Fault distance measurement method and system based on traveling wave signal of circuit breaker

Abstract

The invention provides a circuit breaker traveling wave signal-based fault location method and system, and relates to the technical field of power grid fault detection, and the method comprises the following steps: constructing a line topology library containing tower coordinates and conductor sags; collecting a transient traveling wave signal generated during opening or closing operation of the circuit breaker; determining a traveling wave head arrival time set; calculating the geographic coordinates of the fault point and the tower section to which the fault point belongs; and according to the calculated geographic coordinates of the fault point and the tower section to which the fault point belongs, calculating the electrical distance of the line of the fault point relative to a preset reference point, and obtaining a final fault distance measurement result. An equivalent distance is constructed by using an initial wave and a reflected wave, and a clock synchronization error is eliminated; and in combination with least square solution and coordinate projection, the electrical measurement value is mapped into the geographic coordinate of the specific tower section, so that the fault finding time can be shortened from the hour level to the minute level, and the power supply reliability is greatly improved.

Description

Technical Field

[0001] This invention relates to the field of power grid fault detection technology, and more specifically, to a fault location method and system based on circuit breaker traveling wave signals. Background Technology

[0002] As a major component of my country's power system, the distribution network undertakes the crucial task of transmitting and distributing electrical energy. It is the network in close contact with users within the transmission chain, and its safe and stable operation is inextricably linked to the normal operation and development of various industries in my country. Therefore, rapid fault detection and troubleshooting in the distribution network are particularly important. While short-circuit faults in distribution networks are mostly single-phase grounding faults, in actual engineering projects, the complex network structure with numerous feeder branches, flexible load transfer operations, and potential hazards such as trees, wires, and foreign objects make the operating environment of the distribution network quite complex.

[0003] Currently, fault location technology based on the traveling wave principle has attracted widespread attention due to its advantages such as high positioning accuracy and fast response speed. However, existing traveling wave location methods still have the following technical problems in practical applications: Traditional traveling wave location methods usually use a fixed theoretical wave velocity for distance calculation, but the actual traveling wave velocity is affected by various factors such as line type, conductor sag, ambient temperature, and load current, exhibiting dynamic variation characteristics. The assumption of a fixed wave velocity leads to a large error in the location measurement results. At the same time, in the actual complex power grid environment, the traveling wave signal is often mixed with power frequency components, high-frequency noise, and mechanical vibration interference generated by circuit breaker operation. Traditional methods are difficult to accurately extract the arrival time of the traveling wave surge from the strong noise background, which easily leads to missed or false detections of the wavefront. Especially when the fault occurs near the voltage zero crossing, the traveling wave signal energy is weak, making wavefront identification even more difficult.

[0004] No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention

[0005] In view of this, the present invention provides a fault location method and system based on circuit breaker traveling wave signals to solve the aforementioned problems.

[0006] To solve the above problems, the specific technical solution adopted by the present invention is as follows: According to one aspect of the present invention, a fault location method based on circuit breaker traveling wave signals is provided, comprising the following steps: S1. Collect point cloud data of circuit breaker lines and generate line point cloud models through 3D reconstruction; use the line point cloud models to extract tower coordinates and conductor sag status, and construct a line topology library containing tower coordinates and conductor sag. S2. Based on the current transformer pre-installed on the circuit breaker, collect the transient traveling wave signal generated during the opening or closing operation of the circuit breaker. S3. Preprocess the transient traveling wave signal and determine the arrival time of the traveling wave surge based on the preprocessed transient traveling wave signal; based on the energy flow direction and combined with the three-phase different opening and closing periodicity parameters of the circuit breaker and the contact closing bounce time, perform error analysis on the arrival time of the traveling wave surge and determine the arrival time set of the traveling wave wavehead. S4. Based on the arrival time set of the traveling wave front, use the time difference positioning algorithm for two-way ranging to determine the propagation path and reflection point of the fault traveling wave, and combine it with the line topology library to calculate the geographical coordinates of the fault point and the tower section to which it belongs. S5. Based on the calculated geographical coordinates of the fault point and its corresponding tower section, calculate the electrical distance of the fault point along the line relative to the preset reference point to obtain the final fault distance measurement result.

[0007] Preferably, the preprocessing of the transient traveling wave signal and the determination of the arrival time of the traveling wave surge based on the preprocessed transient traveling wave signal; and the error analysis of the arrival time of the traveling wave surge based on the energy flow direction and combined with the three-phase different opening and closing periodicity parameters of the circuit breaker and the contact closing bounce time, to determine the arrival time set of the traveling wave front, include the following steps: S31. Perform filtering and noise reduction preprocessing on the transient traveling wave signal, and perform wavelet transform processing on the preprocessed transient traveling wave signal to obtain a two-dimensional time spectrum. S32. Construct a weighting function matrix using a two-dimensional time-frequency spectrum diagram, and determine the velocity and direction of each energy point by calculating the optimized direction field of the energy points to form an energy flow distribution; S33. Based on the energy flow distribution and the sliding analysis window along the time axis, calculate the direction entropy and amplitude entropy of the energy flow vector in each window; combine the average movement velocity of the energy points in the window to construct the traveling wave disorder index, and combine it with the preset disorder threshold to determine the arrival time of the candidate traveling wave surge. S34. Using the three-phase different opening and closing period parameters of the circuit breaker and the contact closing bounce time as prior constraints, error analysis is performed on the arrival time of candidate traveling wave surges, and the final set of traveling wave front arrival times is obtained through screening and interpolation.

[0008] Preferably, the step of constructing a weighting function matrix using a two-dimensional time-frequency spectrum and determining the velocity and direction of each energy point by calculating the optimized direction field of the energy points to form an energy flow distribution includes the following steps: S321. Based on the energy amplitude of each pixel in the two-dimensional time-frequency spectrum and the distribution gradient of each pixel on the frequency axis, construct a composite weighting function matrix that integrates amplitude intensity and frequency change rate. S322. Based on the composite weight function matrix and combined with the covariance manifold method, calculate the initial direction field of energy flow; S323. By utilizing the range constraint of the propagation speed of the transient traveling wave signal in the time-frequency plane and the prior of frequency attenuation, the initial direction field of the energy flow is iteratively smoothed and corrected to obtain the optimized direction field of the energy point. S324. Based on the optimized direction field of the energy points, the velocity of each energy point is estimated by calculating the rate of change of energy amplitude along the direction field, and the velocity is combined with the optimized direction field to obtain the velocity and direction of each energy point, so as to form an energy flow distribution.

[0009] Preferably, the calculation of the initial direction field of the energy flow based on the composite weight function matrix and combined with the covariance manifold method includes the following steps: S3221. Treat the composite weight function matrix as a two-dimensional image, and divide the two-dimensional image into several overlapping rectangular regions; for each rectangular region, extract the multi-dimensional features of each pixel within the rectangular region to form a feature vector set; S3222. Based on the feature vector set of each rectangular region, calculate the covariance matrix of all feature vectors within the rectangular region to obtain the covariance descriptor of the rectangular region. S3223. Iteratively solve the covariance descriptor of all rectangular regions to obtain the Karcher mean point used as a reference benchmark. S3224. Using logarithmic mapping, the covariance descriptor of each rectangular region is projected onto the tangent space at the Karcher mean point to obtain the vector in the tangent space. S3225. For each rectangular region, extract the components corresponding to the frequency gradient features and time gradient features from the tangent space vector, and construct the local gradient covariance matrix of the rectangular region; calculate the eigenvector corresponding to the largest eigenvalue of the gradient covariance matrix, and use its direction as the initial direction of energy flow at the center point of the rectangular region. S3226. Smooth the initial energy flow direction at the center point of all regions to obtain the initial energy flow direction field.

[0010] Preferably, the step of iteratively smoothing and correcting the initial direction field of the energy flow by utilizing the propagation speed range constraint of the transient traveling wave signal in the time-frequency plane and the frequency attenuation prior to obtain the optimized direction field of the energy point includes the following steps: S3231. Based on the type of circuit breaker line, determine the propagation characteristics of transient traveling wave signal in the line, determine the propagation speed range constraint and frequency attenuation prior of transient traveling wave signal in the time-frequency plane, and construct the direction field optimized energy functional according to the propagation speed range constraint and frequency attenuation prior. S3232, Iteratively solve the energy functional for the direction field optimization, and smooth and correct the initial direction field of the energy flow in each iteration; S3233. When the change in the orientation field between two adjacent iterations is less than the preset convergence threshold or the preset maximum number of iterations is reached, the iteration is stopped, and the optimized orientation field of each energy point is obtained.

[0011] Preferably, the step of using the three-phase different opening and closing periodicity parameters of the circuit breaker and the contact closing bounce time as prior constraints to perform error analysis on the arrival times of candidate traveling wave surges, and obtaining the final set of traveling wave front arrival times through screening and interpolation processing, includes the following steps: S341. Arrange the candidate wave surge arrival times in chronological order and calculate the pairwise time difference between each phase. If the time difference between any two phases exceeds the allowable range of the three-phase opening and closing asynchrony parameter of the circuit breaker, it is judged as abnormal data and deleted to obtain the candidate times after preliminary screening. S342. Starting from each candidate moment after initial screening, extend the duration corresponding to the contact closing bounce time parameter forward and backward as the failure-to-operate time window; if multiple candidate moments appear in the same failure-to-operate time window, retain the moment with the largest energy amplitude as the effective wavehead, and the rest are judged as pseudo waveheads generated by mechanical bounce and removed, to obtain the wavehead moment after secondary screening. S343. Based on the tower coordinates and line length parameters in the line topology library, calculate the electrical distance corresponding to the time difference between adjacent wavefront times. If the electrical distance has an integer multiple relationship with the line length or tower spacing, it is determined to be a reflected wave or a refracted wave and retained; otherwise, it is discarded as noise interference, and the wavefronts after three screenings are obtained. S344. For each wavefront time after three screenings, extract the original waveform segments of preset length before and after it, and use spline interpolation to perform local fitting. Take the position with the largest energy change rate as the final arrival time of the traveling wavefront, and output the complete set of traveling wavefront arrival times.

[0012] Preferably, the step of determining the propagation path and reflection point of the fault traveling wave based on the arrival time set of the traveling wave wavefront, using a two-way ranging time difference localization algorithm, and calculating the geographical coordinates of the fault point and its corresponding tower section in conjunction with the line topology library includes the following steps: S41. Based on the arrival time set of the traveling wave front, extract the arrival time of the initial traveling wave surge and the time of the reflected wave at the fault point, and determine the wave velocity of the transient traveling wave signal by combining the line topology library. S42. Based on the arrival time of the initial traveling wave surge, the time of the reflected wave at the fault point, and the wave velocity of the transient traveling wave signal, calculate the equivalent distance between the circuit breaker and the fault point. S43. Take the circuit breaker that first receives the traveling wave surge as the reference station, and combine the location information of the circuit breaker to establish a linear equation for the coordinates of the fault point. S44. Solve the linear equation between the fault point coordinates and the reference distance to obtain the preliminary geographical coordinates of the fault point. Combine this with the line topology library to determine the geographical coordinates of the fault point and the tower section to which it belongs.

[0013] Preferably, determining the wave velocity of the transient traveling wave signal by combining the line topology library includes the following steps: Extract information on all known lengths of intact line sections without faults from the line topology library, and extract the arrival time of the traveling wave front generated by the last operation of the circuit breakers at both ends of each line section before the fault from the traveling wave front arrival time set. For each intact line segment, the time difference is calculated based on the arrival time of the traveling wave fronts of the circuit breakers at both ends; the length of the line segment in the line topology library is used to calculate the candidate value of dynamic wave velocity under the current environment. The consistency of the candidate dynamic wave velocity values ​​calculated from multiple intact line sections is checked, and after removing outliers, a weighted average is taken to obtain the final dynamic wave velocity of the current line transient traveling wave signal.

[0014] Preferably, the step of using the circuit breaker that first receives the traveling wave surge as a reference station and, in conjunction with the location information of the circuit breaker, establishing a linear equation for the fault point coordinates includes the following steps: S431. Based on the arrival time set of the traveling wave front and the equivalent distance, the circuit breaker that first receives the traveling wave surge is determined as the reference station. S432. For each other circuit breaker, extract the coordinates from the line topology library and calculate the reference station distance difference by combining them with the equivalent distance of the reference station. S433. Based on the principle of time of arrival difference positioning, the quadratic term is eliminated by the square subtraction method, and a linear equation about the coordinates of the fault point is established by combining the coordinates of each circuit breaker and the distance difference of the reference station.

[0015] According to another aspect of the present invention, a fault location system based on circuit breaker traveling wave signals is provided, the system comprising: The data acquisition module is used to collect point cloud data of circuit breaker lines and generate line point cloud models through 3D reconstruction; it uses the line point cloud models to extract tower coordinates and conductor sag status, and constructs a line topology library containing tower coordinates and conductor sag. The signal acquisition module is used to acquire transient traveling wave signals generated during the opening or closing operation of the circuit breaker based on the current transformers pre-installed on the circuit breaker, and to record the arrival time of the transient traveling wave signal of each circuit breaker. The time analysis module is used for preprocessing transient traveling wave signals by filtering and denoising, and then performing wavelet transform on the preprocessed transient traveling wave signals to obtain a two-dimensional time spectrum. A weighting function matrix is ​​constructed using the two-dimensional time spectrum, and the velocity and direction of each energy point are determined by calculating the optimized direction field of the energy points to form an energy flow distribution. Based on the energy flow distribution and a sliding analysis window along the time axis, the directional entropy and amplitude entropy of the energy flow vector within each window are calculated. A traveling wave disorder index is constructed by combining the average velocity of the energy points within the window, and a preset disorder threshold is used to determine the arrival times of candidate traveling wave surges. Using the three-phase different opening and closing periodicity parameters of the circuit breaker and the contact closing bounce time as prior constraints, error analysis is performed on the arrival times of the candidate traveling wave surges, and the final set of traveling wave front arrival times is obtained through screening and interpolation. The coordinate calculation module is used to determine the propagation path and reflection point of the fault traveling wave based on the arrival time set of the traveling wave wavehead, using a two-way ranging time difference positioning algorithm, and combined with the line topology library, to calculate the geographical coordinates of the fault point and the tower section to which it belongs. The distance measurement result calculation module is used to calculate the electrical distance of the fault point along the line relative to the preset reference point based on the calculated geographical coordinates of the fault point and the tower section to which it belongs, and to obtain the final fault distance measurement result.

[0016] The beneficial effects of this invention are as follows: 1. This invention utilizes point cloud data to construct a line topology library containing real-time sag, which can provide an accurate geographical reference for ranging; it extracts the energy flow direction from the time-spectrum map through covariance manifold, significantly improving the wavefront identification accuracy; it constructs an equivalent distance using the initial wave and reflected wave, eliminating clock synchronization errors; and it combines least squares solution and coordinate projection to map electrical measurement values ​​to the geographical coordinates of specific tower sections, which can shorten the fault finding time from hours to minutes, greatly improving power supply reliability. 2. This invention transforms the time-frequency spectrum into a two-dimensional image. By extracting the energy flow direction, it overcomes the shortcomings of traditional threshold methods that are susceptible to noise interference. It utilizes the inherent mechanical parameters of the circuit breaker as prior constraints to effectively eliminate spurious wavefronts generated by equipment operation. Furthermore, it constructs a traveling wave disorder index using directional entropy, amplitude entropy, and average velocity to achieve adaptive detection of wavefront arrival time. Combined with a line topology library, it performs logical verification on the reflected wave to further eliminate noise interference. This enables high-precision extraction of the arrival time of the traveling wavefront from complex transient signals, laying a reliable foundation for subsequent fault location. 3. This invention utilizes the known length of intact sections in the line topology library to invert dynamic wave velocity online, eliminating the influence of environmental factors on wave velocity. It employs a two-way ranging concept and constructs an equivalent distance through the initial wave and reflected wave, thus eliminating the dependence on strict synchronization of multiple-terminal clocks. It linearizes the nonlinear positioning equation using the square subtraction method and combines it with least squares solution to fully utilize multi-station redundancy information to improve the robustness of the solution. Finally, it projects the coordinates onto the line centerline and matches the tower sections, achieving a precise mapping from electrical measurements to geographical coordinates, providing maintenance personnel with directly navigable fault location information. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly described below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a flowchart of a fault location method based on a circuit breaker traveling wave signal according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a fault location system based on a circuit breaker traveling wave signal according to an embodiment of the present invention.

[0018] In the picture: 1. Data acquisition module; 2. Signal acquisition module; 3. Time analysis module; 4. Coordinate calculation module; 5. Distance measurement result calculation module. Detailed Implementation

[0019] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0020] According to an embodiment of the present invention, a fault location method and system based on circuit breaker traveling wave signals are provided.

[0021] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figure 1 As shown, according to an embodiment of the present invention, a fault location method based on circuit breaker traveling wave signals is provided, comprising the following steps: S1. Collect point cloud data of circuit breaker lines and generate line point cloud models through 3D reconstruction; use the line point cloud models to extract tower coordinates and conductor sag status, and construct a line topology library containing tower coordinates and conductor sag. Specifically, the circuit breaker in this invention is a ZW32-12 series outdoor high-voltage vacuum circuit breaker, hereinafter referred to as the circuit breaker, which is an outdoor power distribution device with a rated voltage of 12KV and a three-phase current of 50Hz. It is mainly used for breaking and closing load currents in power systems. It is suitable for protection and control in substations and industrial and mining enterprise power distribution systems, and is even more suitable for rural power grids and locations with frequent operations. This instruction manual specifies the reference standards, operating environment conditions, model and rated parameters, structural characteristics, working principle, ordering information, and the principles and methods for operation, installation, use, and maintenance of the circuit breaker. The parameters of the circuit breaker are shown in Table 1.

[0022] Table 1 Circuit breaker parameters

[0023] It should be noted that a drone-based LiDAR system is used to scan the circuit breaker line under test. By emitting laser pulses and receiving reflected signals, high-density, high-precision three-dimensional point cloud data of conductors, towers, insulator strings, hardware, and other objects within the passageway is acquired. Then, specialized point cloud data processing software is used to denoise, classify, and reconstruct the 3D structure of the original point cloud. Specifically, this includes distinguishing between ground points, tower points, and conductor points; performing spatial curve fitting on individual conductors to generate a line point cloud model that accurately reflects the spatial direction and shape of the conductors. This line point cloud model contains the geometric information of the circuit breaker line. The construction of the line point cloud model is existing technology and will not be elaborated upon further here.

[0024] Based on this, the following key parameters are extracted from the line point cloud model and a line topology library is constructed: (1) Extract the precise latitude and longitude coordinates and elevation information of the center point or feature point of each tower to form a spatial skeleton coordinate sequence of the line. This coordinate sequence can be directly substituted into the linear equation as the location information of the circuit breaker to establish the geometric relationship between the fault point coordinates and the reference station; it can also be used to project the calculated preliminary coordinates of the fault point onto the line centerline, and determine the specific tower section to which the fault point belongs by calculating the perpendicular distance with the coordinates of adjacent towers. (2) Based on the point cloud model, the spatial curve of the conductor at the current moment is extracted. Combined with the synchronously acquired ambient temperature, solar radiation intensity and line load data, the real-time sag value of the conductor under different operating conditions is calculated, and then the accurate line length of each section after sag correction is obtained. The corrected line length can be used for subsequent dynamic wave velocity inversion calculation, etc. (3) Automatically identify and measure the clearance distance between the line and the road, river, other power line or building below, and record the location and attribute information of the crossing point. This layer is used to identify abnormal reflected waves generated by the crossing object to avoid misjudging them as reflected waves of the fault point; at the same time, it provides environmental background information for the subsequent output of the geographical coordinates of the fault point; (4) Divide the entire line into several sections according to adjacent circuit breakers or towers, and record the starting and ending numbers, the precise length after sag correction, the line type, the conductor material, the theoretical wave velocity, and other basic parameters of each section. This parameter table can be used as a candidate source for intact sections and as a basis for reading the theoretical wave velocity when there is no effective inversion data.

[0025] S2. Based on the current transformers pre-installed on the circuit breakers, the transient traveling wave signals generated during the opening or closing operations of the circuit breakers are collected; at the same time, the arrival time of the transient traveling wave signals of each circuit breaker is also recorded. It should be noted that when acquiring and timing transient traveling wave signals, broadband electronic voltage transformers and electronic current transformers pre-installed on the circuit breaker are used. Electronic transformers have wideband response characteristics, typically reaching several MHz or even higher, and can accurately reflect the high-frequency transient components contained in the traveling wave signal, avoiding signal distortion. The transformers are installed on the outgoing line side or bus side of the circuit breaker to sense the transient voltage and current changes on the line in real time.

[0026] Specifically, when a line fault occurs or during normal operation, the opening or closing action of the circuit breaker generates a transient traveling wave signal that propagates to both ends of the line at the fault point or the circuit breaker disconnection. A combined continuous sampling and trigger-based storage operating mode is adopted: the instrument transformer continuously collects analog voltage and current signals of the line, which are then sent to the digital signal processor after analog-to-digital conversion. When a sudden change in signal amplitude or energy exceeding a preset threshold is detected, automatic recording is triggered, saving complete traveling wave waveform data for a period of time before and after the trigger moment, including waveforms before, during, and after the fault, ensuring the integrity of the traveling wave front. Simultaneously, to achieve subsequent multi-terminal collaborative positioning, the precise arrival time of each traveling wave signal needs to be labeled. Specifically, a BeiDou / GPS dual-mode timing module can be used to provide high-precision clock synchronization for the system. The timing module receives satellite signals and performs nanosecond-level synchronization calibration of the local clock using pulse-per-second (PPS) signals and serial port time information. When a traveling wave trigger signal is generated, the system immediately latches the current local clock count, obtains the absolute arrival timestamp of the traveling wave signal, and stores this timestamp along with the corresponding waveform data.

[0027] Specifically, for each circuit breaker terminal, the data to be recorded includes: raw waveform data, which is a sequence of voltage and current samples taken for a certain duration before and after the triggering moment, with a sampling rate typically not less than 10MHz; absolute arrival timestamp, the precise moment when the traveling wave surge first crosses the triggering threshold; channel identifier, recording which phase (A, B, C) the signal originates from and the corresponding circuit breaker number; and operation type identifier, recording whether the triggering signal was generated by a tripping or closing operation, facilitating targeted processing by subsequent algorithms.

[0028] In addition, the raw traveling wave data collected is processed locally on the terminal, including zero drift correction and power frequency component filtering. Then, the traveling wave waveform data with precise time tags is uploaded in real time to the fault ranging master station or edge computing node via fiber optic Ethernet or 4G / 5G wireless communication network for subsequent wavehead identification and collaborative positioning calculation.

[0029] S3. Preprocess the transient traveling wave signal and determine the arrival time of the traveling wave surge based on the preprocessed transient traveling wave signal; based on the energy flow direction and combined with the three-phase different opening and closing periodicity parameters of the circuit breaker and the contact closing bounce time, perform error analysis on the arrival time of the traveling wave surge and determine the arrival time set of the traveling wave wavehead. In a preferred embodiment, the preprocessing of the transient traveling wave signal and the determination of the arrival time of the traveling wave surge based on the preprocessed transient traveling wave signal, and the error analysis of the arrival time of the traveling wave surge based on the energy flow direction and combined with the three-phase different opening and closing periodicity parameters of the circuit breaker and the contact closing bounce time, to determine the arrival time set of the traveling wave front, include the following steps: S31. Perform filtering and noise reduction preprocessing on the transient traveling wave signal, and perform wavelet transform processing on the preprocessed transient traveling wave signal to obtain a two-dimensional time spectrum. It should be noted that the original transient traveling wave signal contains various interference components, mainly including: power frequency components and their harmonics, high-frequency noise, and mechanical vibration interference caused by circuit breaker operation. By employing a digital notch filter, such as an IIR bilinear notch filter, the 50Hz power frequency and its harmonics components can be filtered out. Alternatively, wavelet decomposition can be performed on the signal, and the wavelet coefficients can be processed using soft or hard thresholding methods to remove noise figures before reconstructing the signal. This achieves preprocessing for filtering and denoising the transient traveling wave signal. After preprocessing, a clean traveling wave signal sequence with a significantly improved signal-to-noise ratio is obtained.

[0030] Wavelet transform is a multi-scale time-frequency localization analysis method. It uses a scalable and translational basis function, the wavelet basis, to perform inner product operations with the original signal, enabling signal decomposition at different scales: at smaller scales, the wavelet is compressed, corresponding to the high-frequency component, achieving higher time resolution and locating abrupt changes in traveling wave surges; at larger scales, the wavelet is stretched, corresponding to the low-frequency component, achieving higher frequency resolution and analyzing the oscillation characteristics of the traveling wave signal. It achieves adaptive adjustment of the time-frequency window, making wavelet transform perfectly suited to the needs of traveling wave signal analysis. In practical processing, wavelet bases similar to the oscillation attenuation characteristics of traveling wave signals, such as Morlet wavelets or complex Gaussian wavelets, are selected to perform continuous wavelet transform on the preprocessed signal. The transform result is a two-dimensional coefficient matrix, where rows correspond to different frequency scales and columns correspond to different time points. The value of each element reflects the energy intensity of that frequency component at that moment. Presenting this coefficient matrix as an image—with time as the horizontal axis, frequency as the vertical axis, and energy intensity as the pixel grayscale value—results in a two-dimensional time-spectrum.

[0031] S32. Construct a weighting function matrix using a two-dimensional time-frequency spectrum diagram, and determine the velocity and direction of each energy point by calculating the optimized direction field of the energy points to form an energy flow distribution; In a preferred embodiment, the step of constructing a weighting function matrix using a two-dimensional time-frequency spectrum and determining the velocity and direction of each energy point by calculating the optimized direction field of the energy points to form an energy flow distribution includes the following steps: S321. Based on the energy amplitude of each pixel in the two-dimensional time-frequency spectrum and the distribution gradient of each pixel on the frequency axis, construct a composite weighting function matrix that integrates amplitude intensity and frequency change rate. It should be noted that the time-frequency spectrogram contains energy distribution information of the traveling wave signal, but the energy amplitude and frequency variation characteristics of different regions contribute differently to the subsequent directional field calculation. Specifically, for each pixel in the time-frequency spectrogram, its energy amplitude is first extracted, which reflects the energy intensity of the frequency component at that moment. Simultaneously, the gradient of that pixel along the frequency axis is calculated, i.e., the rate of energy change along the frequency direction. The larger the gradient value, the more drastic the energy change with frequency at that point. Based on these two types of information, a composite weighting function is constructed. : ; In the formula, and These represent the normalized energy amplitude and frequency gradient, respectively. and All represent weighting coefficients, typically taken as... .

[0032] S322. Based on the composite weight function matrix and combined with the covariance manifold method, calculate the initial direction field of energy flow; As a preferred embodiment, the calculation of the initial direction field of the energy flow based on the composite weight function matrix and combined with the covariance manifold method includes the following steps: S3221. Treat the composite weight function matrix as a two-dimensional image, and divide the two-dimensional image into several overlapping rectangular regions; for each rectangular region, extract the multi-dimensional features of each pixel within the rectangular region to form a feature vector set; It's important to note that the composite weight function matrix is ​​essentially a two-dimensional array. Its rows correspond to sampling points on the time axis, and its columns to sampling points on the frequency axis. The value of each matrix element is the composite weight function value for that pixel. Each element of the matrix corresponds to a pixel in the image, and the element's value corresponds to the pixel's grayscale value. Therefore, the composite weight function matrix can be directly viewed as a grayscale image, where brighter areas represent higher weight function values ​​(i.e., strong energy and dramatic frequency changes), and darker areas represent lower weight function values ​​(i.e., background noise or stable regions).

[0033] The composite weight function image is divided into several overlapping rectangular regions. The size of these rectangular regions is typically set to [value missing]. Pixels, such as 15×15 pixels, cover a certain time span and frequency range. Using overlapping partitions can ensure a smooth transition between regions, avoiding discontinuities or abrupt changes in subsequent orientation field calculations due to boundary effects. The overlap rate is preferably around 50%.

[0034] For each rectangular region, multidimensional features of each pixel within the region are extracted to form the feature vector set of that region. These multidimensional features include the pixel's energy amplitude, frequency axis gradient, temporal axis gradient, and local binary pattern texture features. S3222. Based on the feature vector set of each rectangular region, calculate the covariance matrix of all feature vectors within the rectangular region to obtain the covariance descriptor of the rectangular region. S3223. Iteratively solve the covariance descriptor of all rectangular regions to obtain the Karcher mean point used as a reference benchmark. It should be noted that since the covariance descriptors are all symmetric positive definite matrices, the space they form is not Euclidean space, but a Riemannian manifold. By solving for the Karcher mean point of the covariance descriptors for all rectangular regions—that is, the point on the manifold that minimizes the sum of the squared geodesic distances to all points—it can serve as a reference for subsequent unified mapping. The specific steps for solving for the Karcher mean point include: Initialize the mean point as any covariance descriptor, typically the covariance matrix of the first region. For each iteration, calculate the sum of the tangent vectors from the current mean point to all covariance descriptors, and update the mean point along this direction. Repeat the iteration until the change in the mean point between two consecutive iterations is less than a preset threshold, for example, a threshold of 10. -6 .

[0035] S3224. Using logarithmic mapping, the covariance descriptor of each rectangular region is projected onto the tangent space at the Karcher mean point to obtain the vector in the tangent space. It's important to note that the logarithmic mapping in Riemannian geometry can be used to project points on these manifolds onto the tangent space at the Karcher mean point, transforming the nonlinear problem into a linear one. Specifically, a Riemannian manifold is a curved space where the distance between covariance descriptors is defined by geodesics and cannot be directly measured by straight-line distance. The tangent space, on the other hand, is a linear approximation space at a point on the manifold; it is a regular Euclidean space where vectors can be added, subtracted, and linearly combined. The Karcher mean point, as the center of all covariance descriptors, has a tangent space that can be viewed as a unified workbench. After mapping covariance descriptors from different regions onto this workbench, linear algebraic methods can be used for processing.

[0036] The logarithmic mapping involves finding the shortest path (geodesic) from the Karcher mean point to the covariance descriptor for each rectangular region. The initial direction vector of this path at the Karcher mean point is then taken as the projection of the covariance descriptor onto the tangent space. This initial direction vector contains information about the change in the original covariance descriptor relative to the mean point, including the magnitude and direction of the change. After the logarithmic mapping, each symmetric positive definite matrix originally on the Riemannian manifold is transformed into a symmetric matrix in the tangent space, which can be expanded into a vector.

[0037] S3225. For each rectangular region, extract the components corresponding to the frequency gradient features and time gradient features from the tangent space vector, and construct the local gradient covariance matrix of the rectangular region; calculate the eigenvector corresponding to the largest eigenvalue of the gradient covariance matrix, and use its direction as the initial direction of energy flow at the center point of the rectangular region. It should be noted that the tangent space vector is composed of all the independent components of the covariance descriptor. For a vector derived from... d The covariance descriptor constructed from the dimensional features has a tangent space vector with dimension . In this invention, the multidimensional features of each pixel include four components: energy amplitude, frequency axis gradient, time axis gradient, and local binary pattern texture features. Therefore... d=4, and the dimension of the tangent space vector is 10.

[0038] These 10 components are arranged in a fixed order, corresponding to the upper or lower triangular elements of the covariance matrix. Specifically, they include: the variance of energy amplitude, the covariance of energy amplitude and frequency axis gradient, the covariance of energy amplitude and time axis gradient, the covariance of energy amplitude and local binary pattern texture features, the variance of frequency axis gradient, the covariance of frequency axis gradient and time axis gradient, the covariance of frequency axis gradient and local binary pattern texture features, the variance of time axis gradient, the covariance of time axis gradient and local binary pattern texture features, and the variance of local binary pattern texture features. Then, the components directly related to the energy flow direction are extracted from these 10 components. Specifically, the variance of the frequency axis gradient reflects the degree of energy change in the frequency direction, the variance of the time axis gradient reflects the degree of energy change in the time direction, and the covariance of the frequency axis gradient and time axis gradient reflects the coupling relationship between these two changes, i.e., the tilt of energy flow in the time-frequency plane.

[0039] After extracting these three key components from the tangent space vector, they are constructed into a 2×2 local gradient covariance matrix as follows: the variance of the frequency axis gradient is used as the first diagonal element of the matrix, the variance of the time axis gradient is used as the second diagonal element, and the covariance of both is used as the two off-diagonal elements. The constructed matrix describes the gradient variation characteristics of energy in the time-frequency plane within this rectangular region.

[0040] Furthermore, to determine the dominant direction of energy flow, eigenvalues ​​and corresponding eigenvectors are obtained by performing eigenvalue decomposition on the local gradient covariance matrix. In the eigenvalue decomposition results, the eigenvectors corresponding to larger eigenvalues ​​point to the direction of the most dramatic gradient change, which is the dominant direction of energy flow; the eigenvectors corresponding to smaller eigenvalues ​​point to secondary directions orthogonal to the dominant direction. The eigenvector itself is a two-dimensional vector containing frequency and time components, and its orientation angle can be calculated from the ratio of these two components. This orientation angle is used as the initial direction of energy flow at the center point of the rectangular region.

[0041] S3226. Smooth the initial energy flow direction at the center point of all regions to obtain the initial energy flow direction field.

[0042] It should be noted that the smoothing process employs anisotropic diffusion filtering, adaptively adjusting the smoothing intensity based on local gradient information. Specifically, the orientation angle of each center point is converted into a unit vector. For each pixel, a neighborhood window is defined centered on it, and the orientation vectors of all center points within the window are weighted and averaged. The weights are determined by both spatial distance and orientation similarity. The spatial distance weight is calculated based on the Euclidean distance between the neighborhood center points and the current pixel. The orientation similarity weight is calculated based on the difference between the original orientation of the neighborhood center points and the current iteration orientation of the current pixel. The final weight is the product of the spatial distance weight and the orientation similarity weight.

[0043] S323. By utilizing the range constraint of the propagation speed of the transient traveling wave signal in the time-frequency plane and the prior of frequency attenuation, the initial direction field of the energy flow is iteratively smoothed and corrected to obtain the optimized direction field of the energy point. As a preferred embodiment, the step of iteratively smoothing and correcting the initial direction field of the energy flow by utilizing the propagation speed range constraint of the transient traveling wave signal in the time-frequency plane and the frequency attenuation prior to obtain the optimized direction field of the energy point includes the following steps: S3231. Based on the type of circuit breaker line, determine the propagation characteristics of transient traveling wave signal in the line, determine the propagation speed range constraint and frequency attenuation prior of transient traveling wave signal in the time-frequency plane, and construct the direction field optimized energy functional according to the propagation speed range constraint and frequency attenuation prior. It should be noted that the propagation characteristics of transient traveling wave signals in the circuit can be determined based on the type and parameters of the circuit breaker line. These characteristics manifest in the time-frequency plane as two key priors: Due to propagation speed constraints, the propagation of traveling waves in the time-frequency plane is not arbitrary but limited by the actual physical velocity of the wave. The wave speed typically ranges from 0.95 times the speed of light to 1.0 times the speed of light or lower; this speed range translates into a reasonable range of variation in the energy flow direction angle on the time-frequency spectrum. Directions outside this range are considered physically unrealizable pseudo-directions.

[0044] Frequency attenuation is a priori; during the propagation of a traveling wave, due to the skin effect and dielectric loss of the line, high-frequency components attenuate faster than low-frequency components. This characteristic is manifested in the time-spectrum diagram as an overall trend of energy flow migrating from high frequencies to low frequencies, meaning the energy flow direction tends to point towards decreasing frequencies.

[0045] Specifically, based on the above priors, an optimized energy functional for the direction field can be constructed, consisting of three parts: ; In the formula, This is a data fidelity term, ensuring that the optimized orientation field maintains a certain degree of consistency with the initial orientation field; As a smoothing constraint, it ensures a smooth transition in the direction of adjacent pixels; As a physical prior, it penalizes abnormal directions that violate the propagation speed range and frequency decay laws; and To balance the weighting coefficients of each contribution, the optimal orientation field that conforms to both the data characteristics and physical laws can be obtained by minimizing this energy functional.

[0046] S3232, Iteratively solve the energy functional for the direction field optimization, and smooth and correct the initial direction field of the energy flow in each iteration; Specifically, the gradient descent method is used to iteratively solve the energy functional of the orientation field optimization. This includes: for the current orientation field... Calculate the gradient of the energy functional with respect to the orientation value of each pixel, according to Update the direction field, where, The step size factor controls the correction magnitude in each iteration. The updated orientation field may exceed the propagation speed range constraint, requiring projection onto the feasible region. For pixels exceeding the allowed angle range, their orientation is constrained to the nearest boundary value; for pixels violating the frequency decay trend, their orientation is appropriately adjusted to conform to prior laws. During iteration, the smoothing intensity is dynamically adjusted based on the local consistency of the current orientation field. In regions with high orientation consistency, such as continuous wave surge regions, smoothing is appropriately enhanced to suppress noise; in regions with abrupt orientation changes, such as wavefront boundaries, smoothing is weakened to preserve details.

[0047] S3233. When the change in the orientation field between two adjacent iterations is less than the preset convergence threshold or the preset maximum number of iterations is reached, the iteration is stopped, and the optimized orientation field of each energy point is obtained.

[0048] It should be noted that the difference between two adjacent iterations of the directional field is usually measured by the average or maximum value of the directional changes of all pixels.

[0049] Specifically, a maximum number of iterations, such as 100, can be set as a safety mechanism to prevent excessive computation time due to slow convergence. When the maximum number of iterations is reached, the iteration terminates regardless of whether the convergence threshold has been reached.

[0050] S324. Based on the optimized direction field of the energy points, the velocity of each energy point is estimated by calculating the rate of change of energy amplitude along the direction field, and the velocity is combined with the optimized direction field to obtain the velocity and direction of each energy point, so as to form an energy flow distribution.

[0051] It should be noted that for each pixel... Along its optimization direction Calculate the rate of change of energy amplitude. Direction The physical meaning of this direction on the time-frequency spectrum is the tangential direction of energy flow. The angle between this direction and the time axis determines the proportion of energy migration in both time and frequency dimensions. The rate of change of energy amplitude along this direction can be calculated using numerical differentiation. In practice, for each pixel, this direction is decomposed into a time axis component and a frequency axis component based on its optimized direction angle. These two components represent the projection proportions of energy flow in the time and frequency directions, respectively. Centered on the current pixel, neighboring pixels are selected along both the time and frequency axes. By calculating the energy amplitude difference between the current pixel and its neighbors, and combining this with the step size in each direction, the rate of change of energy amplitude along the time and frequency directions can be obtained. The rates of change in these two directions are then weighted and synthesized according to the projection proportions corresponding to the optimized direction angle, thus obtaining the rate of change of energy amplitude along the optimized direction itself.

[0052] Furthermore, after obtaining the optimized orientation field for each pixel, the motion velocity is estimated by calculating the rate of change of energy amplitude along that direction. Specifically, for each pixel, the rate of change of energy amplitude is calculated along its optimized direction using numerical differentiation. This rate of change reflects the attenuation or amplification rate of traveling wave energy propagating along that direction in the time-frequency plane, and is used as the estimated motion velocity for that point. This motion velocity is then combined with the corresponding optimized direction: the optimized direction determines the direction of energy flow, and the motion velocity determines the speed of energy flow; together, they constitute the energy flow vector for that point. By traversing all pixels in the entire time-frequency spectrum, each point is assigned an energy flow vector with a definite direction and magnitude; the set of these vectors forms the complete energy flow distribution.

[0053] S33. Based on the energy flow distribution and the sliding analysis window along the time axis, calculate the direction entropy and amplitude entropy of the energy flow vector in each window; combine the average movement velocity of the energy points in the window to construct the traveling wave disorder index, and combine it with the preset disorder threshold to determine the arrival time of the candidate traveling wave surge. It should be noted that for each window location, the energy flow vector of all pixels within that window is extracted. The directional distribution of the energy flow vectors within the window is statistically analyzed to calculate the directional entropy. The amplitude distribution of the energy flow vectors within the window is statistically analyzed to calculate the amplitude entropy. Then, the average velocity of all pixels within the window is calculated; this velocity reflects the average drastic degree of energy change in that region. Combining these three indicators, a traveling wave disorder index is constructed, typically expressed as the product of directional entropy, amplitude entropy, and average velocity. This index reaches its peak in regions where energy flow is both orderly and drastically changing, representing the characteristics of the arrival time of the traveling wave surge.

[0054] S34. Using the three-phase different opening and closing period parameters of the circuit breaker and the contact closing bounce time as prior constraints, error analysis is performed on the arrival time of candidate traveling wave surges, and the final set of traveling wave front arrival times is obtained through screening and interpolation.

[0055] As a preferred embodiment, the step of using the three-phase different opening and closing periodicity parameters of the circuit breaker and the contact closing bounce time as prior constraints to perform error analysis on the arrival times of candidate traveling wave surges, and obtaining the final set of traveling wave front arrival times through screening and interpolation processing, includes the following steps: S341. Arrange the candidate wave surge arrival times in chronological order and calculate the pairwise time difference between each phase. If the time difference between any two phases exceeds the allowable range of the three-phase opening and closing asynchrony parameter of the circuit breaker, it is judged as abnormal data and deleted to obtain the candidate times after preliminary screening. It should be noted that all candidate surge arrival times are arranged in chronological order and labeled with their respective phases, including phases A, B, and C. Then, the pairwise time differences between each phase are calculated, including the difference between adjacent times within the same phase and the difference between different phases. If the time difference between any two phases exceeds the allowable range for asynchrony of three-phase opening and closing specified in the circuit breaker's technical parameters—for example, if one phase significantly lags behind or leads the other two phases—then that time is considered abnormal data. This abnormal data may be caused by noise interference, external transient disturbances, or abnormal signal acquisition, rather than a genuine fault surge. After removing these abnormal data, a preliminary set of candidate times is obtained.

[0056] S342. Starting from each candidate moment after initial screening, extend the duration corresponding to the contact closing bounce time parameter forward and backward as the failure-to-operate time window; if multiple candidate moments appear in the same failure-to-operate time window, retain the moment with the largest energy amplitude as the effective wavehead, and the rest are judged as pseudo waveheads generated by mechanical bounce and removed, to obtain the wavehead moment after secondary screening. It should be noted that during the closing operation of a circuit breaker, the contacts may experience brief mechanical bounce, typically ≤2ms. This bounce can cause multiple pre-breakdowns or re-breakdowns in the contact gap, resulting in a series of traveling wave pulses with extremely short time intervals. These pulses appear as multiple closely adjacent candidate moments in the time domain, but only the first moment that truly represents contact closure or opening is a valid traveling wave surge; subsequent bounce pulses are pseudo-waveheads generated by mechanical characteristics. Specifically, starting from each candidate moment after initial screening, the duration corresponding to the contact closing bounce time parameter is extended forward and backward, for example, 1ms forward and 1ms backward, with a total window length of 2ms, forming a failure-to-operate time window. If multiple candidate moments appear within the same failure-to-operate time window, the moment with the largest energy amplitude is retained as the valid wavehead.

[0057] S343. Based on the tower coordinates and line length parameters in the line topology library, calculate the electrical distance corresponding to the time difference between adjacent wavefront times. If the electrical distance has an integer multiple relationship with the line length or tower spacing, it is determined to be a reflected wave or a refracted wave and retained; otherwise, it is discarded as noise interference, and the wavefronts after three screenings are obtained. It should be noted that during the propagation of traveling wave signals, in addition to the initial wavefront generated at the fault point, they will also encounter points of impedance discontinuity in the line, such as the opposite bus, branch points, and towers, generating reflected or refracted waves. Redundant information is introduced into the candidate time points. Specifically, for two adjacent wavefront times, their time difference is calculated and converted into the corresponding electrical distance by combining it with the current dynamic wave velocity of the line. Then, the length parameters of the line segment in the line topology database are queried, such as the spacing between adjacent towers and the total line length, to determine whether the electrical distance has an integer multiple relationship with the line length or tower spacing. If an integer multiple relationship exists, the wavefront is determined to be a reflected or refracted wave from the far end of the line or a specific reflection point and is retained for subsequent ranging analysis; otherwise, it is discarded as unexplained noise interference. After the above processing, the wavefront times after three rounds of filtering are obtained.

[0058] S344. For each wavefront time after three screenings, extract the original waveform segments of preset length before and after it, and use spline interpolation to perform local fitting. Take the position with the largest energy change rate as the final arrival time of the traveling wavefront, and output the complete set of traveling wavefront arrival times.

[0059] It should be noted that the accuracy of the candidate wavefront times obtained after the above multiple rounds of screening is still limited by the center position of the sliding window and the sampling interval, which cannot meet the requirements of high-precision fault location. Subsampling point level localization of wavefront times is achieved using the original waveform information. Specifically, for each wavefront time after three rounds of screening, a pre-defined length of original waveform segment before and after that time (e.g., 5 μs before and 10 μs after) is extracted from the acquired original traveling wave data. Then, cubic spline interpolation is used to locally fit this waveform segment. Cubic spline interpolation can construct a smooth and continuous cubic polynomial curve between sampling points, thereby reconstructing the continuous change shape of the waveform. On the interpolated and reconstructed continuous waveform, the rate of energy change at each point is calculated, i.e., the waveform slope or first derivative. The position with the largest rate of energy change is found, which is the true arrival time of the traveling wave surge. By performing precise localization on all screened wavefront times, a complete set of traveling wavefront arrival times containing precise arrival times can be finally output.

[0060] S4. Based on the arrival time set of the traveling wave front, use the time difference positioning algorithm for two-way ranging to determine the propagation path and reflection point of the fault traveling wave, and combine it with the line topology library to calculate the geographical coordinates of the fault point and the tower section to which it belongs. As a preferred embodiment, the step of determining the propagation path and reflection point of the fault traveling wave based on the arrival time set of the traveling wave wavehead using a two-way ranging time difference localization algorithm, and calculating the geographical coordinates of the fault point and its corresponding tower section by combining the line topology library, includes the following steps: S41. Based on the arrival time set of the traveling wave front, extract the arrival time of the initial traveling wave surge and the time of the reflected wave at the fault point, and determine the wave velocity of the transient traveling wave signal by combining the line topology library. As a preferred embodiment, determining the wave velocity of the transient traveling wave signal by combining the line topology library includes the following steps: Extract information on all known lengths of intact line sections without faults from the line topology library, and extract the arrival time of the traveling wave front generated by the last operation of the circuit breakers at both ends of each line section before the fault from the traveling wave front arrival time set. It should be noted that the line topology library stores segmented parameters for the entire line, including the precise length of each segment after real-time sag correction, the circuit breaker numbers at both ends, and the line type. All known-length, currently fault-free, healthy line segments can be extracted from the library. These segments were in a healthy state before the fault occurred, and the traveling wave signals generated by the circuit breakers at both ends during historical operations can be used for wave velocity inversion. Simultaneously, the arrival times of the traveling wave fronts generated by the last operation of the circuit breakers at both ends of these healthy segments before the fault can be extracted from the set of traveling wave front arrival times.

[0061] For each intact line segment, the time difference is calculated based on the arrival time of the traveling wave fronts of the circuit breakers at both ends; the length of the line segment in the line topology library is used to calculate the candidate value of dynamic wave velocity under the current environment. It should be noted that for each selected intact line segment, its length is assumed to be... The arrival times of the traveling wave fronts of the circuit breakers at both ends are respectively and The earlier one is the starting end, and the later one is the ending end. The time difference for the traveling wave to propagate from one end of the segment to the other is... According to the basic principles of traveling wave propagation, there is a definite relationship between this time difference and the segment length: for one-way propagation, the wave speed... If self-verification is performed using reflected waves, then the following can be adopted: Therefore, the candidate dynamic wave velocity values ​​for this section under the current environment are calculated. .

[0062] The consistency of the candidate dynamic wave velocity values ​​calculated from multiple intact line sections is checked, and after removing outliers, a weighted average is taken to obtain the final dynamic wave velocity of the current line transient traveling wave signal.

[0063] Specifically, the consistency test employs statistical methods, such as the 3σ criterion, to identify and eliminate outliers. For example, the mean and standard deviation of all candidate values ​​are calculated, and candidate values ​​exceeding the mean ± 3 times the standard deviation are identified as outliers and eliminated. For the remaining valid candidate values, a weighted average is used to calculate the final dynamic wave velocity, with the weights typically taken as the length of each segment or the signal-to-noise ratio.

[0064] It should be noted that when there are no intact sections available in the line or the number of valid candidate values ​​is insufficient, such as less than 3, a reliable dynamic wave velocity cannot be obtained through online inversion. In this case, the system automatically switches to the backup scheme: it reads the theoretical wave velocity preset from the line topology library based on the line type, conductor material, and ambient temperature as the initial wave velocity.

[0065] S42. Based on the arrival time of the initial traveling wave surge, the time of the reflected wave at the fault point, and the wave velocity of the transient traveling wave signal, calculate the equivalent distance between the circuit breaker and the fault point. It should be noted that for each circuit breaker, two key moments are extracted from the arrival times of the traveling wave front: the arrival time of the initial traveling wave surge. That is, the moment when the traveling wave generated by the fault first arrives at the circuit breaker, and the moment when the reflected wave from the fault point arrives. This refers to the time it takes for the traveling wave to reach the circuit breaker after being reflected from the fault point. The time difference between these two moments is... This represents the round-trip propagation time of the traveling wave from the circuit breaker to the fault point and back to the circuit breaker. Combined with a determined dynamic wave velocity... The equivalent distance between the circuit breaker and the fault point is calculated according to the principle of bidirectional ranging. This calculation is based on the physical fact that the one-way distance of a traveling wave from a circuit breaker to a fault point is exactly half the round-trip distance. In this way, each circuit breaker obtains an equivalent distance observation from the fault point.

[0066] S43. Take the circuit breaker that first receives the traveling wave surge as the reference station, and combine the location information of the circuit breaker to establish a linear equation for the coordinates of the fault point. In a preferred embodiment, the step of using the circuit breaker that first receives the traveling wave surge as a reference station and establishing a linear equation for the fault point coordinates based on the circuit breaker's location information includes the following steps: S431. Based on the arrival time set of the traveling wave front and the equivalent distance, the circuit breaker that first receives the traveling wave surge is determined as the reference station. It should be noted that in multi-station positioning, selecting a reference station can convert absolute distance measurement into relative distance difference measurement, thereby eliminating common systematic errors.

[0067] S432. For each other circuit breaker, extract the coordinates from the line topology library and calculate the reference station distance difference by combining them with the equivalent distance of the reference station. It should be noted that for each other circuit breaker except the reference station... First, extract its coordinates from the line topology library. And obtain its equivalent distance from the calculation results of S42. Then calculate the equivalent distance difference between the circuit breaker and the reference station. This distance difference reflects the difference in distance from the fault point to the two circuit breakers, corresponding to the TDOA measurement value in the Time Difference of Arrival (TDOA) location principle.

[0068] S433. Based on the principle of time of arrival difference positioning, the quadratic term is eliminated by the square subtraction method, and a linear equation about the coordinates of the fault point is established by combining the coordinates of each circuit breaker and the distance difference of the reference station.

[0069] Specifically, based on the time-of-arrival (TOA) positioning principle, the difference between the distance from the fault point to any non-reference station circuit breaker and the distance to the reference station circuit breaker should be equal to the equivalent distance difference. Geometrically, this describes a hyperboloid with reference and non-reference stations as foci. Its mathematical expression contains two square root terms of distance, making it a typical nonlinear equation that is difficult to solve directly. Therefore, this invention employs a square subtraction method for linearization. Specifically, this involves squaring both sides of the equation containing square roots to eliminate the radical sign; then, using the common term resulting from the square expansion, the square of the reference distance, and subtracting the equations to cancel out the square term of the reference distance; after this process, the original nonlinear equation is transformed into a linear combination of the fault point coordinates and the reference distance. Each non-reference station can establish an equation of the form "a linear combination of the coordinate difference and the unknown coordinates plus the product of the distance difference and the reference distance equals a constant." In this linear equation, the three coordinate components of the fault point and the four unknowns (the reference distance) all appear linearly. Multiple circuit breakers can establish multiple linear equations, forming a system of linear equations for subsequent solution.

[0070] S44. Solve the linear equation between the fault point coordinates and the reference distance to obtain the preliminary geographical coordinates of the fault point. Combine this with the line topology library to determine the geographical coordinates of the fault point and the tower section to which it belongs.

[0071] It should be noted that the established system of linear equations typically contains four unknowns and three-dimensional coordinates. and reference distance The system of equations consists of multiple equations. When the number of equations exceeds the number of unknowns, the system is overdetermined and has no strictly exact solution. This invention uses the least squares method to solve the system, that is, to find the solution that minimizes the sum of squared residuals of all equations. The least squares method obtains the optimal estimate by solving the normal equations or by using matrix pseudo-inverses. When the number of equations is exactly equal to the number of unknowns, linear algebra methods can be used to solve the system directly. The solution yields the preliminary geographical coordinates of the fault point, which are spatial location estimates based on traveling wave measurements and geometric calculations.

[0072] Furthermore, the initial geographic coordinates may deviate from the actual centerline of the line due to measurement or calculation errors. Considering that the fault point must be located on the line conductor, the initial coordinates are projected onto the line centerline in a constructed line topology library. The coordinate sequences of all adjacent towers near the fault point are extracted from the line topology library, and the spatial curve of the line centerline for that section is fitted. Then, the perpendicular distance from the initial coordinates to this centerline is calculated to obtain the coordinates of the perpendicular foot point, which is the more physically accurate location of the fault point. Based on this, the tower section to which the fault point belongs is determined by comparing the distance along the line between the perpendicular foot point and the coordinates of each tower.

[0073] S5. Based on the calculated geographical coordinates of the fault point and its corresponding tower section, calculate the electrical distance of the fault point along the line relative to the preset reference point to obtain the final fault distance measurement result.

[0074] It should be noted that by using coordinate projection and tower segment matching, the fault point is located to a specific line location, that is, it is determined which two adjacent towers the fault point is located between, and its projected position within that segment. However, for power system operation and maintenance personnel, a more intuitive and commonly used way to describe the fault location is the electrical distance along the line relative to a preset reference point, such as a substation outlet, the line starting point, or a specific important tower. Specifically, relevant information about the preset reference point is collected, including the location of the substation busbar, which is usually chosen as the line starting point, or a benchmark tower, with its mileage along the line being 0. Simultaneously, the precise coordinates of the towers at both ends of the tower segment to which the fault point belongs, as well as the precise length of the segment after real-time sag correction, are extracted from the line topology database. Based on the determined projected position of the fault point, its distance along the line from the starting tower of the segment is calculated. This distance can be obtained through spatial interpolation between the coordinates of the projection point and the coordinates of the towers at both ends, or a more precise calculation of the length along the line can be performed using the sag curve parameters at the projection point. The cumulative distance is obtained by accumulating the sag-corrected lengths of all complete segments from the preset reference point to the starting tower of the fault location segment. Adding this cumulative distance to the distance within each segment yields the electrical distance of the fault location along the line relative to the preset reference point. This can be output as the final fault location result. The output information typically includes: the electrical distance of the fault location relative to the preset reference point, the tower segment to which the fault location belongs, the specific location of the fault location within that segment, and the determined geographical coordinates of the fault location, etc.

[0075] like Figure 2 As shown, according to an embodiment of the present invention, a fault location system based on circuit breaker traveling wave signals is provided, the system comprising: Data acquisition module 1 is used to collect point cloud data of circuit breaker lines and generate line point cloud models through three-dimensional reconstruction; it uses the line point cloud model to extract tower coordinates and conductor sag status, and constructs a line topology library containing tower coordinates and conductor sag. Signal acquisition module 2 is used to acquire transient traveling wave signals generated during the opening or closing operation of the circuit breaker based on the current transformers pre-installed on the circuit breaker, and to record the arrival time of the transient traveling wave signal of each circuit breaker. The time analysis module 3 is used to preprocess the transient traveling wave signal by filtering and denoising, and then to perform wavelet transform processing on the preprocessed transient traveling wave signal to obtain a two-dimensional time spectrum. A weight function matrix is ​​constructed using the two-dimensional time spectrum, and the velocity and direction of each energy point are determined by calculating the optimized direction field of the energy points to form an energy flow distribution. Based on the energy flow distribution and a sliding analysis window along the time axis, the directional entropy and amplitude entropy of the energy flow vector within each window are calculated. A traveling wave disorder index is constructed by combining the average velocity of the energy points within the window, and a preset disorder threshold is used to determine the arrival times of candidate traveling wave surges. Using the three-phase different opening and closing periodicity parameters of the circuit breaker and the contact closing bounce time as prior constraints, error analysis is performed on the arrival times of the candidate traveling wave surges, and the final set of traveling wave front arrival times is obtained through screening and interpolation processing. The coordinate calculation module 4 is used to determine the propagation path and reflection point of the fault traveling wave based on the arrival time set of the traveling wave wavehead, using a two-way ranging time difference positioning algorithm, and combined with the line topology library, to calculate the geographical coordinates of the fault point and the tower section to which it belongs. The ranging result calculation module 5 is used to calculate the electrical distance of the fault point along the line relative to the preset reference point based on the calculated geographical coordinates of the fault point and the tower section to which it belongs, and to obtain the final fault ranging result.

[0076] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, etc.) containing computer-usable program code.

[0077] 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. 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.

Claims

1. A fault location method based on circuit breaker traveling wave signals, characterized in that, Includes the following steps: S1. Collect point cloud data of circuit breaker lines and generate line point cloud models through 3D reconstruction; use the line point cloud models to extract tower coordinates and conductor sag status, and construct a line topology library containing tower coordinates and conductor sag. S2. Based on the current transformer pre-installed on the circuit breaker, collect the transient traveling wave signal generated during the opening or closing operation of the circuit breaker. S3. Perform filtering and denoising preprocessing on the transient traveling wave signal, and then perform wavelet transform processing on the preprocessed transient traveling wave signal to obtain a two-dimensional time spectrum; construct a weight function matrix using the two-dimensional time spectrum, and determine the motion velocity and direction of each energy point by calculating the optimized direction field of the energy points to form an energy flow distribution; calculate the direction entropy and amplitude entropy of the energy flow vector in each window based on the energy flow distribution and the sliding analysis window along the time axis. By combining the average movement velocity of energy points within the window, a traveling wave disorder index is constructed, and by combining the preset disorder threshold, the arrival time of candidate traveling wave surges is determined. Using the three-phase different opening and closing periodicity parameters of the circuit breaker and the contact closing bounce time as prior constraints, error analysis is performed on the arrival time of candidate traveling wave surges, and the final set of traveling wave front arrival times is obtained through screening and interpolation. S4. Based on the arrival time set of the traveling wave front, use the time difference positioning algorithm for two-way ranging to determine the propagation path and reflection point of the fault traveling wave, and combine it with the line topology library to calculate the geographical coordinates of the fault point and the tower section to which it belongs. S5. Based on the calculated geographical coordinates of the fault point and its corresponding tower section, calculate the electrical distance of the fault point along the line relative to the preset reference point to obtain the final fault distance measurement result.

2. The fault location method based on circuit breaker traveling wave signal according to claim 1, characterized in that, The process of constructing a weighting function matrix using a two-dimensional time-frequency spectrum and determining the velocity and direction of each energy point by calculating the optimized direction field of the energy points to form an energy flow distribution includes the following steps: S321. Based on the energy amplitude of each pixel in the two-dimensional time-frequency spectrum and the distribution gradient of each pixel on the frequency axis, construct a composite weighting function matrix that integrates amplitude intensity and frequency change rate. S322. Based on the composite weight function matrix and combined with the covariance manifold method, calculate the initial direction field of energy flow; S323. By utilizing the range constraint of the propagation speed of the transient traveling wave signal in the time-frequency plane and the prior of frequency attenuation, the initial direction field of the energy flow is iteratively smoothed and corrected to obtain the optimized direction field of the energy point. S324. Based on the optimized direction field of the energy points, the velocity of each energy point is estimated by calculating the rate of change of energy amplitude along the direction field, and the velocity is combined with the optimized direction field to obtain the velocity and direction of each energy point, so as to form an energy flow distribution.

3. The fault location method based on circuit breaker traveling wave signal according to claim 2, characterized in that, The calculation of the initial direction field of energy flow based on the composite weight function matrix and combined with the covariance manifold method includes the following steps: S3221. Treat the composite weight function matrix as a two-dimensional image, and divide the two-dimensional image into several overlapping rectangular regions; for each rectangular region, extract the multi-dimensional features of each pixel within the rectangular region to form a feature vector set; S3222. Based on the feature vector set of each rectangular region, calculate the covariance matrix of all feature vectors within the rectangular region to obtain the covariance descriptor of the rectangular region. S3223. Iteratively solve the covariance descriptor of all rectangular regions to obtain the Karcher mean point used as a reference benchmark. S3224. Using logarithmic mapping, the covariance descriptor of each rectangular region is projected onto the tangent space at the Karcher mean point to obtain the vector in the tangent space. S3225. For each rectangular region, extract the components corresponding to the frequency gradient features and time gradient features from the tangent space vector, and construct the local gradient covariance matrix of the rectangular region; calculate the eigenvector corresponding to the largest eigenvalue of the gradient covariance matrix, and use its direction as the initial direction of energy flow at the center point of the rectangular region. S3226. Smooth the initial energy flow direction at the center point of all regions to obtain the initial energy flow direction field.

4. The fault location method based on circuit breaker traveling wave signal according to claim 3, characterized in that, The process of using the propagation speed range constraint of the transient traveling wave signal in the time-frequency plane and the frequency attenuation prior to iteratively smooth and correct the initial direction field of the energy flow to obtain the optimized direction field of the energy point includes the following steps: S3231. Based on the type of circuit breaker line, determine the propagation characteristics of transient traveling wave signal in the line, determine the propagation speed range constraint and frequency attenuation prior of transient traveling wave signal in the time-frequency plane, and construct the direction field optimized energy functional according to the propagation speed range constraint and frequency attenuation prior. S3232, Iteratively solve the energy functional for the direction field optimization, and smooth and correct the initial direction field of the energy flow in each iteration; S3233. When the change in the orientation field between two adjacent iterations is less than the preset convergence threshold or the preset maximum number of iterations is reached, the iteration is stopped, and the optimized orientation field of each energy point is obtained.

5. The fault location method based on circuit breaker traveling wave signal according to claim 4, characterized in that, The method of using the three-phase different opening and closing periodic parameters of the circuit breaker and the contact closing bounce time as a priori constraints to perform error analysis on the arrival times of candidate traveling wave surges, and obtaining the final set of traveling wave front arrival times through screening and interpolation processing, includes the following steps: S341. Arrange the candidate wave surge arrival times in chronological order and calculate the pairwise time difference between each phase. If the time difference between any two phases exceeds the allowable range of the three-phase opening and closing asynchrony parameter of the circuit breaker, it is judged as abnormal data and deleted to obtain the candidate times after preliminary screening. S342. Starting from each candidate moment after initial screening, extend the duration corresponding to the contact closing bounce time parameter forward and backward as the failure-to-operate time window; if multiple candidate moments appear in the same failure-to-operate time window, retain the moment with the largest energy amplitude as the effective wavehead, and the rest are judged as pseudo waveheads generated by mechanical bounce and removed, to obtain the wavehead moment after secondary screening. S343. Based on the tower coordinates and line length parameters in the line topology library, calculate the electrical distance corresponding to the time difference between adjacent wavefront times. If the electrical distance has an integer multiple relationship with the line length or tower spacing, it is determined to be a reflected wave or a refracted wave and retained; otherwise, it is discarded as noise interference, and the wavefronts after three screenings are obtained. S344. For each wavefront time after three screenings, extract the original waveform segments of preset length before and after it, and use spline interpolation to perform local fitting. Take the position with the largest energy change rate as the final arrival time of the traveling wavefront, and output the complete set of traveling wavefront arrival times.

6. The fault location method based on circuit breaker traveling wave signal according to claim 1, characterized in that, The process of determining the propagation path and reflection point of the fault traveling wave based on the arrival time set of the traveling wave front, using a two-way ranging time difference localization algorithm, and calculating the geographical coordinates of the fault point and its corresponding tower section by combining the line topology library includes the following steps: S41. Based on the arrival time set of the traveling wave front, extract the arrival time of the initial traveling wave surge and the time of the reflected wave at the fault point, and determine the wave velocity of the transient traveling wave signal by combining the line topology library. S42. Based on the arrival time of the initial traveling wave surge, the time of the reflected wave at the fault point, and the wave velocity of the transient traveling wave signal, calculate the equivalent distance between the circuit breaker and the fault point. S43. Take the circuit breaker that first receives the traveling wave surge as the reference station, and combine the location information of the circuit breaker to establish a linear equation for the coordinates of the fault point. S44. Solve the linear equation between the fault point coordinates and the reference distance to obtain the preliminary geographical coordinates of the fault point. Combine this with the line topology library to determine the geographical coordinates of the fault point and the tower section to which it belongs.

7. The fault location method based on circuit breaker traveling wave signal according to claim 6, characterized in that, The determination of the wave velocity of the transient traveling wave signal by combining the line topology library includes the following steps: Extract information on all known lengths of intact line sections without faults from the line topology library, and extract the arrival time of the traveling wave front generated by the last operation of the circuit breakers at both ends of each line section before the fault from the traveling wave front arrival time set. For each intact line segment, the time difference is calculated based on the arrival time of the traveling wave fronts of the circuit breakers at both ends; the length of the line segment in the line topology library is used to calculate the candidate value of dynamic wave velocity under the current environment. The consistency of the candidate dynamic wave velocity values ​​calculated from multiple intact line sections is checked, and after removing outliers, a weighted average is taken to obtain the final dynamic wave velocity of the current line transient traveling wave signal.

8. The fault location method based on circuit breaker traveling wave signal according to claim 6, characterized in that, The process of using the circuit breaker that first receives the traveling wave surge as a reference station and combining it with the circuit breaker's location information to establish a linear equation for the fault point coordinates includes the following steps: S431. Based on the arrival time set of the traveling wave front and the equivalent distance, the circuit breaker that first receives the traveling wave surge is determined as the reference station. S432. For each other circuit breaker, extract the coordinates from the line topology library and calculate the reference station distance difference by combining them with the equivalent distance of the reference station. S433. Based on the principle of time of arrival difference positioning, the quadratic term is eliminated by the square subtraction method, and a linear equation about the coordinates of the fault point is established by combining the coordinates of each circuit breaker and the distance difference of the reference station.

9. A fault location system based on circuit breaker traveling wave signals, used to implement the fault location method based on circuit breaker traveling wave signals as described in any one of claims 1-8, characterized in that, The system includes: The data acquisition module is used to collect point cloud data of circuit breaker lines and generate line point cloud models through 3D reconstruction; it uses the line point cloud models to extract tower coordinates and conductor sag status, and constructs a line topology library containing tower coordinates and conductor sag. The signal acquisition module is used to acquire transient traveling wave signals generated during the opening or closing operation of the circuit breaker based on the current transformers pre-installed on the circuit breaker, and to record the arrival time of the transient traveling wave signal of each circuit breaker. The time analysis module is used for preprocessing transient traveling wave signals by filtering and denoising, and then performing wavelet transform on the preprocessed transient traveling wave signals to obtain a two-dimensional time spectrum. A weighting function matrix is ​​constructed using the two-dimensional time spectrum, and the velocity and direction of each energy point are determined by calculating the optimized direction field of the energy points to form an energy flow distribution. Based on the energy flow distribution and a sliding analysis window along the time axis, the directional entropy and amplitude entropy of the energy flow vector within each window are calculated. A traveling wave disorder index is constructed by combining the average velocity of the energy points within the window, and a preset disorder threshold is used to determine the arrival times of candidate traveling wave surges. Using the three-phase different opening and closing periodicity parameters of the circuit breaker and the contact closing bounce time as prior constraints, error analysis is performed on the arrival times of the candidate traveling wave surges, and the final set of traveling wave front arrival times is obtained through screening and interpolation. The coordinate calculation module is used to determine the propagation path and reflection point of the fault traveling wave based on the arrival time set of the traveling wave wavehead, using a two-way ranging time difference positioning algorithm, and combined with the line topology library, to calculate the geographical coordinates of the fault point and the tower section to which it belongs. The distance measurement result calculation module is used to calculate the electrical distance of the fault point along the line relative to the preset reference point based on the calculated geographical coordinates of the fault point and the tower section to which it belongs, and to obtain the final fault distance measurement result.

Images