Digital Distribution Network Fault Traveling Wave Distance Measuring Instrument and Method

Abstract

This application relates to the technical field of fault location. Specifically, it discloses a digital distribution network fault traveling wave distance measuring instrument and method. It uses fault traveling wave acquisition devices deployed at various key nodes of the distribution network to collect fault traveling wave signals in real time, and obtains stable line-mode traveling wave signals through signal filtering and line-mode component extraction. Then, it introduces a signal processing algorithm based on deep learning to perform time-domain feature analysis on the line-mode traveling wave signals of each key node to detect the wavehead position of the signal. Furthermore, by comparing the signal arrival times of each key node, it determines the section of the distribution network where the fault point is located, and based on the double-ended traveling wave distance measurement algorithm, according to the signal arrival time difference between the two ends of this section and combined with the signal propagation speed, it calculates the distances between the fault point and the two end nodes. In this way, through in-depth time-domain analysis of the fault traveling wave signals, the wavehead position of the signals can be accurately identified, improving the reliability of distribution network fault traveling wave distance measurement.

Description

Technical Field

[0001] This application relates to the technical field of fault location, and more specifically, to a digital distribution network fault traveling wave distance measuring instrument and method. Background Art

[0002] As a key link connecting the power system and the user side, the operation reliability of the distribution network directly affects the power supply quality and user satisfaction. With the continuous expansion of the scale of the power system, the distribution network has a complex topology (multiple branches and nodes), strong heterogeneity of line parameters, and is often affected by factors such as the access of distributed power sources and the impact of non-linear loads, resulting in frequent faults. Traditional fault location methods (such as impedance method, traveling wave method, etc.) are relatively mature in the transmission network, but face significant challenges when applied to the distribution network.

[0003] Specifically, the impedance method determines the fault location by measuring the impedance of the fault line. However, the distribution network has a complex structure, numerous branches, and the line parameters are easily affected by the operating environment, such as changes in temperature and humidity, resulting in large impedance calculation errors and difficulty in accurately locating the fault point. In addition, when there are multiple fault points or transition resistances in the distribution network, the measurement accuracy of the impedance method will be further reduced. The traveling wave ranging technology is based on the propagation characteristics of the traveling wave generated by the fault on the transmission line, and calculates the fault location by measuring the time difference of the traveling wave arriving at different measurement points. Compared with traditional methods, the traveling wave ranging technology has higher accuracy and faster response speed, can quickly locate the fault point, shorten the power outage time, and improve the power supply reliability.

[0004] However, due to the short length and numerous branches of the distribution network lines, the fault traveling wave is easily affected by factors such as discontinuous line impedance and equipment reflection during the propagation process, resulting in rapid signal amplitude attenuation and severe waveform distortion, making it difficult for traditional traveling wave detection methods based on set thresholds to accurately identify the signal wavefront, and further generating time measurement errors, which affect the accuracy of fault location.

[0005] Therefore, an optimized digital distribution network fault traveling wave distance measuring instrument and method are expected. Summary of the Invention

[0006] To solve the above technical problems, the present application is proposed. Embodiments of the present application provide a digital distribution network fault traveling wave ranging device and method, which uses fault traveling wave acquisition devices deployed at various key nodes of the distribution network to collect fault traveling wave signals in real time, and obtains stable line-mode traveling wave signals through signal filtering and line-mode component extraction. Then, a signal processing algorithm based on deep learning is introduced to perform time-domain feature analysis on the line-mode traveling wave signals of each key node to detect the wavehead position of the signal, and accordingly mark the time when the fault traveling wave signal arrives at the acquisition device. Furthermore, by comparing the signal arrival times of each key node, the distribution network section where the fault point is located is determined, and based on the double-end traveling wave ranging algorithm, according to the signal arrival time difference between the two ends of the section and combined with the signal propagation speed, the distances between the fault point and the two end nodes are calculated. This method can accurately identify the wavehead position of the signal through in-depth time-domain analysis of the fault traveling wave signal, improving the reliability of distribution network fault traveling wave ranging.

[0007] According to one aspect of the present application, there is provided a digital distribution network fault traveling wave ranging method, which includes:

[0008] Using fault traveling wave acquisition devices deployed at various key nodes of the distribution network to collect fault traveling wave signals in real time, where the key nodes include all branch nodes of the distribution network and both ends of the main line;

[0009] Filtering and extracting line-mode components from the fault traveling wave signals of each key node to obtain a set of line-mode traveling wave signals;

[0010] Performing wavehead detection based on deep learning on each line-mode traveling wave signal in the set of line-mode traveling wave signals and determining the wavehead arrival time to obtain a set of key node fault traveling wave arrival times;

[0011] Processing the set of key node fault traveling wave arrival times based on a section location algorithm to determine the distribution network section where the fault point is located between two adjacent key nodes;

[0012] Processing the fault traveling wave arrival times of the two adjacent key nodes based on the double-end traveling wave ranging algorithm to determine the distances between the fault point and the two adjacent key nodes.

[0013] According to another aspect of the present application, there is provided a digital distribution network fault traveling wave ranging device, which includes:

[0014] A fault traveling wave signal acquisition module, configured to use fault traveling wave acquisition devices deployed at various key nodes of the distribution network to collect fault traveling wave signals in real time, where the key nodes include all branch nodes of the distribution network and both ends of the main line;

[0015] The fault traveling wave signal preprocessing module is used to filter the fault traveling wave signals of each key node and extract the line mode components to obtain a set of line mode traveling wave signals;

[0016] The wavefront detection module is used to perform wavefront detection based on deep learning on each line mode traveling wave signal in the set of line mode traveling wave signals and determine the wavefront arrival time to obtain a set of fault traveling wave arrival times of key nodes;

[0017] The distribution network section determination module is used to process the set of fault traveling wave arrival times of key nodes based on the section location algorithm to determine the distribution network section where the fault point is located between two adjacent key nodes;

[0018] The double - ended traveling wave ranging module is used to process the fault traveling wave arrival times of the two adjacent key nodes based on the double - ended traveling wave ranging algorithm to determine the distances between the fault point and the two adjacent key nodes.

[0019] Compared with the prior art, the digital distribution network fault traveling wave ranging instrument and method provided by this application utilize the fault traveling wave acquisition devices deployed at each key node of the distribution network to collect fault traveling wave signals in real time, and obtain stable line mode traveling wave signals through signal filtering and line mode component extraction. Then, a signal processing algorithm based on deep learning is introduced to perform time - domain feature analysis on the line mode traveling wave signals of each key node to detect the wavefront position of the signal, thereby marking the time when the fault traveling wave signal arrives at the acquisition device. Furthermore, by comparing the signal arrival times of each key node, the distribution network section where the fault point is located is determined, and based on the double - ended traveling wave ranging algorithm, according to the signal arrival time difference between the two end nodes of this section and combined with the signal propagation speed, the distances between the fault point and the two end nodes are calculated. This method can accurately identify the wavefront position of the signal through in - depth time - domain analysis of the fault traveling wave signals, improving the reliability of distribution network fault traveling wave ranging. Description of the Drawings

[0020] By describing the embodiments of the present application in more detail in combination with the drawings, the above - mentioned and other objects, features, and advantages of the present application will become more obvious. The drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification. They are used to explain the present application together with the embodiments of the present application, and do not constitute a limitation to the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0021] Figure 1 It is a flowchart of the digital distribution network fault traveling wave ranging method according to the embodiment of the present application.

[0022] Figure 2 It is a schematic diagram of data flow of the digital distribution network fault traveling wave ranging method according to the embodiment of the present application.

[0023] Figure 3 It is a flowchart of sub-step S3 of the digital distribution network fault traveling wave ranging method according to an embodiment of the present application.

[0024] Figure 4 It is a flowchart of sub-step S32 of the digital distribution network fault traveling wave ranging method according to an embodiment of the present application.

[0025] Figure 5 It is a flowchart of sub-step S323 of the digital distribution network fault traveling wave ranging method according to an embodiment of the present application.

[0026] Figure 6 It is a flowchart of sub-step S3232 of the digital distribution network fault traveling wave ranging method according to an embodiment of the present application.

[0027] Figure 7 It is a block diagram of a digital distribution network fault traveling wave ranging device according to an embodiment of the present application. Detailed implementation manners

[0028] As shown in the present application and the claims, unless the context clearly indicates otherwise, words such as "a", "an", "one" and / or "the" are not specifically singular and may also include plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of the clearly identified steps and elements, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0029] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or the server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0030] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the operations before or below do not necessarily need to be executed precisely in sequence. On the contrary, as needed, various steps can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several steps can be removed from these processes.

[0031] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0032] It should be noted that in this application, all actions of obtaining data are carried out on the premise of complying with the corresponding data protection regulations and policies of the country where the data is located and obtaining the authorization given by the owner of the corresponding device.

[0033] In the power system, when a fault occurs in a transmission line (such as a short circuit or a broken wire), voltage and current mutations will occur at the fault point. These mutations will form a transient electromagnetic wave that propagates along the transmission line, namely the so-called fault traveling wave. The fault traveling wave propagates along the line towards both ends at a speed close to the speed of light and can be captured by sensors installed at both ends of the line. The "wavefront" refers to the starting part of the fault traveling wave, that is, the steep voltage or current change front generated at the moment of the fault occurrence, which marks the start of the fault traveling wave and is the exact time point of the fault occurrence. For fault location, accurately detecting the wavefront of the traveling wave can calculate the location of the fault point by comparing the time difference of the traveling wave reaching both ends of the line.

[0034] However, due to the short length and many branches of the distribution network line, the fault traveling wave is easily affected by factors such as discontinuous line impedance and equipment reflection during the propagation process, resulting in rapid attenuation of the signal amplitude and serious waveform distortion. This makes it difficult for traditional traveling wave detection methods based on set thresholds to accurately identify the signal wavefront, thereby generating time measurement errors and affecting the accuracy of fault location. In response to this, this application proposes a digital distribution network fault traveling wave ranging method. It uses fault traveling wave acquisition devices deployed at each key node of the distribution network to collect fault traveling wave signals in real time, and obtains stable line-mode traveling wave signals through signal filtering and line-mode component extraction. Then, it introduces a signal processing algorithm based on deep learning to perform time-domain feature analysis on the line-mode traveling wave signals of each key node to detect the wavefront position of the signal, marks the time when the fault traveling wave signal reaches the acquisition device accordingly, and further determines the section of the distribution network where the fault point is located by comparing the signal arrival times of each key node. Based on the double-end traveling wave ranging algorithm, according to the time difference of the signal arrival at both ends of this section and combined with the signal propagation speed, the distances between the fault point and both ends of the node are calculated. This method can accurately identify the wavefront position of the signal through in-depth time-domain analysis of the fault traveling wave signal, improving the reliability of fault traveling wave ranging in the distribution network.

[0035] Figure 1 It is a flowchart of the digital distribution network fault traveling wave ranging method according to an embodiment of the present application. Figure 2 It is a schematic diagram of data flow of the digital distribution network fault traveling wave ranging method according to an embodiment of the present application. As Figure 1 and Figure 2As shown in the figure, the digital distribution network fault traveling wave ranging method includes the following steps: S1, using fault traveling wave acquisition devices deployed at each key node of the distribution network to collect fault traveling wave signals in real time, where the key nodes include all branch nodes of the distribution network and both ends of the main line; S2, filtering and extracting line-mode components from the fault traveling wave signals of each key node to obtain a set of line-mode traveling wave signals; S3, performing deep learning-based wavefront detection on each line-mode traveling wave signal in the set of line-mode traveling wave signals and determining the wavefront arrival time to obtain a set of key node fault traveling wave arrival times; S4, processing the set of key node fault traveling wave arrival times based on the section location algorithm to determine the distribution network section where the fault point is located between two adjacent key nodes; S5, processing the fault traveling wave arrival times of the two adjacent key nodes based on the double-end traveling wave ranging algorithm to determine the distances between the fault point and the two adjacent key nodes.

[0036] In the above digital distribution network fault traveling wave ranging method, in step S1, fault traveling wave acquisition devices deployed at each key node of the distribution network are used to collect fault traveling wave signals in real time, where the key nodes include all branch nodes of the distribution network and both ends of the main line. It should be understood that the distribution network has the topological characteristics of multiple branches and short lines. During the propagation of fault traveling waves, reflections and transmissions will occur due to sudden changes in line impedance (such as branch points and transformers), resulting in rapid attenuation of signal energy and complex signal paths. To ensure that the characteristics of fault traveling waves on the key propagation paths can be captured, fault traveling wave acquisition devices are deployed at all branch nodes of the distribution network and both ends of the main line in this application to cover the main channels where fault traveling waves may arrive and provide basic data support for subsequent section location. Specifically, the fault traveling wave acquisition device uses the principle of electromagnetic induction to convert the electromagnetic energy carried by the traveling wave signal propagating in the line into an electrical signal for acquisition, and through analog-to-digital conversion technology, discretizes the continuous analog electrical signal and converts it into a digital signal that is convenient for storage, transmission, and processing, so as to realize the transformation of the traveling wave signal from a physical quantity to a digital quantity.

[0037] Specifically, any fault that occurs on the power transmission line will generate traveling waves in a specific form. These traveling waves carry rich information, including important parameters such as the location where the fault occurs. When traveling waves propagate along the transmission line, reflections or transmissions will occur when encountering impedance mutation points such as branch points and transformers, resulting in attenuation of signal energy and a more complex signal path. Therefore, in order to accurately capture and analyze these complex signals, fault traveling wave acquisition devices need to be deployed at all branch nodes of the distribution network and both ends of the main line. These positions are regarded as key nodes because they cover the main channels where fault traveling waves may arrive and provide solid basic data support for subsequent data processing and fault location.

[0038] The core function of the fault traveling wave acquisition device is to convert the electromagnetic energy carried by the traveling wave signal propagating in the line into an electrical signal. In the specific implementation process, sensors that require precise design are needed. These sensors can efficiently detect weak traveling wave signals without affecting the normal operation of the power system. After the traveling wave signal is captured, the next step is to convert it from an analog signal to a digital signal. In this way, the originally continuously changing physical quantity can be represented by discrete numerical values, which is convenient for storage, transmission, and further computational processing. The technology to achieve this transformation is called analog-to-digital conversion technology. Through this technology, the original traveling wave signal can not only be accurately recorded but also quickly transmitted to the background system for in-depth analysis.

[0039] Considering the complexity of the distribution network topology, especially the characteristics of multi-branch and short lines, the real-time acquisition of fault traveling wave signals is particularly important. Since fault traveling waves may experience multiple reflections and transmissions when passing through multiple branch nodes and different types of equipment, the signal intensity may rapidly weaken, and the waveform may be severely distorted. Therefore, relying solely on monitoring at a single location often fails to comprehensively understand the true situation of the fault. Installing fault traveling wave acquisition devices at all key nodes can effectively solve this problem. Each acquisition point can work independently and cooperate with each other to form a complete monitoring network. In this way, no matter where the fault occurs, the specific location of the fault can be determined by comparing the time difference of the arrival of traveling wave signals from different directions.

[0040] In addition, considering the influence of environmental factors such as temperature and humidity on line parameters, the fault traveling wave acquisition device also needs to have a certain adaptability. This means that in the hardware design, high-precision and high-stability components must be used to ensure the quality of signal acquisition even in harsh environments. At the same time, software algorithms also need to be continuously optimized to more accurately identify and compensate for errors caused by external conditions. In this way, not only can the accuracy of fault location be improved, but also the robustness of the entire system can be enhanced, enabling it to operate stably under various complex conditions.

[0041] In the above digital distribution network fault traveling wave ranging method, in step S2, the fault traveling wave signals of the respective key nodes are filtered and line mode component extraction is performed to obtain a set of line mode traveling wave signals. Specifically, due to the presence of power electronic device switching noise, arc high-frequency interference, and power frequency harmonic superposition in the distribution network, the signal-to-noise ratio (SNR) of the original traveling wave signal is low, seriously affecting the accurate analysis of the fault traveling wave signal. At the same time, the fault traveling wave contains a line mode component (propagation between lines) and a ground mode component (propagation between line and ground). Among them, the line mode component is less affected by factors such as line parameter changes and electromagnetic interference during propagation, and has higher stability and reliability, while the ground mode component is easily distorted by the influence of the grounding impedance. Therefore, in order to extract signal features with higher stability, the present application adopts signal filtering and line mode component extraction technologies to preprocess the fault traveling wave signals of the respective key nodes. Specifically, first, the wavelet threshold denoising technology is used to eliminate the baseline noise to improve the signal-to-noise ratio of the signal. Wavelet threshold denoising is an effective signal denoising method. It uses the multi-scale analysis characteristics of wavelet transform to decompose the signal into wavelet coefficients of different frequency components, and then by setting a threshold, the wavelet coefficients below the threshold are regarded as noise and removed, and then through wavelet inverse transform, the denoised signal is reconstructed, thereby retaining the main features of the signal. The extraction of the line mode component is based on the multi-phase transmission line theory, and through specific matrix transformation of the collected three-phase traveling wave signals, the separation of the line mode component from other components is realized. In the embodiment of the present application, the Db4 wavelet basis is selected to decompose the signal into 6 layers, and the high-frequency coefficients are adaptively shrunk by improving the threshold function (such as SUREShrink), while retaining the steep wavefront characteristics of the traveling wave, the baseline noise is filtered out. Then, based on the KLM (Karenbauer) transformation, the filtered three-phase voltage / current signals are converted into a line mode component (α mode) and a ground mode component (β mode, γ mode), and the α mode signal is extracted through matrix operations. The propagation speed of the line mode traveling wave signal is stable and less affected by line parameters, and is suitable as a ranging reference.

[0042] In the specific implementation process, for the baseline noise existing in the original traveling wave signal, using the wavelet threshold denoising technology for processing is a very effective means. The multi-scale analysis characteristics of wavelet transform enable the signal to be decomposed into wavelet coefficients of different frequency components. By setting a reasonable threshold, the wavelet coefficients below the threshold are regarded as noise and removed, and then the denoised signal is reconstructed through wavelet inverse transform. This method can not only effectively eliminate noise, but also maximize the retention of the main features of the signal. Taking the Db4 wavelet basis as an example, selecting it to decompose the signal into 6 layers and combining the SUREShrink improved threshold function to adaptively shrink the high-frequency coefficients can filter out the baseline noise while retaining the steep wavefront characteristics of the traveling wave. In this way, the quality of the processed signal is significantly improved, providing clear data support for further analysis.

[0043] Meanwhile, considering that the fault traveling wave contains both line-mode components and ground-mode components, and the ground-mode components are distorted due to being vulnerable to the influence of grounding impedance, it is more inclined to use the line-mode components with higher stability as the main analysis object. Based on the polyphase transmission line theory, by performing specific matrix transformations on the collected three-phase traveling wave signals, the effective separation of the line-mode components from other components can be achieved. Specifically, the KLM (Karenbauer) transformation, as a commonly used transformation method, can convert the filtered three-phase voltage or current signals into line-mode components (α-mode) and ground-mode components (β-mode, γ-mode). Through a series of precise matrix operations, the α-mode signal, that is, the line-mode component, is extracted therefrom. This line-mode component is less affected by the changes in line parameters and electromagnetic interference during propagation, has high stability and reliability, and is very suitable for use as a ranging reference.

[0044] Meanwhile, it is also crucial to ensure that the fault traveling wave acquisition devices at all key nodes can accurately capture the original signals. These devices not only need to have high sensitivity but also good anti-interference ability so as to obtain high-quality data in a complex environment. Then, for the signals obtained at each acquisition point, they all need to go through the above-mentioned wavelet threshold denoising technique and the line-mode component extraction step. In this process, selecting appropriate parameters such as the wavelet basis type, decomposition level, and threshold function plays a decisive role in ensuring the processing effect. For example, selecting the Db4 wavelet basis for 6-layer decomposition and applying the SUREShrink improved threshold function can achieve a better denoising effect while avoiding signal feature loss caused by over-smoothing.

[0045] In the above digital distribution network fault traveling wave ranging method, in step S3, perform deep learning-based wavefront detection on each line-mode traveling wave signal in the set of line-mode traveling wave signals and determine the wavefront arrival time to obtain a set of fault traveling wave arrival times at key nodes. It should be understood that the complex operating environment of the distribution network, such as the discontinuity of line impedance, reflection and refraction of electrical equipment, etc., will cause the fault traveling wave signal to be distorted. Traditional wavefront detection methods (such as the wavelet transform modulus maximum method) rely on artificially setting thresholds and are prone to misjudgment in scenarios with low signal-to-noise ratio or complex waveforms (such as multiple reflection superposition). Therefore, in order to improve the accuracy and robustness of wavefront detection, this application introduces a deep learning-based signal processing algorithm. With the powerful feature extraction and pattern recognition capabilities of the deep learning model, perform time-domain feature analysis on the line-mode traveling wave signals of each key node to adaptively identify the wavefront positions under different noise levels and distortion forms and improve the time marking accuracy. Among them, Figure 3 is the flowchart of sub-step S3 of the digital distribution network fault traveling wave ranging method according to the embodiment of the present application. As Figure 3As shown, step S3 includes steps: S31, performing local time-domain encoding on the line-mode traveling wave signal to obtain a set of local time-domain feature encoding vectors of the line-mode traveling wave signal; S32, performing time-domain-semantic double-constraint optimization on the set of local time-domain feature encoding vectors of the line-mode traveling wave signal to obtain a set of enhanced local time-domain feature encoding vectors of the line-mode traveling wave signal; S33, performing wavefront detection on the set of enhanced local time-domain feature encoding vectors of the line-mode traveling wave signal to determine the wavefront position and mark the wavefront arrival time.

[0046] Specifically, in a specific example of the present application, step S31 includes: performing local waveform feature extraction on the line-mode traveling wave signal based on one-dimensional convolutional encoding to obtain a set of local time-domain feature encoding vectors of the line-mode traveling wave signal. It should be understood that the wavefront of the traveling wave in the distribution network fault appears as a steep mutation point (such as a microsecond-level rising edge) in the time-domain signal, but its shape is easily affected by line parameters, noise interference, and multiple reflection superposition, presenting local non-linear characteristics. Traditional manual features (such as gradients, curvatures) are difficult to comprehensively describe the difference between the wavefront and background noise, while a one-dimensional convolutional neural network (1D-CNN) can automatically learn the abstract representation of the local time-domain pattern of the signal using a one-dimensional convolutional kernel, extract more discriminative feature encodings, and capture the microscopic fluctuation characteristics near the wavefront, thus laying a foundation for subsequent refined wavefront detection. In the embodiment of the present application, a one-dimensional convolutional layer with a sliding step of 1 is designed to perform a convolutional operation on the line-mode traveling wave signal, and the one-dimensional convolutional layer has multiple convolutional kernels of different sizes for capturing local features of different scales, such as high-frequency mutation details, low-frequency trend components, etc., so as to enhance the sensitivity and generalization ability of the model to wavefront features. In this way, at each time step of the line-mode traveling wave signal, multiple one-dimensional convolutional kernels of different scales are stacked to perform a convolutional operation on the local neighborhood of this time step to extract local waveform features of different dimensions, and the outputs of each convolutional kernel are concatenated to obtain the local time-domain feature encoding vector at this time step. In addition, since the sliding step of each convolutional kernel is 1, it ensures that each local time-domain feature encoding vector of the line-mode traveling wave signal extracted is strictly aligned with the time resolution of the original signal (that is, each time step corresponds to a local time-domain feature encoding vector of the line-mode traveling wave signal), avoiding the loss of feature information.

[0047] Specifically, in step S32, the set of local time-domain feature encoding vectors of the line-mode traveling wave signals is optimized with time-domain and semantic double constraints to obtain a set of enhanced local time-domain feature encoding vectors of the line-mode traveling wave signals. It should be understood that in the scenario of traveling wave ranging for distribution network faults, due to the short line length, many branches, and frequent impedance mutations, the fault traveling wave signals are vulnerable to reflection superposition and noise interference during propagation, resulting in distortion or ambiguity of local time-domain features (such as wavefront steepness, phase jump). Therefore, in order to distinguish the real wavefront features from noise, the present application further optimizes each local time-domain feature representation of the line-mode traveling wave signals with time-domain and semantic double constraints, so as to improve the discriminability of the feature representations of each local time-domain feature encoding vector of the line-mode traveling wave signals by jointly modeling the physical time-domain laws of signal propagation (such as wave velocity consistency, path topology dependence) and high-order semantic discriminability (such as wavefront mutation mode, fault type association), thereby laying a foundation for subsequent wavefront detection. Among them, Figure 4 is a flowchart of sub-step S32 of the digital distribution network fault traveling wave ranging method according to an embodiment of the present application. As Figure 4 shown, step S32 includes the steps of: S321, extracting the local time-domain feature encoding vector of the line-mode traveling wave signal to be enhanced from the set of local time-domain feature encoding vectors of the line-mode traveling wave signals; S322, performing n random samplings on the set of local time-domain feature encoding vectors of the line-mode traveling wave signals to obtain n local time-domain feature encoding vectors of the line-mode traveling wave signals as a sparse set of local time-domain reference feature encoding vectors of the line-mode traveling wave signals; S323, based on the sparse set of local time-domain reference feature encoding vectors of the line-mode traveling wave signals, performing time-domain and semantic double-domain information compensation optimization on the local time-domain feature encoding vector of the line-mode traveling wave signal to be enhanced to obtain the enhanced local time-domain feature encoding vector of the line-mode traveling wave signal.

[0048] More specifically, step S321 is expressed by the formula:

[0049]

[0050] Among them, represents the set of local time-domain feature encoding vectors of the line-mode traveling wave signals, represents the local time-domain feature encoding vector of the line-mode traveling wave signals at the th position in the set of local time-domain feature encoding vectors of the line-mode traveling wave signals, represents the local time-domain feature encoding vector of the line-mode traveling wave signal to be enhanced.

[0051] Here, the local time-domain feature coding vector of the line-mode traveling wave signal to be enhanced can be any one of the local time-domain feature coding vectors of the line-mode traveling wave signal in the set. By traversing the entire set, each local time-domain feature coding vector of the line-mode traveling wave signal is enhanced one by one, so that it can more accurately represent the true fluctuation characteristics of the traveling wave signal of the distribution network fault at each time step.

[0052] More specifically, step S322 is expressed by the formula:

[0053]

[0054] Where 、 、 and respectively represent the 1st, 2nd, th, and th local time-domain reference feature coding vectors of the line-mode traveling wave signal in the sparse set of local time-domain reference feature coding vectors of the line-mode traveling wave signal, represents the sparse set of local time-domain reference feature coding vectors of the line-mode traveling wave signal.

[0055] That is, by introducing a random scanning strategy, the selective attention mechanism of biological vision can be simulated, and local signal fluctuation patterns are dynamically selected from the set of local time-domain feature coding vectors of the line-mode traveling wave signal to form a sparse set of local time-domain reference feature coding vectors of the line-mode traveling wave signal covering different time-domain features. It not only retains the key signal fluctuation characteristics, but also suppresses redundant noise interference through sparse sampling. At the same time, through the complementary context information of multiple sampling points in the sparse set, the time-domain feature distribution of the fault traveling wave can be more comprehensively represented, and the computational efficiency advantage of sparse sampling is used to balance the requirements of feature diversity and real-time performance.

[0056] Figure 5 It is a flowchart of sub-step S323 of the digital distribution network fault traveling wave ranging method according to the embodiment of the present application. As Figure 5As shown, step S323 includes the steps of: S3231, calculating an information compensation coding vector between each line-mode traveling wave signal local time-domain reference feature coding vector in the sparse set of line-mode traveling wave signal local time-domain reference feature coding vectors and the to-be-enhanced line-mode traveling wave signal local time-domain feature coding vector to obtain a set of to-be-enhanced line-mode traveling wave signal local time-domain feature information compensation coding vectors; S3232, based on the time-domain-semantic dual-domain correlation between each line-mode traveling wave signal local time-domain reference feature coding vector in the sparse set of line-mode traveling wave signal local time-domain reference feature coding vectors and the to-be-enhanced line-mode traveling wave signal local time-domain feature coding vector, performing explicit modulation aggregation on the set of to-be-enhanced line-mode traveling wave signal local time-domain feature information compensation coding vectors to obtain a to-be-enhanced line-mode traveling wave signal local time-domain feature enhancement component implicit coding vector; S3233, fusing the to-be-enhanced line-mode traveling wave signal local time-domain feature enhancement component implicit coding vector and the to-be-enhanced line-mode traveling wave signal local time-domain feature coding vector to obtain the enhanced line-mode traveling wave signal local time-domain feature coding vector.

[0057] In a specific example of the present application, step S3231 is represented by the formula:

[0058]

[0059] Wherein, represents with respect to the to-be-enhanced line-mode traveling wave signal local time-domain feature information compensation coding vector.

[0060] That is, by introducing an information interaction mechanism between multiple line-mode traveling wave signal local time-domain reference feature coding vectors and the to-be-enhanced line-mode traveling wave signal local time-domain feature coding vector, the complementary information between different local features can be dynamically captured, the feature loss in a single perspective can be compensated, and the problem of local feature one-sidedness can be solved.

[0061] Figure 6 FIG. is a flowchart of sub-step S3232 of the digital distribution network fault traveling wave ranging method according to an embodiment of the present application. As Figure 6As shown in the figure, step S3232 includes the following steps: S32321, calculating the Poincaré distance between the local time-domain feature coding vector of the line-mode traveling wave signal to be enhanced and each local time-domain reference feature coding vector in the sparse set of local time-domain reference feature coding vectors of the line-mode traveling wave signal to obtain a local time-domain feature time-domain modulation matrix of the line-mode traveling wave signal to be enhanced composed of multiple Poincaré distances; S32322, calculating the implicit semantic association between the local time-domain feature coding vector of the line-mode traveling wave signal to be enhanced and each local time-domain reference feature coding vector in the sparse set of local time-domain reference feature coding vectors of the line-mode traveling wave signal to obtain a set of local time-domain feature semantic association coding matrices of the line-mode traveling wave signal to be enhanced; S32323, based on the local time-domain feature time-domain modulation matrix of the line-mode traveling wave signal to be enhanced and the set of local time-domain feature semantic association coding matrices of the line-mode traveling wave signal to be enhanced, performing explicit modulation aggregation coding on the set of local time-domain feature information compensation coding vectors of the line-mode traveling wave signal to be enhanced to obtain the local time-domain feature enhanced component implicit coding vector of the line-mode traveling wave signal to be enhanced.

[0062] In a specific example of the present application, step S32321 is expressed by the formula:

[0063]

[0064]

[0065] Wherein, represents the square of the norm of the vector, represents the inverse hyperbolic cosine function, represents and the Poincaré distance between them, represents and the Poincaré distance between them, represents the local time-domain feature time-domain modulation matrix of the line-mode traveling wave signal to be enhanced.

[0066] It should be understood that through the above compensation information extraction, the complementary information of each reference feature relative to the local time-domain feature coding vector of the line-mode traveling wave signal to be enhanced is captured, but not all of this complementary information is beneficial to the enhancement of the local time-domain feature coding vector of the line-mode traveling wave signal to be enhanced. Therefore, in order to more accurately guide each local time-domain feature information compensation coding vector of the line-mode traveling wave signal to be enhanced to participate in the final feature enhancement process, the present application further uses the time-domain correlation and semantic correlation between each reference feature and the local time-domain feature coding vector of the line-mode traveling wave signal to be enhanced to screen and optimize the set of local time-domain feature information compensation coding vectors of the line-mode traveling wave signal to be enhanced.

[0067] Here, first, by introducing the Poincaré distance metric, an attention mechanism based on hyperbolic geometry is constructed. By quantifying the relative position relationship between the local time-domain feature encoding vector of the line-mode traveling wave signal to be enhanced and the local time-domain reference feature encoding vector of the line-mode traveling wave signal in the hyperbolic space, the key reference patterns that have time-domain correlation with the local time-domain feature encoding vector of the line-mode traveling wave signal to be enhanced are dynamically identified. In this way, through the non-linear distribution modeling ability of the Poincaré distance for signal features, the generated local time-domain feature time modulation matrix of the line-mode traveling wave signal to be enhanced can enhance the adaptability of the model to the time-varying traveling wave signal under the complex distribution network topology.

[0068] In a specific example of the present application, the step S32322 is expressed by the formula:

[0069]

[0070] where represents the weight matrix, represents the transpose of the vector, represents the vector multiplication, represents the normalized exponential function, represents and the local time-domain feature semantic association coding matrix between the line-mode traveling wave signal to be enhanced.

[0071] Here, by introducing a latent semantic association mechanism, the deep semantic connections beyond the surface features between the local time-domain feature encoding vector of the line-mode traveling wave signal to be enhanced and each local time-domain reference feature encoding vector of the line-mode traveling wave signal are mined, and a feature enhancement framework based on a semantic attention map is constructed. In this way, by quantifying the similarity or complementarity between the local time-domain feature encoding vector of the line-mode traveling wave signal to be enhanced and each local time-domain reference feature encoding vector of the line-mode traveling wave signal in the semantic space, the reference patterns with strong semantic associations are dynamically screened, and the interference of irrelevant noise or conflicting features is suppressed. The deep semantic decoupling of complex fault patterns in the distribution network can be achieved through latent semantic associations, and the semantic understanding ability of the model for traveling wave signals in multi-branch and high-impedance heterogeneous scenarios can be enhanced.

[0072] Specifically, in a preferred example of the present application, the step S32323 includes: First, based on the set of the local time-domain feature time modulation matrix of the line-mode traveling wave signal to be enhanced and the set of the local time-domain feature semantic association coding matrix of the line-mode traveling wave signal to be enhanced, each local time-domain feature information compensation coding vector in the set of local time-domain feature information compensation coding vectors of the line-mode traveling wave signal to be enhanced is subjected to double-domain nested constraint optimization to obtain a set of optimized local time-domain feature information compensation coding vectors of the line-mode traveling wave signal to be enhanced, which is expressed by the formula:

[0073]

[0074]

[0075] Among them, denotes the compensated coding vector of the local time-domain feature information of the corresponding semantic nested optimized pending enhanced line-mode traveling wave signal, denotes the compensated coding vector of the local time-domain feature information of the corresponding optimized pending enhanced line-mode traveling wave signal.

[0076] That is, by maintaining the structural stability of semantic association through the constraint of gauge transformation invariance, while using the covariance transformation invariance to adapt to the dynamic geometric changes of the signal in the curved space, and enhancing the interaction intensity between the double-layer mask and information compensation with the coupling mechanism in the form of partial derivatives in field theory, so as to improve the robustness and representation accuracy of the compensated coding vector of the local time-domain feature information of the pending enhanced line-mode traveling wave signal in the complex distribution network environment. In this way, through the double-domain nested constraint, the multi-scale features of the fault traveling wave signal are synergistically enhanced, generating the compensated coding vector of the local time-domain feature information of the optimized pending enhanced line-mode traveling wave signal, effectively suppressing the feature distortion caused by the sudden change of line impedance and equipment reflection, and improving the positioning reliability of the double-ended traveling wave ranging algorithm in multi-branch and high-noise scenarios.

[0077] Then, taking each compensated coding matrix of the local time-domain feature semantic association of the pending enhanced line-mode traveling wave signal in the set of compensated coding matrices of the local time-domain feature semantic association of the pending enhanced line-mode traveling wave signal as the primary mask modulation unit, and taking the time-domain modulation degree matrix of the local time-domain feature of the pending enhanced line-mode traveling wave signal as the secondary mask modulation unit, performing explicit modulation aggregation coding on the set of compensated coding vectors of the local time-domain feature information of the optimized pending enhanced line-mode traveling wave signal to obtain the latent coding vector of the enhanced component of the local time-domain feature of the pending enhanced line-mode traveling wave signal, which is expressed by the formula:

[0078]

[0079] Among them, denotes the latent coding vector of the enhanced component of the local time-domain feature of the pending enhanced line-mode traveling wave signal.

[0080] Here, an explicit modulation mechanism is further introduced. Based on the regulation weights of the reference feature complementary information by the semantic association coding matrix of the local time-domain features of the line-mode traveling wave signal to be enhanced and the time-domain modulation matrix of the local time-domain features of the line-mode traveling wave signal to be enhanced, first, the complementary information semantically consistent with the local time-domain features of the line-mode traveling wave signal to be enhanced is screened by a semantic mask, and then the complementary information temporally related to the local time-domain features of the line-mode traveling wave signal to be enhanced is further screened by a time-domain mask. Through a hierarchical feature enhancement framework, a hidden coding vector of the enhanced component of the local time-domain features of the line-mode traveling wave signal to be enhanced is generated, enabling the information compensation process to focus on the key difference components that are semantically consistent and temporally adaptable.

[0081] In a specific example of the present application, step S3233 includes: calculating the position-wise weighted sum between the hidden coding vector of the enhanced component of the local time-domain features of the line-mode traveling wave signal to be enhanced and the coding vector of the local time-domain features of the line-mode traveling wave signal to be enhanced to obtain the coding vector of the local time-domain features of the enhanced line-mode traveling wave signal, which is expressed by the formula:

[0082]

[0083] Where and represent different fusion weight parameters, represents the coding vector of the local time-domain features of the enhanced line-mode traveling wave signal.

[0084] That is, by introducing a position-wise weighting mechanism, the fusion ratio between the hidden coding vector of the enhanced component of the local time-domain features of the line-mode traveling wave signal to be enhanced and the coding vector of the local time-domain features of the line-mode traveling wave signal to be enhanced is dynamically adjusted, solving the problem that local key information is submerged by global noise due to the staticization of feature fusion weights. In this way, the expression ability of the fault traveling wave signal features is significantly improved through dynamic weighted fusion. The generated coding vector of the local time-domain features of the enhanced line-mode traveling wave signal can more accurately characterize the true fluctuation features of the distribution network fault traveling wave signal at each time step, thereby improving the positioning accuracy and anti-interference ability of the double-ended traveling wave ranging algorithm in complex topologies and high-impedance heterogeneous scenarios.

[0085] Specifically, the step S33 performs wave head detection on the set of local time domain feature coding vectors of the enhanced line mode traveling wave signal to determine the wave head position and mark the wave head arrival time. In a specific example of the present application, the step S33 includes: inputting the set of local time domain feature coding vectors of the enhanced line mode traveling wave signal into a wave head detection module based on a bidirectional LSTM model to determine the wave head position. It should be understood that when performing wave head detection, it is necessary to comprehensively consider the before and after timing correlation of the signal: for example, the signal before the wave head is usually relatively stable, while a significant mutation will occur when the wave head arrives, and the reflected wave is manifested as multiple small fluctuations after the main wave head. As a special recurrent neural network (RNN), the bidirectional LSTM (Long Short-Term Memory) model can simultaneously capture the forward and backward timing dependencies of the signal, and perform bidirectional modeling on the set of local time domain feature coding vectors of the enhanced line mode traveling wave signal to avoid misjudgment caused by the unidirectional LSTM due to "invisible future information" (such as mistaking the reflected secondary peak for the main wave head). Specifically, the bidirectional LSTM model receives the set of local time domain feature encoding vectors of the enhanced line mode traveling wave signal as input, and uses the forward LSTM layer and the backward LSTM layer to traverse the entire signal sequence from front to back and from back to front respectively, so as to capture the dependency between the local features of the signal at each time step and the features of the previous and next time steps, and integrates the context information of the previous and next time steps into the current time step to obtain the hidden state of each time step, and maps the hidden state of each time step to 2 dimensions through full connection encoding at the output layer, and calculates the probability distribution of binary classification (wave head and non-wave head) through the Softmax function to obtain the prediction result of whether it is a wave head at each time step. Then, by extracting the time step with the largest wave head probability in the prediction result as the wave head position, and marking the corresponding wave head arrival time, the accurate detection of the arrival time of the fault traveling wave head is achieved.

[0086] In the above-mentioned digital distribution network fault traveling wave ranging method, the step S4 processes the set of key node fault traveling wave arrival times based on the section positioning algorithm to determine the distribution network section where the fault point is located between two adjacent key nodes. It should be understood that the time when the distribution network fault traveling wave arrives at different nodes is directly related to its propagation path. The section positioning algorithm is based on the propagation characteristics of the fault traveling wave in the distribution network, that is, the fault traveling wave will arrive at different key nodes in a certain order. By comparing the order of arrival time of the fault traveling wave of each key node, and combining the topological structure and connection relationship of the distribution network, a corresponding logical judgment model is established, so that the section where the fault point is located can be determined. In this way, the scope of fault location can be quickly narrowed, and the originally complex full-network search problem can be converted into a positioning problem within a specific section, which greatly improves the efficiency of fault location and saves a lot of time for the subsequent accurate calculation of the fault point location.

[0087] In the above digital distribution network fault traveling wave ranging method, in step S5, the arrival times of the fault traveling waves at the two adjacent key nodes are processed based on the double - end traveling wave ranging algorithm to determine the distances between the fault point and the two adjacent key nodes. It should be understood that after determining the section where the fault point is located, in order to accurately locate the fault point, it is necessary to accurately calculate the distances between the fault point and the key nodes at both ends of the section. Specifically, the double - end traveling wave ranging algorithm is based on the physical principle that traveling waves propagate at a constant speed on a uniform transmission line. Based on the traveling wave propagation speed of , and the time difference when the fault traveling wave arrives at the key nodes at both ends of the section , through the formula , where is the distance from the fault point to one end node, is the total length of the distribution network section between two adjacent key nodes, the relative distances between the fault point and the key nodes at both ends of the section can be calculated, and then the exact location of the fault point can be determined, providing an accurate basis for power maintenance personnel to quickly find the fault point and carry out fault repair work.

[0088] In summary, the digital distribution network fault traveling wave ranging method based on the embodiments of the present application is elucidated. It uses the fault traveling wave acquisition devices deployed at each key node of the distribution network to collect fault traveling wave signals in real - time, and obtains stable line - mode traveling wave signals through signal filtering and line - mode component extraction. Then, a signal processing algorithm based on deep learning is introduced to perform time - domain feature analysis on the line - mode traveling wave signals of each key node to detect the wave - head position of the signal, mark the time when the fault traveling wave signal arrives at the acquisition device accordingly. Furthermore, by comparing the signal arrival times of each key node, the distribution network section where the fault point is located is determined, and based on the double - end traveling wave ranging algorithm, according to the time difference between the signal arrival times of the nodes at both ends of this section, combined with the signal propagation speed, the distances between the fault point and the nodes at both ends are calculated. This method can accurately identify the wave - head position of the signal through in - depth time - domain analysis of the fault traveling wave signal, improving the reliability of distribution network fault traveling wave ranging.

[0089] Furthermore, a digital distribution network fault traveling wave range finder is also provided.

[0090] Figure 7 It is a block diagram of the digital distribution network fault traveling wave range finder according to the embodiments of the present application. As Figure 7As shown in the figure, the digital distribution network fault traveling wave distance measuring instrument 100 according to the embodiment of the present application includes: a fault traveling wave signal acquisition module 110, configured to use fault traveling wave acquisition devices deployed at each key node of the distribution network to collect fault traveling wave signals in real time, where the key nodes include all branch nodes of the distribution network and both ends of the main line; a fault traveling wave signal preprocessing module 120, configured to filter and extract the line mode component of the fault traveling wave signals at each key node to obtain a set of line mode traveling wave signals; a wavefront detection module 130, configured to perform wavefront detection based on deep learning on each line mode traveling wave signal in the set of line mode traveling wave signals and determine the wavefront arrival time to obtain a set of key node fault traveling wave arrival times; a distribution network section determination module 140, configured to process the set of key node fault traveling wave arrival times based on a section location algorithm to determine the distribution network section where the fault point is located between two adjacent key nodes; a double-end traveling wave distance measuring module 150, configured to process the fault traveling wave arrival times of the two adjacent key nodes based on a double-end traveling wave distance measuring algorithm to determine the distances between the fault point and the two adjacent key nodes.

[0091] Here, those skilled in the art can understand that the specific operations of each module in the above digital distribution network fault traveling wave distance measuring instrument have been described in detail in the description of the digital distribution network fault traveling wave distance measuring method above with reference to Figures 1 to 6 and thus, the repeated description thereof will be omitted.

[0092] The basic principles of the present invention have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, advantages, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, advantages, effects, etc. are essential for each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and easy understanding, rather than limitations, and the above details do not limit the present invention to necessarily adopt the above specific details to implement.

[0093] In the above embodiments, the descriptions of each embodiment have their own focuses. For the parts not detailed or recorded in a certain embodiment, reference may be made to the relevant descriptions of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative. For example, the unit division is only a logical function division, and there may be other division methods in actual implementation. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0094] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above-mentioned exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present invention. Any reference signs in the claims should not be regarded as limiting the claims involved.

[0095] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0096] Finally, it should be noted that the above description has been given for purposes of illustration and description. In addition, the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A digital distribution network fault traveling wave ranging method, characterized in that: include: The fault traveling wave acquisition device deployed at each key node of the power distribution network is used to collect the fault traveling wave signal in real time, wherein the key nodes include all branch nodes and both ends of the trunk line of the power distribution network; Filtering the fault traveling wave signals of each key node and extracting the line mode components to obtain a set of line mode traveling wave signals; Performing wave head detection based on deep learning on each line mode traveling wave signal in the set of line mode traveling wave signals and determining the wave head arrival time to obtain a set of critical node fault traveling wave arrival times; Processing the set of arrival times of the key node fault traveling waves based on a section location algorithm to determine the distribution network section where the fault point is located between two adjacent key nodes; Processing the arrival time of the fault traveling waves of the two adjacent key nodes based on a double-terminal traveling wave ranging algorithm to determine the distance between the fault point and the two adjacent key nodes; The method of performing deep learning-based wave head detection on each line mode traveling wave signal in the set of line mode traveling wave signals and determining the wave head arrival time to obtain a set of critical node fault traveling wave arrival times includes: Performing local time-domain coding on the line-mode traveling wave signal to obtain a set of local time-domain feature coding vectors of the line-mode traveling wave signal; Performing time domain-semantic dual constraint optimization on the set of local time domain feature coding vectors of the line mode traveling wave signal to obtain a set of enhanced local time domain feature coding vectors of the line mode traveling wave signal; The wave head detection is performed on the set of local time domain feature coding vectors of the enhanced line mode traveling wave signal to determine the wave head position, and the wave head arrival time is marked.

2. The digital distribution network fault traveling wave ranging method according to claim 1 is characterized in that: Performing local time domain coding on the line mode traveling wave signal to obtain a set of local time domain feature coding vectors of the line mode traveling wave signal includes: A local waveform feature extraction based on one-dimensional convolution coding is performed on the line mode traveling wave signal to obtain a set of local time domain feature coding vectors of the line mode traveling wave signal.

3. The digital distribution network fault traveling wave ranging method according to claim 2 is characterized in that: The set of local time domain feature coding vectors of the line mode traveling wave signal is subjected to time domain-semantic dual constraint optimization to obtain a set of local time domain feature coding vectors of the enhanced line mode traveling wave signal, including: Extracting the local time domain feature coding vector of the line mode traveling wave signal to be enhanced from the set of local time domain feature coding vectors of the line mode traveling wave signal; Performing n random sampling on the set of local time domain feature coding vectors of the line mode traveling wave signal to obtain n local time domain feature coding vectors of the line mode traveling wave signal as a sparse set of local time domain reference feature coding vectors of the line mode traveling wave signal; Based on the sparse set of local time domain reference feature coding vectors of the line mode traveling wave signal, the local time domain feature coding vectors of the line mode traveling wave signal to be enhanced are optimized by time domain-semantic dual domain information compensation to obtain the enhanced local time domain feature coding vectors of the line mode traveling wave signal.

4. The digital distribution network fault traveling wave distance measurement method according to claim 3 is characterized in that: Based on the sparse set of local time domain reference feature coding vectors of the line mode traveling wave signal, the local time domain feature coding vectors of the line mode traveling wave signal to be enhanced are subjected to time domain-semantic dual-domain information compensation optimization to obtain the local time domain feature coding vectors of the enhanced line mode traveling wave signal, including: Calculating the information compensation coding vector between each local time domain reference feature coding vector of the line mode traveling wave signal in the sparse set of the local time domain reference feature coding vector of the line mode traveling wave signal and the local time domain feature coding vector of the line mode traveling wave signal to be enhanced to obtain a set of local time domain feature information compensation coding vectors of the line mode traveling wave signal to be enhanced; Based on the time domain-semantic dual-domain association between each local time domain reference feature coding vector of the line mode traveling wave signal in the sparse set of local time domain reference feature coding vectors of the line mode traveling wave signal and the local time domain feature coding vector of the line mode traveling wave signal to be enhanced, the set of local time domain feature information compensation coding vectors of the line mode traveling wave signal to be enhanced is explicitly modulated and aggregated to obtain an implicit coding vector of the local time domain feature enhancement component of the line mode traveling wave signal to be enhanced; The implicit coding vector of the enhanced component of the local time domain feature of the line mode traveling wave signal to be enhanced and the coding vector of the local time domain feature of the line mode traveling wave signal to be enhanced are fused to obtain the enhanced local time domain feature coding vector of the line mode traveling wave signal.

5. The digital distribution network fault traveling wave distance measurement method according to claim 4, characterized in that: Based on the time domain-semantic dual-domain association between each local time domain reference feature coding vector of the line mode traveling wave signal in the sparse set of the local time domain reference feature coding vector of the line mode traveling wave signal and the local time domain feature coding vector of the line mode traveling wave signal to be enhanced, the set of the local time domain feature information compensation coding vector of the line mode traveling wave signal to be enhanced is explicitly modulated and aggregated to obtain an implicit coding vector of the local time domain feature enhancement component of the line mode traveling wave signal to be enhanced, including: Calculating the Poincare distance between the local time domain feature coding vector of the line mode traveling wave signal to be enhanced and each local time domain reference feature coding vector of the line mode traveling wave signal in the sparse set of the local time domain reference feature coding vector of the line mode traveling wave signal to obtain a local time domain feature time domain modulation matrix of the line mode traveling wave signal to be enhanced composed of multiple Poincare distances; Calculating the implicit semantic association between the local time domain feature coding vector of the line mode traveling wave signal to be enhanced and each line mode traveling wave signal local time domain reference feature coding vector in the sparse set of the line mode traveling wave signal local time domain reference feature coding vector to obtain a set of semantic association coding matrices of the local time domain features of the line mode traveling wave signal to be enhanced; Based on the set of the time domain modulation matrix of the local time domain characteristics of the line mode traveling wave signal to be enhanced and the set of the semantic association coding matrix of the local time domain characteristics of the line mode traveling wave signal to be enhanced, the set of local time domain characteristic information compensation coding vectors of the line mode traveling wave signal to be enhanced is explicitly modulated and aggregated coded to obtain the implicit coding vector of the local time domain characteristic enhancement component of the line mode traveling wave signal to be enhanced.

6. The digital distribution network fault traveling wave distance measurement method according to claim 5, characterized in that: Based on the set of the time domain modulation matrix of the local time domain feature of the line mode traveling wave signal to be enhanced and the set of the semantic association coding matrix of the local time domain feature of the line mode traveling wave signal to be enhanced, the set of the local time domain feature information compensation coding vector of the line mode traveling wave signal to be enhanced is subjected to explicit modulation aggregation coding to obtain the implicit coding vector of the local time domain feature enhancement component of the line mode traveling wave signal to be enhanced, including: Based on the set of the time domain modulation matrix of the local time domain characteristics of the line mode traveling wave signal to be enhanced and the set of the semantic association coding matrix of the local time domain characteristics of the line mode traveling wave signal to be enhanced, performing double-domain nested constraint optimization on each local time domain characteristic information compensation coding vector of the line mode traveling wave signal to be enhanced in the set of the local time domain characteristic information compensation coding vector of the line mode traveling wave signal to be enhanced to obtain a set of optimized local time domain characteristic information compensation coding vectors of the line mode traveling wave signal to be enhanced; Using each of the semantic association coding matrices of the local time domain characteristics of the line mode traveling wave signal to be enhanced in the set of the semantic association coding matrices of the local time domain characteristics of the line mode traveling wave signal to be enhanced as the first-level mask modulation unit, and using the time domain modulation degree matrix of the local time domain characteristics of the line mode traveling wave signal to be enhanced as the second-level mask modulation unit, the set of the optimized local time domain characteristic information compensation coding vectors of the line mode traveling wave signal to be enhanced is explicitly modulated and aggregated encoded to obtain the implicit coding vector of the local time domain characteristic enhancement component of the line mode traveling wave signal to be enhanced.

7. The digital distribution network fault traveling wave distance measurement method according to claim 6, characterized in that: The method of fusing the implicit coding vector of the enhanced component of the local time domain feature of the line mode traveling wave signal to be enhanced and the local time domain feature coding vector of the line mode traveling wave signal to be enhanced to obtain the enhanced local time domain feature coding vector of the line mode traveling wave signal comprises: The position-weighted sum between the implicit coding vector of the local time domain feature enhancement component of the line mode traveling wave signal to be enhanced and the local time domain feature coding vector of the line mode traveling wave signal to be enhanced is calculated to obtain the enhanced local time domain feature coding vector of the line mode traveling wave signal.

8. The digital distribution network fault traveling wave distance measurement method according to claim 7, characterized in that: Performing wave head detection on the set of local time domain feature coding vectors of the enhanced line mode traveling wave signal to determine the wave head position and marking the wave head arrival time, including: The set of local time domain feature coding vectors of the enhanced line mode traveling wave signal is input into a wave head detection module based on a bidirectional LSTM model to determine the wave head position.

9. A digital distribution network fault traveling wave ranging instrument, characterized in that: include: A fault traveling wave signal acquisition module is used to collect fault traveling wave signals in real time using fault traveling wave acquisition devices deployed at each key node of the power distribution network, wherein the key nodes include all branch nodes and both ends of the trunk line of the power distribution network; A fault traveling wave signal preprocessing module, used for filtering the fault traveling wave signals of each key node and extracting the line mode components to obtain a set of line mode traveling wave signals; A wave head detection module, used for performing wave head detection based on deep learning on each line mode traveling wave signal in the set of line mode traveling wave signals and determining the wave head arrival time to obtain a set of critical node fault traveling wave arrival times; A distribution network section determination module, used for processing the set of arrival times of the key node fault traveling waves based on a section location algorithm to determine the distribution network section where the fault point is located between two adjacent key nodes; A double-end traveling wave ranging module, used for processing the arrival time of the fault traveling waves of the two adjacent key nodes based on the double-end traveling wave ranging algorithm to determine the distance between the fault point and the two adjacent key nodes; The method of performing deep learning-based wave head detection on each line mode traveling wave signal in the set of line mode traveling wave signals and determining the wave head arrival time to obtain a set of critical node fault traveling wave arrival times includes: Performing local time-domain coding on the line-mode traveling wave signal to obtain a set of local time-domain feature coding vectors of the line-mode traveling wave signal; Performing time domain-semantic dual constraint optimization on the set of local time domain feature coding vectors of the line mode traveling wave signal to obtain a set of enhanced local time domain feature coding vectors of the line mode traveling wave signal; The wave head detection is performed on the set of local time domain feature coding vectors of the enhanced line mode traveling wave signal to determine the wave head position, and the wave head arrival time is marked.

Images