value to determine whether it is an outlier. For the identified outliers, linear interpolation of adjacent points is used to replace them. Polynomial fitting baseline correction method is used to eliminate DC offset and low-frequency drift in the signal; An adaptive notch filter is used to remove the power frequency and its harmonic components, while retaining the high-frequency components of the transient traveling wave, thus obtaining a pure traveling wave signal.

[0007] Preferably, singular value decomposition is used for traveling wave signal denoising, specifically including: The signals are arranged into a matrix according to the set embedding dimension and time delay to obtain the Hankel matrix. Singular value decomposition is performed on the Hankel matrix, and the difference spectrum of the singular values ​​is calculated. The order corresponding to the first maximum point in the difference spectrum is used as the boundary between effective singular values ​​and noisy singular values. Select the aforementioned number of valid singular values, reconstruct the corresponding singular vectors to obtain the denoised matrix, and then obtain the denoised traveling wave signal by the diagonal averaging method.

[0008] Preferably, the nonlinear enhancement of the denoised signal using the Teager energy operator specifically includes: The Teager energy value is calculated point by point for the denoised traveling wave signal, and the sequence of all calculated Teager energies is smoothed; the smoothed Teager energy sequence is normalized to obtain the enhanced traveling wave signal.

[0009] Preferably, the improved AIC criterion for wavehead detection, which introduces a sliding window variable step size search strategy, specifically includes: Set the sliding window length for AIC calculation, and slide the window across the entire signal range with a preset step size to calculate the AIC value point by point. Use the area corresponding to the AIC value being less than the preset threshold to coarsely locate the wavefront. Within the area obtained by coarse localization, the AIC value is calculated point by point using a sliding window with sampling points as the step size, and the minimum AIC point is found. The minimum AIC point and its three adjacent points with preceding and following AIC values ​​are fitted, and the extreme point of the fitted curve is taken as the arrival time of the wavefront.

[0010] Preferably, the process for determining the reflected wave at the fault point specifically includes: Waveform segments of a set length before and after the arrival time of the initial traveling wave are extracted as reference waveform templates. In the signal after the initial traveling wave, waveform segments are truncate according to a set time step. The correlation coefficient of each segment is calculated with the reference waveform template, and the peak point of all correlation coefficients is extracted. The corresponding time is used as the arrival time of the candidate reflected wave. The theoretical arrival time ranges for the reflected waves from the fault point and the reflected waves from the opposite end are set. If the arrival time of a candidate reflected wave falls within the theoretical arrival time range of the reflected wave from the fault point, it is determined to be a reflected wave from the fault point.

[0011] Preferably, the adaptive estimation of traveling wave velocity based on the actual structural and operational parameters of the overhead line specifically includes: Obtain the structural parameters of the overhead line, including conductor type, conductor radius, split spacing, tower height, phase-to-phase distance, and ground resistivity; based on the line structural parameters, use a distributed parameter model to calculate the resistance, inductance, capacitance, and conductance per unit length of the line; Based on the frequency range of the traveling wave, the wave impedance and propagation coefficient at the corresponding frequency are calculated, and then the propagation speed of the traveling wave is calculated as the wave velocity of the traveling wave.

[0012] Preferably, after calculating the fault distance, a multi-scale consistency verification step is also included, specifically including: The pure traveling wave signal is decomposed into multiple scales to obtain traveling wave signals at different scales; wavefront detection and fault distance calculation are performed at each scale to obtain positioning results at multiple scales. Calculate the dispersion between positioning results at each scale. If the dispersion is less than a set threshold, the positioning result is reliable. If the dispersion is greater than the threshold, remove the abnormal scales and recalculate the average positioning result. Output the final fault distance and location confidence that have been verified for consistency.

[0013] Preferably, it also includes a positioning error correction step, specifically including: The entire line is divided into several segments according to tower sections, and the wave velocity of the traveling wave corresponding to each segment is calculated separately to locate faults based on the segmentation. An error correction database is established based on the historical fault location results and the actual fault location. For new fault location results, the corresponding correction amount is retrieved from the error correction database based on the fault location to compensate for the error in the location results.

[0014] This application embodiment also provides an overhead line single-end fault traveling wave location device, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the above-described overhead line single-end fault traveling wave location methods.

[0015] As can be seen from the above, the traveling wave location method and equipment for single-end faults in overhead lines provided in this application have at least the following beneficial effects: This application presents a complete single-end traveling wave positioning solution from raw signal to positioning result. Unlike traditional methods that only improve a single step, this application features an overall system architecture design with each module working together to achieve high accuracy and high reliability in single-end positioning.

[0016] This application proposes a traveling wave signal processing framework combining singular value decomposition (SVD) and the Teager energy operator. First, SVD is used to denoise the signal, and then the Teager energy operator enhances the wavefront features. Singular value decomposition is a matrix-based denoising method that does not require pre-selection of basis functions, has strong adaptability, and can effectively suppress random noise. The Teager energy operator is a nonlinear energy calculation method that can significantly enhance the abrupt changes in the signal, making the traveling wave wavefront more prominent. The combination of these two methods improves the signal-to-noise ratio and highlights the features, laying a solid foundation for subsequent accurate wavefront detection.

[0017] This application employs an improved AIC criterion for traveling wave front detection. Based on information theory, the AIC criterion automatically detects changes in signal statistical characteristics, offering high detection accuracy and strong anti-interference capabilities. In particular, a variable step-size search strategy is introduced: first, a large step size is used for coarse positioning to narrow the range, then a small step size is used for fine positioning to accurately locate the point, significantly improving computational efficiency while maintaining accuracy. Furthermore, sub-sampling level detection accuracy can be achieved through quadratic polynomial fitting, an order of magnitude improvement over traditional sampling level accuracy, greatly enhancing positioning precision.

[0018] This application proposes a reflected wave identification method based on waveform correlation analysis. Using the initial traveling wave as a reference template, it automatically searches for similar reflected waves in subsequent signals by calculating the correlation coefficient, thus accurately identifying the arrival time of the reflected wave at the fault point. Simultaneously, by combining the physical laws of traveling wave propagation and information on the total line length, it can effectively distinguish between reflected waves from those at the fault point and those from the opposite substation, solving the problem of reflected wave confusion in single-end positioning and significantly improving the accuracy and reliability of single-end positioning.

[0019] This application proposes an adaptive wave velocity estimation method based on line parameters. Instead of using a fixed wave velocity, it calculates the actual propagation velocity of the traveling wave online according to the actual structural and operational parameters of the overhead line. By considering the influence of various factors such as conductor parameters, tower structure, and ground resistivity on the wave velocity, it obtains a wave velocity value that more closely reflects the actual situation, fundamentally solving the positioning error problem caused by wave velocity uncertainty and significantly improving positioning accuracy.

[0020] This application introduces a multi-scale consistency verification mechanism, performing positioning calculations at multiple time scales and then comparing the consistency of the results at each scale. If the results at all scales are consistent, the positioning result is reliable; if a result at a certain scale is abnormal, it is discarded. This multi-scale cross-validation mechanism can effectively detect and eliminate abnormal results, greatly improving the reliability and fault tolerance of the positioning results and avoiding positioning failures caused by interference at a single scale. Attached Figure Description

[0021] Figure 1 A flowchart illustrating the steps of a traveling wave location method for a single-end fault in an overhead line provided in this application. Detailed Implementation

[0022] The following description, in conjunction with the accompanying drawings, details the specific scheme of the traveling wave location method and equipment for single-end faults in overhead lines provided in this application.

[0023] Please see Figure 1 The diagram illustrates a flowchart of a traveling wave localization method for a single-end fault in an overhead line according to an embodiment of this application, including the following steps: Step 1: Perform outlier removal, baseline correction, and power frequency component filtering preprocessing on the transient voltage and current traveling wave signals after the fault occurs to obtain a clean traveling wave signal.

[0024] This embodiment uses a 220kV overhead transmission line as an example to describe in detail the single-end fault traveling wave location method. The line is approximately 120 kilometers long, with double-split conductors of type LGJ-400 / 35. The average tower height is approximately 30 meters, and the phase-to-phase distance is approximately 4.5 meters. The terrain along the line is complex, including mountains, hills, and plains, with uneven soil resistivity distribution. This line experiences frequent lightning strikes, and the traditional single-end traveling wave location method has a large error, often resulting in inaccurate location. Therefore, there is an urgent need to improve the location accuracy.

[0025] First, a traveling wave acquisition device is installed at one end of the substation. In this embodiment, the sampling frequency is 1MHz, meaning one point is sampled every microsecond, which meets the high-frequency sampling requirements of traveling wave signals. The acquired signals include transient waveforms of three-phase voltage and three-phase current. The triggering method is abrupt triggering; acquisition is automatically triggered when the signal change exceeds a set threshold, recording waveform data before and after the fault. The acquisition time window is from 2 milliseconds before the fault to 10 milliseconds after the fault, ensuring complete recording of the initial traveling wave and each reflected wave.

[0026] The Dixon test was used to identify outliers in the original sampled data. The Dixon test is an outlier detection method based on the range ratio, particularly suitable for small sample data. The specific procedure is as follows: sort the data in ascending order, calculate the difference between the maximum or minimum value and its adjacent values, and then divide this difference by the data range to obtain the test statistic; look up the Dixon critical value table based on the sample size and significance level; if the test statistic is greater than the critical value, it is determined to be an outlier.

[0027] In this embodiment, the significance level is set to 0.05. Upon inspection, a small number of outliers caused by sampling glitches and electromagnetic interference were found in the original data. All of these were identified and removed. The outlier locations were then filled using linear interpolation between two adjacent normal sampling points.

[0028] Due to the characteristics of the current transformer and the influence of the sampling circuit, the acquired traveling wave signal may exhibit DC offset and low-frequency drift. A polynomial fitting baseline correction method is used to address this: a third-order polynomial is used to fit the baseline of the signal, and then the fitted baseline is subtracted from the original signal to obtain a signal with a stable baseline. This method is computationally simple and can effectively eliminate DC offset and slow baseline drift.

[0029] Traveling wave signals are mainly high-frequency transient components, while the amplitude of the power frequency component is very large, which can interfere with traveling wave analysis. Therefore, it is necessary to filter out the power frequency and its harmonic components and retain the high-frequency components of the transient traveling wave.

[0030] An adaptive notch filter is used for power frequency filtering. A notch filter is a filter that can produce deep attenuation at a specific frequency, effectively filtering out the 50Hz power frequency and its harmonic components, while having minimal impact on signals at other frequencies. The adaptive notch filter can also track minute fluctuations in the power frequency to ensure filtering effectiveness.

[0031] After the above preprocessing, outliers, baseline drift, and power frequency interference in the original signal were effectively removed, resulting in a relatively pure transient traveling wave signal, which is ready for subsequent analysis and processing.

[0032] Step 2: Construct the Hankel matrix by performing phase space reconstruction on the pure traveling wave signal, perform singular value decomposition on the matrix, select effective singular values ​​according to the distribution characteristics of the singular values ​​for signal reconstruction, and use the Teager energy operator to perform nonlinear enhancement on the denoised signal.

[0033] Although preprocessing was performed, random noise still exists in the signal, especially for weak reflected waves, where the signal-to-noise ratio remains low, making direct wavefront detection prone to errors. Therefore, further noise reduction and enhancement of wavefront features are needed.

[0034] First, the phase space of the one-dimensional traveling wave signal is reconstructed to construct a Hankel matrix. A Hankel matrix is ​​a special type of matrix where all elements on each anti-diagonal line are equal. The construction method is as follows: set the embedding dimension *m* and the time delay *τ*, then arrange the one-dimensional signal in rows, taking *m* samples per row, with adjacent rows staggered by *τ* samples. This transforms the one-dimensional signal into a two-dimensional matrix.

[0035] The choice of embedding dimension m is crucial; too small a value results in poor noise reduction, while too large a value increases computational complexity. In this embodiment, the embedding dimension is set to 50, and the time delay is set to 1, meaning each row contains 50 points, with adjacent rows staggered by 1 point. For a signal of length N, the constructed Hankel matrix has N-m+1 rows and m columns.

[0036] Singular value decomposition (SVD) is performed on the constructed Hankel matrix. SVD is a matrix factorization method that decomposes any matrix into a product of three matrices: a left singular vector matrix, a diagonal matrix of singular values, and a right singular vector matrix. The singular values ​​are arranged in descending order along the diagonal of the diagonal matrix. Larger singular values ​​correspond more closely to the principal components of the signal, while smaller singular values ​​correspond more closely to noise components. This is because the signal energy is mainly concentrated in the first few large singular values, while noise is evenly distributed across all singular values.

[0037] The key to noise reduction is determining how many effective singular values ​​to retain, which means finding the boundary between effective singular values ​​and noise singular values. If too many are retained, the noise will not be completely removed; if too few are retained, the useful signal will be lost.

[0038] This embodiment uses the singular value difference spectrum method to determine the effective order. Specifically, it calculates the difference between adjacent singular values ​​to obtain a difference sequence, i.e., a difference spectrum. Since the first few effective singular values ​​decrease rapidly, the difference is relatively large; in the noise region, the decrease in singular values ​​becomes slower, and the difference becomes smaller; and at the boundary between effective and noise, the difference exhibits a significant abrupt change, i.e., a maximum point. Therefore, in this embodiment, the position corresponding to the first maximum point in the difference spectrum is obtained, and this position is taken as the effective singular value order, serving as the boundary point distinguishing between effective and noise singular values.

[0039] In this embodiment, the effective singular value order is calculated to be 6, meaning that the first 6 singular values ​​are effective signal components, while the rest are noise components.

[0040] After determining the order of the effective singular values, that is, after selecting the effective singular values ​​through the boundary points, in this embodiment, only the first 6 large singular values ​​are retained, and the smaller singular values ​​are all set to zero. Then, the matrix is ​​reconstructed using the retained singular values ​​and the corresponding singular vectors, and the denoised matrix is ​​obtained.

[0041] Finally, the two-dimensional matrix is ​​restored to a one-dimensional signal, i.e., the denoised traveling wave signal, using the diagonal averaging method. The diagonal averaging method involves averaging the elements along each anti-diagonal line of the matrix to obtain the signal value at the corresponding time. Thus, according to the above process in this embodiment, the denoised one-dimensional traveling wave signal can be obtained.

[0042] After singular value decomposition for noise reduction, the signal-to-noise ratio of the signal is significantly improved, random noise is effectively suppressed, and the waveform characteristics of the traveling wave are clearer.

[0043] After noise reduction, the Teager energy operator is used to enhance the signal features, further highlighting the abrupt changes in the wavefront.

[0044] The Teager energy operator is a nonlinear energy calculation method. Its calculation is very simple: for a discrete signal, the Teager energy value at each time step equals the square of the current sample value minus the product of the previous and next sample values. In words, the current energy equals the square of the current value minus the product of the previous and next sample values.

[0045] The Teager energy operator has a crucial characteristic: it is highly sensitive to changes in both the amplitude and frequency of a signal, significantly enhancing abrupt changes in the signal. The wavefront of a traveling wave is precisely where signal abrupt changes are most dramatic; after calculation by the Teager energy operator, a distinct energy peak is formed at the wavefront position, making it very easy to detect.

[0046] All the calculated Teager energies are combined into a Teager energy sequence. The directly calculated Teager energy sequence has large fluctuations and many high-frequency spikes. Therefore, the Teager energy sequence is smoothed. Preferably, in this embodiment, a sliding window weighted average is used for smoothing. The window length is 5 sampling points. After weighted averaging, the energy sequence becomes smoother while retaining the main peak features.

[0047] Finally, the smoothed Teager energy sequence is normalized, with the maximum value normalized to 1, to obtain the enhanced traveling wave characteristic signal. Normalization facilitates subsequent threshold comparison and peak detection.

[0048] After two steps of SVD noise reduction and Teager energy enhancement, the originally weak and blurry traveling wave front exhibits significantly prominent features, improving the signal-to-noise ratio and creating conditions for subsequent accurate wave front detection.

[0049] Step 3: The enhanced traveling wave signal is subjected to wavefront detection using the improved AIC criterion, wherein a sliding window variable step size search strategy is introduced into the improved AIC criterion.

[0050] Wavefront detection is the core component of traveling wave positioning, and the accuracy of wavefront arrival time detection directly determines the positioning accuracy. This embodiment uses an improved AIC criterion for wavefront detection.

[0051] The AIC criterion, also known as the Akaike Information Criterion, is a classic criterion in information theory. Its basic idea is that a statistical model can be used to fit a signal, and the quality of the model can be measured by the AIC value; the smaller the AIC value, the better the model. For the traveling wavefront detection problem, the arrival time of the wavefront is the moment when the signal's statistical characteristics undergo a sudden change, which is also the moment when the AIC value is minimum.

[0052] Set up a sliding window with a length of N sampling points. Divide the window into two segments: the front segment is the noise segment before the wavefront arrives, and the back segment is the signal segment after the wavefront arrives. The dividing point of the window is the current point to be detected.

[0053] For each segmentation point, the variances of the preceding and following signals are calculated separately, and then the AIC value for that point is calculated using the AIC formula. It should be noted that the specific calculation process is existing technology and will not be elaborated upon in this embodiment.

[0054] When the dividing point is exactly at the wavefront, the first segment is pure noise with low variance, while the second segment contains signal with high variance. At this point, the overall AIC value is at its minimum. Therefore, the minimum point of the AIC curve corresponds to the arrival time of the wavefront.

[0055] Calculating the AIC value for each sampling point would be computationally intensive. To improve efficiency, a variable step-size search strategy is adopted, performing a coarse search followed by a fine search.

[0056] Therefore, this embodiment first performs a large-step coarse search, with a step size of 10 sampling points, meaning the AIC value is calculated every 10 points. A sliding window is used across the entire signal range to calculate the AIC value at each point, identifying regions with relatively small AIC values ​​to determine the approximate location range of the wavefront. Using regions where the AIC value is less than a preset threshold, the wavefront is coarsely located. In this embodiment, the max-min normalization method is used to normalize all AIC values, and the preset threshold is set to 0.35. The specific normalization process is existing technology and will not be elaborated in this embodiment. In practical applications, implementers can also use other existing normalization methods for normalization; this embodiment does not impose any special restrictions. The coarse search can quickly narrow down the search range and improve the efficiency of subsequent searches.

[0057] In this embodiment, after a coarse search, the wavefront is determined to be within the range of the 320th and 360th sampling points, with a range length of 40 sampling points.

[0058] Then, a fine-grained search with small steps is performed. Within the area of ​​40 points obtained from the coarse localization, the AIC value is calculated point by point with a step size of 1 sampling point to find the minimum point of the AIC curve. This minimum point is the wavefront arrival time with sampling-level precision.

[0059] In this embodiment, after a fine search, the minimum AIC value appears at the 342nd sampling point.

[0060] Furthermore, considering that the sampling level precision is the precision of one sampling point, corresponding to a time error of 1 microsecond, which translates to an error of approximately 300 meters in distance, a quadratic polynomial fitting method is used to obtain sub-sampling level precision in order to further improve the accuracy.

[0061] The specific method is as follows: take the minimum point of AIC and one point before and after it, for a total of three points, and use these three points to perform a quadratic polynomial fitting to obtain a quadratic curve; then find the extreme point of this quadratic curve, which is the minimum point. The position corresponding to this extreme point is the arrival time of the wavefront with subsampling precision.

[0062] By using quadratic fitting, the wavefront detection accuracy can be improved from the sampling level to the sub-sampling level, which is more accurate than one sampling interval. In this embodiment, the wavefront position obtained after fitting is the 342.37th sampling point, which is about 3 times more accurate than the simple sampling level.

[0063] Through the improved AIC criterion and three-step detection, a very accurate initial traveling wave arrival time was obtained, and the detection accuracy reached the sub-microsecond level, laying the foundation for high-precision positioning.

[0064] Step 4: Based on the waveform characteristics of the initial traveling wave and using the waveform correlation coefficient, determine whether it is a fault point reflected wave, so as to determine the arrival time of the fault point reflected wave.

[0065] After obtaining the arrival time of the initial traveling wave, it is also necessary to find the arrival time of the reflected wave from the fault point in order to calculate the time difference and fault distance. However, there are two types of reflected waves: one is reflected back from the fault point, called the fault point reflected wave; the other is reflected back after traveling to the opposite substation, called the opposite end reflected wave. These two types of reflected waves need to be distinguished, otherwise the fault distance will be calculated incorrectly.

[0066] This embodiment uses waveform correlation analysis to identify and distinguish reflected waves.

[0067] First, taking the arrival time of the initial traveling wave as the center, waveform segments of a certain length are taken before and after it as reference waveform templates. The template length must be appropriate; too short and the features are insufficient, too long and it contains too much irrelevant information. In this embodiment, the template length is 20 microseconds, which is 20 sampling points.

[0068] The initial traveling wave is the first traveling wave generated by the fault. Its waveform is the clearest and its characteristics are the most obvious, making it the most suitable reference template.

[0069] Furthermore, in the signal following the initial traveling wave, waveform segments are truncated according to a set time step, with each segment having the same length as the reference template. For each truncated segment, its correlation coefficient with the reference template is calculated.

[0070] It should be noted that the correlation coefficient is an indicator that measures the similarity between two waveforms. The value ranges from -1 to 1. The closer the correlation coefficient is to 1, the more similar the two waveforms are; the closer it is to 0, the less similar they are.

[0071] In this embodiment, the correlation coefficient is calculated as follows: first, the mean of each waveform is calculated; then, the deviation of each point from the mean is calculated; the deviations of corresponding points are multiplied and summed; finally, this summation is divided by the product of the standard deviations of the two waveforms to obtain the correlation coefficient. Implementers may also use other existing methods for calculating correlation coefficients; this embodiment does not impose any special restrictions on this.

[0072] After calculating the correlation coefficients at all locations, a correlation coefficient curve will be obtained. Several peaks will appear on the curve. Each peak corresponds to a candidate arrival time of the reflected wave, and each peak corresponds to a candidate reflected wave. The higher the peak, the more similar the waveform of the reflected wave is to the initial traveling wave, and the more likely it is to be the real reflected wave.

[0073] In this embodiment, two distinct peaks appeared on the correlation coefficient curve, near the 856th and 1623rd sampling points respectively, indicating that there are two main reflected waves.

[0074] After identifying candidate reflected waves, it's necessary to determine which is the reflected wave from the fault point and which is the reflected wave from the opposite end. This can be determined based on the physical laws of traveling wave propagation. Using the total line length and theoretical wave speed, the theoretical arrival time of the reflected wave from the opposite end can be calculated. Since the reflected wave travels from the measuring end to the opposite end and then reflects back, the propagation distance is twice the total line length, and the time is twice the line length divided by the wave speed.

[0075] The propagation distance of the reflected wave from the fault point is twice the fault distance, so its arrival time must be less than the arrival time of the reflected wave from the opposite end. Therefore, the reflected wave that arrives first is the reflected wave from the fault point, and the reflected wave that arrives later is the reflected wave from the opposite end.

[0076] Further verification can be achieved by calculating the theoretical time range. Specifically, if the fault point is between 0 and the total length, then the arrival time of the reflected wave from the fault point is between 0 and twice the total length divided by the wave speed, while the arrival time of the reflected wave from the other end is near twice the total length divided by the wave speed. Based on which interval the arrival time of the candidate reflected wave falls within, its type can be determined. If the arrival time of the candidate reflected wave falls within the theoretical arrival time range of the fault point reflected wave, it is identified as a fault point reflected wave; otherwise, it is an other end reflected wave.

[0077] In this embodiment, the line is 120 kilometers long, with a theoretical wave speed of approximately 297 meters per microsecond. The theoretical arrival time of the reflected wave at the other end is approximately 808 microseconds, which is near the 808th sampling point. The first peak is at the 856th sampling point, close to the theoretical time of the reflected wave at the other end, so it is the reflected wave at the other end. The second peak is at the 1623rd sampling point, approximately twice the time of the reflected wave at the other end, therefore it is the reflected wave from the fault point after two reflections.

[0078] By combining waveform correlation analysis with physical law verification, the reflected waves of the fault point can be accurately and automatically identified, solving the problem of difficult identification of reflected waves in single-end positioning.

[0079] Step 5: Based on the actual structural and operational parameters of the overhead line, adaptively estimate the traveling wave velocity, and then combine the time difference between the initial traveling wave and the reflected wave at the fault point to obtain the distance from the fault point to the measurement end, so as to obtain the fault location result.

[0080] Once the arrival times of the initial wave and the reflected wave from the fault point are obtained, the time difference can be calculated. However, to calculate the fault distance, the propagation speed of the traveling wave also needs to be known. Traditional methods mostly use a fixed wave speed, such as 98% of the speed of light, but the actual wave speed is affected by many factors, and a fixed wave speed will introduce errors.

[0081] This embodiment proposes an adaptive wave velocity estimation method based on line parameters, which calculates the wave velocity online according to the actual line parameters.

[0082] First, obtain the various structural parameters of the overhead line, including: conductor type, conductor radius, number of split conductors and spacing between split conductors, tower height, conductor height above ground, phase-to-phase distance, and earth resistivity. These parameters can be obtained from the line's design documents and records.

[0083] In this embodiment, the conductor type is LGJ-400 / 35, the aluminum wire diameter is about 26.8 mm, the steel core diameter is about 7.5 mm; it is a double-split conductor with a split spacing of 400 mm; the average conductor height above the ground is about 25 meters; the phase-to-phase distance is about 4.5 meters; and the average earth resistivity is about 200 ohms.

[0084] Based on the line structure parameters, the resistance, inductance, capacitance, and conductance per unit length are calculated using a distributed parameter model.

[0085] Resistance per unit length: The DC resistance is calculated based on the material and cross-sectional area of ​​the wire, and then the effect of the skin effect on the AC resistance is considered to obtain the resistance per unit length at the corresponding frequency.

[0086] Inductance per unit length comprises two parts: internal inductance and external inductance. Internal inductance is the inductance within the conductor and is related to the conductor material and frequency; external inductance is the inductance between the conductors and is related to their geometric position. The external inductance is calculated using electromagnetic field theory formulas based on parameters such as conductor radius, height above ground, and phase distance. The total inductance per unit length is obtained by adding the two components together.

[0087] Capacitance per unit length: Based on the geometry and location of the conductor, calculate the capacitance between the conductor and the ground, and between conductors, to obtain the positive-sequence capacitance and zero-sequence capacitance per unit length.

[0088] Conductivity per unit length: mainly composed of leakage conductivity and corona loss of the insulator. For normally operating lines, the conductivity is generally very small, and in this embodiment it is set to 0.

[0089] The propagation speed of a traveling wave is related to the distributed parameters of the line. For transient traveling waves, the main consideration is the modulus propagation speed, which is the propagation speed of the linear modulus component.

[0090] The specific method for obtaining wave speed is existing technology. In this embodiment, the calculation method is: wave speed equals 1 divided by the square root of (inductance per unit length multiplied by capacitance per unit length). This formula is derived from transmission line theory and is the basic formula for traveling wave propagation.

[0091] However, both inductance and capacitance are frequency-dependent, especially inductance, where the skin effect varies at different frequencies, resulting in different internal inductances. Traveling waves, on the other hand, are broadband signals, and the wave speeds of different frequency components differ slightly; this is the phenomenon of wave dispersion.

[0092] In actual calculations, the wave velocity corresponding to the frequency range where the main energy of the traveling wave is concentrated is taken as the equivalent wave velocity. In this embodiment, the main frequency range of the traveling wave is between tens of kHz and hundreds of kHz, and the wave velocity corresponding to the center frequency of 100 kHz is taken as the calculated wave velocity.

[0093] Calculations show that the traveling wave propagation speed of this circuit is approximately 296.8 meters per microsecond, which is about 99% of the speed of light, slightly higher than the traditional 98%.

[0094] Once the wave speed and time difference are obtained, the fault distance can be calculated. The fault distance equals the wave speed multiplied by the time difference between the initial wave and the reflected wave from the fault point, then divided by 2. Dividing by 2 is because the reflected wave travels a round trip, and the time difference corresponds to twice the propagation time of the fault distance.

[0095] In this embodiment, the initial wave arrives at the 342.37th sampling point, and the reflected wave from the fault point arrives at the 856.23rd sampling point, with a time difference of 513.86 microseconds. Multiplying this by the wave speed of 296.5 meters per microsecond and dividing by 2, we obtain a fault distance of approximately 76.1 kilometers. That is, the fault point is approximately 76.1 kilometers from the measurement end.

[0096] After on-site verification, the actual fault location was found to be at 75.8 kilometers, with a positioning error of about 300 meters and a relative error of less than 0.4%, which is much higher than that of traditional methods.

[0097] Furthermore, in order to improve the reliability of the positioning results, this embodiment also introduces a multi-scale consistency verification mechanism.

[0098] The clean traveling wave signal is decomposed into traveling wave signals at different scales. The larger the scale, the lower the corresponding frequency; the smaller the scale, the higher the corresponding frequency. In this embodiment, it is decomposed into 4 scales, each corresponding to a different frequency band.

[0099] At each scale, all the preceding steps are repeated: noise reduction and enhancement, wavefront detection, reflected wave identification, wave velocity estimation, and distance calculation, to obtain the fault location results at each scale. This results in four different scales of fault location results.

[0100] Calculate the mean and standard deviation of the four positioning results. Divide the standard deviation by the mean to obtain the coefficient of variation, which serves as an indicator of dispersion. If the coefficient of variation is less than a set threshold (2% in this example), it indicates that the positioning results at each scale are relatively consistent and the positioning results are reliable. If the coefficient of variation is greater than the threshold, it indicates that the results at a certain scale may be problematic, and outliers need to be removed.

[0101] The outlier detection method uses the Grubbs criterion, which examines the result that deviates the most from the mean to see if it is an outlier. If it is, it is removed, and the mean is recalculated using the remaining results.

[0102] After consistency verification, the final mean fault distance and location confidence level are output. The confidence level is determined based on the degree of dispersion; the smaller the dispersion, the higher the confidence level, and the larger the dispersion, the lower the confidence level.

[0103] In this embodiment, the positioning results at the four scales are 76.3 km, 75.9 km, 76.5 km and 75.8 km, respectively, with an average of 76.1 km and a coefficient of variation of approximately 0.4%. The consistency is very good and the confidence level is very high.

[0104] Multi-scale consistency verification not only improves the reliability of positioning results, but also provides the confidence level of the positioning, allowing maintenance personnel to know the credibility of the results.

[0105] To further improve positioning accuracy, historical fault data can be used to establish an error correction model to correct the positioning results. Since the terrain, soil resistivity, and tower structure along the line may vary, the actual wave velocity in different sections is not entirely the same. The entire line can be divided into several segments according to tower sections, and the corresponding wave velocity for each segment can be calculated based on local parameters to establish a segmented wave velocity model.

[0106] During fault location, the wave velocity of the corresponding section is checked based on the initial fault location, and then a more accurate fault distance is recalculated. By correcting for segmented wave velocity, the location error caused by uneven line parameters can be eliminated.

[0107] Collect historical fault location results and actual fault locations, calculate the location error for each line segment, establish an error correction database, and record the average error and error distribution for each segment in the database.

[0108] For new fault location results, based on the initial fault location, the average correction amount of the corresponding section is searched from the error correction database, and the location results are corrected for error compensation. The corrected results are closer to the actual fault location, and the positioning accuracy can be further improved after error correction.

[0109] Finally, the final fault location results, location reliability, fault time, and other information will be output and displayed for maintenance personnel to refer to.

[0110] Based on the same inventive concept as the above method, this application embodiment also provides an overhead line single-end fault traveling wave location device, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the above-described overhead line single-end fault traveling wave location methods.

[0111] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the protection scope of this application.

Claims

1. A method for locating traveling waves of a single-end fault in an overhead line, characterized in that, Includes the following steps: The transient voltage and transient current traveling wave signals after the fault are preprocessed by outlier removal, baseline correction and power frequency component filtering to obtain a clean traveling wave signal; A Hankel matrix is ​​constructed by reconstructing the phase space of a clean traveling wave signal. The matrix is ​​then subjected to singular value decomposition. Based on the distribution characteristics of the singular values, effective singular values ​​are selected for signal reconstruction. Finally, the Teager energy operator is used to nonlinearly enhance the denoised signal. The enhanced traveling wave signal is subjected to wavefront detection using an improved AIC criterion, wherein a sliding window variable step size search strategy is introduced into the improved AIC criterion. Based on the waveform characteristics of the initial traveling wave and using the waveform correlation coefficient, it is determined whether it is a fault point reflected wave, so as to determine the arrival time of the fault point reflected wave. Based on the actual structural and operational parameters of the overhead line, the traveling wave velocity is adaptively estimated. Then, by combining the time difference between the initial traveling wave and the reflected wave at the fault point, the distance from the fault point to the measurement end is obtained to obtain the fault location result.

2. The method for locating a traveling wave fault at one end of an overhead line as described in claim 1, characterized in that, The preprocessing specifically includes: The Dixon test is used to identify outliers in the sampling. The data are sorted by size and the test statistic is calculated. The statistic is compared with the critical value to determine whether it is an outlier. For the identified outliers, linear interpolation of adjacent points is used to replace them. Polynomial fitting baseline correction method is used to eliminate DC offset and low-frequency drift in the signal; An adaptive notch filter is used to remove the power frequency and its harmonic components, while retaining the high-frequency components of the transient traveling wave, thus obtaining a pure traveling wave signal.

3. The method for locating a traveling wave fault at one end of an overhead line as described in claim 1, characterized in that, Singular value decomposition is used for traveling wave signal denoising, specifically including: The signals are arranged into a matrix according to the set embedding dimension and time delay to obtain the Hankel matrix. Singular value decomposition is performed on the Hankel matrix, and the difference spectrum of the singular values ​​is calculated. The order corresponding to the first maximum point in the difference spectrum is used as the boundary between effective singular values ​​and noisy singular values. Select the aforementioned number of valid singular values, reconstruct the corresponding singular vectors to obtain the denoised matrix, and then obtain the denoised traveling wave signal by the diagonal averaging method.

4. The method for locating a traveling wave fault at one end of an overhead line as described in claim 3, characterized in that, The nonlinear enhancement of the denoised signal using the Teager energy operator specifically includes: The Teager energy value is calculated point by point for the denoised traveling wave signal, and the sequence of all calculated Teager energies is smoothed; the smoothed Teager energy sequence is normalized to obtain the enhanced traveling wave signal.

5. The method for locating a traveling wave fault at one end of an overhead line as described in claim 1, characterized in that, The improved AIC criterion for wavefront detection incorporates a sliding window variable step size search strategy, specifically including: Set the sliding window length for AIC calculation, and slide the window across the entire signal range with a preset step size to calculate the AIC value point by point. Use the area corresponding to the AIC value being less than the preset threshold to coarsely locate the wavefront. Within the area obtained by coarse localization, the AIC value is calculated point by point using a sliding window with sampling points as the step size, and the minimum AIC point is found. The minimum AIC point and its three adjacent points with preceding and following AIC values ​​are fitted, and the extreme point of the fitted curve is taken as the arrival time of the wavefront.

6. The method for locating a traveling wave fault at one end of an overhead line as described in claim 1, characterized in that, The process for determining the reflected wave at the fault point specifically includes: Waveform segments of a set length before and after the arrival time of the initial traveling wave are extracted as reference waveform templates. In the signal after the initial traveling wave, waveform segments are truncate according to a set time step. The correlation coefficient of each segment is calculated with the reference waveform template, and the peak point of all correlation coefficients is extracted. The corresponding time is used as the arrival time of the candidate reflected wave. The theoretical arrival time ranges for the reflected waves from the fault point and the reflected waves from the opposite end are set. If the arrival time of a candidate reflected wave falls within the theoretical arrival time range of the reflected wave from the fault point, it is determined to be a reflected wave from the fault point.

7. The method for locating a traveling wave fault at one end of an overhead line as described in claim 1, characterized in that, The adaptive estimation of traveling wave velocity based on the actual structural and operational parameters of the overhead line specifically includes: Obtain the structural parameters of the overhead line, including conductor type, conductor radius, split spacing, tower height, phase-to-phase distance, and ground resistivity; based on the line structural parameters, use a distributed parameter model to calculate the resistance, inductance, capacitance, and conductance per unit length of the line; Based on the frequency range of the traveling wave, the wave impedance and propagation coefficient at the corresponding frequency are calculated, and then the propagation speed of the traveling wave is calculated as the wave velocity of the traveling wave.

8. The method for locating a traveling wave fault at one end of an overhead line as described in claim 1, characterized in that, After calculating the fault distance, a multi-scale consistency verification step is also included, specifically: The pure traveling wave signal is decomposed into multiple scales to obtain traveling wave signals at different scales; wavefront detection and fault distance calculation are performed at each scale to obtain positioning results at multiple scales. Calculate the dispersion between positioning results at each scale. If the dispersion is less than a set threshold, the positioning result is reliable. If the dispersion is greater than the threshold, remove the abnormal scales and recalculate the average positioning result. Output the final fault distance and location confidence that have been verified for consistency.

9. The method for locating a traveling wave fault at one end of an overhead line as described in claim 7, characterized in that, It also includes a positioning error correction step, specifically including: The entire line is divided into several segments according to tower sections, and the wave velocity of the traveling wave corresponding to each segment is calculated separately to locate faults based on the segmentation. An error correction database is established based on the historical fault location results and the actual fault location. For new fault location results, the corresponding correction amount is retrieved from the error correction database based on the fault location to compensate for the error in the location results.

10. A traveling wave locating device for a single-end fault in an overhead line, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the traveling wave location method for single-end faults in overhead lines as described in any one of claims 1-9.

Images