Distribution network traveling wave fault early warning positioning analysis method and system

Abstract

The invention provides a distribution network traveling wave fault early warning positioning analysis method and system, and belongs to the technical field of smart power grids, and the method comprises the steps: obtaining a plurality of traveling wave signals of a distribution network line, and filtering the interference to obtain a high-fidelity traveling wave signal; constructing a basic wave velocity model and a dynamic wave velocity correction model, and calculating coordinates of a fault point by adopting a double-end, single-end and branch node three-dimensional positioning algorithm; extracting basic parameters of a time domain, a frequency domain and a polarity domain from the high-fidelity traveling wave signal, and obtaining coupling correlation characteristics among the basic parameters; establishing a fault mechanism model library, and substituting the basic parameters and the coupling correlation features into the model library for adaptation to determine a preliminary fault type; and performing parameter conflict verification on the preliminary fault type. According to the distribution network traveling wave fault early warning positioning analysis method provided by the invention, the signal acquisition dimension is expanded to electrical and vibration three-dimensional signals, the fault type identification accuracy is also improved, and different fault types with similar peak values can be accurately distinguished.

Description

Technical Field

[0001] This application belongs to the field of smart grid technology, and more specifically, relates to a method and system for early warning and location analysis of traveling wave faults in distribution networks. Background Technology

[0002] As the core link connecting the power system and users, the timeliness and accuracy of fault early warning and location in the distribution network directly determine the reliability of power supply. According to statistics from the State Grid Corporation of China, single-phase grounding, lightning strikes, and equipment insulation degradation account for over 80% of distribution network faults. If these faults are not quickly located and handled, the outage area will expand and the power restoration time will be prolonged, severely impacting industrial production and residential life. Traveling wave fault location technology, utilizing the electromagnetic transient traveling wave signal generated at the moment of a fault, has advantages such as fast response speed and insensitivity to line load fluctuations, and has gradually replaced the traditional impedance method as the mainstream technology for distribution network fault location. However, existing traveling wave fault early warning and location technologies have limitations in application, including limited signal acquisition dimensions, insufficient coverage, and low calculation accuracy, failing to meet the accuracy requirements of distribution network operation and maintenance. Summary of the Invention

[0003] The purpose of this application is to provide a method and system for early warning and location analysis of traveling wave faults in distribution networks, so as to solve the technical problems of single signal acquisition dimension, insufficient coverage and low calculation accuracy in the existing technology.

[0004] To achieve the above objectives, the technical solution adopted in this application is as follows: A method for early warning and location analysis of traveling wave faults in distribution networks is provided, comprising: S1: acquiring voltage traveling wave, current traveling wave, and fiber optic vibration traveling wave signals of distribution network lines, filtering out interference to obtain high-fidelity traveling wave signals; S2: constructing a basic wave velocity model, deriving a dynamic wave velocity correction model by combining environmental parameters and transient load fluctuations of the line, calculating the fault point coordinates using a three-dimensional location algorithm for dual-end, single-end, and branch nodes, and triggering graded early warnings; S3: extracting time-domain, frequency-domain, and polarity-domain basic parameters from the high-fidelity traveling wave signals, and obtaining coupling correlation characteristics between basic parameters, including the correlation coefficient between peak value and dominant frequency, the linkage characteristics between rising edge slope and polarity reversal times, and the mapping relationship between signal duration and the proportion of high-frequency components; S4: establishing a fault mechanism model library that dynamically adjusts according to the line's operating status, and adapting the basic parameters and coupling correlation characteristics into the model library to determine the preliminary fault type; S5: performing parameter conflict verification on the preliminary fault type, and if contradictions exist, dynamically adjusting and re-adapting through the weights of the coupling correlation characteristics.

[0005] In one possible implementation, in S1, the traveling wave signal of the distribution network line is acquired by combining a distributed optical fiber sensor with an electrical traveling wave acquisition device. A timing module and a clock synchronization calibration unit are arranged in the combined acquisition device. The timing module provides a high-precision reference time signal, and the clock synchronization calibration unit compares the acquisition timestamps of the three types of traveling wave signals in real time, corrects the acquisition data with time deviations, and ensures that the acquisition times of the three types of signals are completely aligned before storing them in a unified data frame. The high-fidelity traveling wave signal is obtained by fusing the three types of signals using a dynamic weighted average method.

[0006] In one possible implementation, in S2, the initial distance for dual-end positioning is L1 = (v × Δt1) / 2, where Δt1 is the time difference between the arrival of the electrical traveling waves at both ends; the auxiliary distance for single-end positioning is L2 = (v × Δt2) / 2, where Δt2 is the time difference between the incident wave and the reflected wave at the fault point; when the fault is located on a branch line, the branch node correction distance is L3 = L2 - ΔL, where ΔL is the distance from the branch node to the main line; noise suppression is performed on L1, L2, and L3 according to the state equation and observation equation in the Kalman filter equation, and the three-dimensional coordinates of the fault point are output.

[0007] In one possible implementation, when constructing the dynamic wave velocity correction model, real-time environmental factor parameters and line operation status parameters are acquired, and these parameters are collected every minute by the distribution network online monitoring device. Experiments are conducted using the controlled variable method to determine the influence of single factor changes on wave velocity, and then the wave velocity correction coefficient under the coupling effect of multiple factors is quantified through orthogonal experiments. When the change of any factor exceeds a preset threshold, model correction is immediately initiated. When there is no significant change, the model is updated periodically at a 5-minute cycle. The corrected wave velocity value is compared with the measured wave velocity value under the same historical conditions. If the deviation exceeds 3%, the coupling coefficient is readjusted.

[0008] In one possible implementation, in S3, after feature extraction is completed, the extraction results of the same feature in different acquisition channels are compared in real time, and the feature consistency coefficient between channels is calculated. If the consistency coefficient is ≥80%, the feature is determined to be valid. If 60%≤consistency coefficient<80%, the weighted average of the features of each channel is taken as the final result. If the consistency coefficient<60%, the signal re-acquisition and feature re-extraction of the corresponding channel are triggered in real time.

[0009] In one possible implementation, during the feature extraction process in S3, the extracted basic parameters and coupled features are verified in real time, and the correlation between them and the data in S2 is checked. If the correlation between the feature value and the wave velocity / fault energy is <0.3, the feature is determined to be an abnormal feature.

[0010] In one possible implementation, in S4, the membership degree of the fault type is solved by the parameter-mechanism mapping equation. When the membership degree is greater than a set threshold, the preliminary fault type is determined. The insulation resistance value of the line is collected in real time by the online monitoring device. When the insulation resistance value is ≥ 80% of the initial value, the standard mechanism model parameters are adopted. When 60% ≤ insulation resistance value < 80%, the energy threshold of partial discharge and insulation breakdown fault models is reduced by 20% in real time. When the insulation resistance value < 60%, the adaptation priority of the insulation-related fault model is raised to the highest level, and the adaptation tolerance of the corresponding features is expanded.

[0011] In one possible implementation, line load rate data is acquired in real time. When the load rate is ≤50%, the adaptation sensitivity of the transient fault model is increased; when the load rate is 50%-80%, a balanced adaptation mode is adopted; when the load rate is >80%, the weight of the load-related fault model is increased in real time, while the calculation cycle of the mechanism adaptation of the corresponding features is shortened.

[0012] In one possible implementation, in S5, parameter conflicts include time-domain and frequency-domain conflicts, polarity and mechanism conflicts, and multi-dimensional cross-conflicts. The feature weights are adjusted according to the conflict type, and the membership is recalculated. When the parameter conflict is a time-domain and frequency-domain conflict, the frequency-domain feature weight is increased by 20%; when the parameter conflict is a polarity and mechanism conflict, the coupling and correlation feature weight is increased by 30%; and when the parameter conflict is a multi-dimensional cross-conflict, the vibration traveling wave correlation feature weight is increased by 40%.

[0013] The beneficial effects of the distribution network traveling wave fault early warning and location analysis method provided in this application are as follows: Compared with the prior art, the distribution network traveling wave fault early warning and location analysis method of this application firstly acquires three types of signals simultaneously: voltage traveling wave, current traveling wave, and fiber optic vibration traveling wave. Among them, the electrical traveling wave can capture the characteristics of electrical faults such as single-phase grounding and lightning strikes, while the fiber optic vibration traveling wave can accurately sense the physical vibration signals of mechanical faults such as conductor breakage and tower tilting, achieving full coverage of fault types. Then, a targeted interference filtering algorithm is used to eliminate on-site interference such as switch operation and electromagnetic radiation, providing a high-fidelity data foundation for subsequent analysis. Next, a basic wave velocity model is first constructed based on the inherent properties of the line material and cross-sectional dimensions. Then, the coupling influence of real-time environmental parameters such as temperature and humidity and transient load fluctuations of the line is innovatively incorporated to derive a dynamic wave velocity correction model, solving the error problem caused by traditional fixed wave velocity or single parameter correction. At the same time, a three-dimensional positioning algorithm of dual-end, single-end, and branch nodes is adopted, combined with multi-terminal collaborative verification logic, to accurately calculate the coordinates of the fault point and simultaneously trigger a hierarchical early warning mechanism to ensure the timeliness of fault response.

[0014] Subsequently, three fundamental parameters—time domain (peak value, duration), frequency domain (dominant frequency, proportion of high-frequency components), and polarity domain (number of polarity reversals)—are comprehensively extracted from the high-fidelity traveling wave signal. This approach overcomes the limitations of traditional isolated parameter analysis and further explores coupled features such as the correlation coefficient between peak value and dominant frequency, and the linkage between rising edge slope and number of polarity reversals. A multi-dimensional, in-depth feature system is constructed to provide a more comprehensive basis for fault type identification. A fault mechanism model library that dynamically adjusts with the line's operating status is established, abandoning the rigidity of traditional fixed models. The model library can be updated in real time based on equipment aging and insulation status changes. The extracted fundamental parameters and coupled features are substituted into the model library for adaptive adaptation, initially identifying the fault type. Finally, cross-dimensional parameter conflict verification is performed on the preliminary judgment results. If contradictions exist between features and mechanisms, the weights of the coupled features are dynamically adjusted for re-adaptation, ensuring the uniqueness and accuracy of the output results. In this way, the signal acquisition dimension is expanded from a single electrical signal to three-dimensional electrical and vibration signals, solving the problem of incomplete fault coverage and significantly reducing the missed detection rate of mechanical faults. The combination of dynamic wave velocity correction model and three-dimensional positioning algorithm further reduces the positioning error. Coupled with a dynamically adjusted fault mechanism model library, the accuracy of fault type identification is improved, and different fault types with similar peak values ​​can be accurately distinguished.

[0015] Another objective of this application is to provide a distribution network traveling wave fault early warning and location analysis system, including any of the above-mentioned distribution network traveling wave fault early warning and location analysis methods.

[0016] The distribution network traveling wave fault early warning and location analysis system provided in this application adopts the distribution network traveling wave fault early warning and location analysis method. The signal acquisition dimension is expanded from a single electrical signal to three-dimensional electrical and vibration signals, which solves the problem of incomplete fault coverage and significantly reduces the missed detection rate of mechanical faults. The combination of dynamic wave velocity correction model and three-dimensional positioning algorithm further reduces the positioning error. With the addition of a dynamically adjusted fault mechanism model library, the accuracy of fault type identification is improved, and different fault types with similar peak values ​​can be accurately distinguished. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating the distribution network traveling wave fault early warning and location analysis method provided in this application embodiment. Detailed Implementation

[0019] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this application.

[0020] It should be noted that when a component is referred to as being "fixed to" or "set on" another component, it can be directly on or indirectly on that other component. When a component is referred to as being "connected to" another component, it can be directly connected to or indirectly connected to that other component.

[0021] It should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0022] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0023] Please see Figure 1 The method for early warning and location analysis of traveling wave faults in distribution networks provided in this application will now be explained. A method for early warning and location analysis of traveling wave faults in distribution networks includes: S1: acquiring voltage traveling wave, current traveling wave, and fiber optic vibration traveling wave signals of distribution network lines, filtering out interference to obtain high-fidelity traveling wave signals; S2: constructing a basic wave velocity model, deriving a dynamic wave velocity correction model by combining environmental parameters and transient load fluctuations of the line, calculating the fault point coordinates using a three-dimensional location algorithm for dual-end, single-end, and branch nodes, and triggering graded early warnings; S3: extracting time-domain, frequency-domain, and polarity-domain basic parameters from the high-fidelity traveling wave signals, and obtaining the coupling correlation characteristics between the basic parameters, including the correlation coefficient between peak value and dominant frequency, the linkage characteristics between rising edge slope and polarity reversal number, and the mapping relationship between signal duration and high-frequency component proportion; S4: establishing a fault mechanism model library that dynamically adjusts with the line operating status, and adapting the basic parameters and coupling correlation characteristics into the model library to determine the preliminary fault type; S5: performing parameter conflict verification on the preliminary fault type, and if there are contradictions, dynamically adjusting and re-adapting through the weights of the coupling correlation characteristics.

[0024] The distribution network traveling wave fault early warning and location analysis method provided in this application, compared with existing technologies, firstly acquires three types of signals simultaneously: voltage traveling wave, current traveling wave, and fiber optic vibration traveling wave. Among them, the electrical traveling wave can capture the characteristics of electrical faults such as single-phase grounding and lightning strikes, while the fiber optic vibration traveling wave can accurately sense the physical vibration signals of mechanical faults such as conductor breakage and tower tilting, achieving full coverage of fault types. Then, a targeted interference filtering algorithm is used to eliminate on-site interference such as switch operation and electromagnetic radiation, providing a high-fidelity data foundation for subsequent analysis. Next, a basic wave velocity model is constructed based on the inherent properties of the line material and cross-sectional dimensions. Then, the coupling influence of real-time environmental parameters such as temperature and humidity and transient load fluctuations of the line is innovatively incorporated to derive a dynamic wave velocity correction model, solving the error problem caused by traditional fixed wave velocity or single parameter correction. At the same time, a three-dimensional positioning algorithm of dual-end, single-end, and branch nodes is adopted, combined with multi-terminal collaborative verification logic, to accurately calculate the coordinates of the fault point and simultaneously trigger a hierarchical early warning mechanism to ensure the timeliness of fault response.

[0025] Subsequently, three fundamental parameters—time domain (peak value, duration), frequency domain (dominant frequency, proportion of high-frequency components), and polarity domain (number of polarity reversals)—are comprehensively extracted from the high-fidelity traveling wave signal. This approach overcomes the limitations of traditional isolated parameter analysis and further explores coupled features such as the correlation coefficient between peak value and dominant frequency, and the linkage between rising edge slope and number of polarity reversals. A multi-dimensional, in-depth feature system is constructed to provide a more comprehensive basis for fault type identification. A fault mechanism model library that dynamically adjusts with the line's operating status is established, abandoning the rigidity of traditional fixed models. The model library can be updated in real time based on equipment aging and insulation status changes. The extracted fundamental parameters and coupled features are substituted into the model library for adaptive adaptation, initially identifying the fault type. Finally, cross-dimensional parameter conflict verification is performed on the preliminary judgment results. If contradictions exist between features and mechanisms, the weights of the coupled features are dynamically adjusted for re-adaptation, ensuring the uniqueness and accuracy of the output results. In this way, the signal acquisition dimension is expanded from a single electrical signal to three-dimensional electrical and vibration signals, solving the problem of incomplete fault coverage and significantly reducing the missed detection rate of mechanical faults. The combination of dynamic wave velocity correction model and three-dimensional positioning algorithm further reduces the positioning error. Coupled with a dynamically adjusted fault mechanism model library, the accuracy of fault type identification is improved, and different fault types with similar peak values ​​can be accurately distinguished.

[0026] Please see Figure 1As a specific implementation of the distribution network traveling wave fault early warning and location analysis method provided in this application, in S1, the traveling wave signal of the distribution network line is acquired by combining a distributed optical fiber sensor and an electrical traveling wave acquisition device. A timing module and a clock synchronization calibration unit are arranged in the combined acquisition device. The timing module provides a high-precision reference time signal, and the clock synchronization calibration unit compares the acquisition timestamps of the three types of traveling wave signals in real time, corrects for any time deviations in the acquired data, and ensures that the acquisition times of the three types of signals are completely aligned before storing them in a unified data frame. A dynamic weighted average method is used to fuse the three types of signals to obtain the high-fidelity traveling wave signal. A hardware architecture combining a distributed optical fiber sensor and an electrical traveling wave acquisition device is constructed, wherein the distributed optical fiber sensor utilizes the Rayleigh scattering principle, injecting laser pulses into the optical fiber and monitoring changes in the intensity of backscattered light. The system accurately captures vibration traveling wave signals caused by mechanical faults such as broken conductors and tilted towers. The electrical traveling wave acquisition device simultaneously acquires voltage and current traveling waves, achieving full coverage of electrical fault (single-phase grounding, lightning strike) and mechanical fault signals, solving the problem of missed mechanical faults in traditional single electrical signal acquisition. Subsequently, the combined device's built-in timing module and clock synchronization calibration unit are activated. The timing module uses nanosecond-level high-precision reference time signals provided by Beidou or GPS satellite navigation systems (synchronization error can be controlled within ±10ns). The clock synchronization calibration unit extracts the acquisition timestamps of the three types of signals in real time, identifies time deviations through comparison and analysis, and corrects deviations exceeding 50ns to ensure that the acquisition times of voltage, current, and vibration traveling waves are completely aligned. The synchronized three types of signals are then stored in the same data frame in a unified format.

[0027] Finally, a dynamic weighted average method is used for signal fusion. The fault type tendency is judged based on the signal features extracted in real time. If the vibration traveling wave is more similar to the typical features of mechanical faults, it is given a higher weight. If the electrical traveling wave is more similar to the typical features of electrical faults, the weight ratio of voltage and current traveling waves is increased. A high-fidelity traveling wave signal that takes into account the characteristics of different fault types is obtained through weighted calculation.

[0028] In this way, the multi-source acquisition architecture ensures comprehensive signal capture for both mechanical and electrical faults, significantly reducing the rate of missed fault detection. High-precision time synchronization technology avoids feature distortion caused by signal time misalignment. Combined with the stringent time synchronization accuracy requirements of smart grids, this provides accurate data support for subsequent wave velocity calculations and fault location, reducing positioning errors down to the hundreds of meters level caused by time inaccuracies. Dynamic weighted fusion, compared to fixed-weight fusion, can adapt to the differences in signal characteristics across different fault scenarios, effectively filtering out on-site interference such as switching operations and electromagnetic radiation.

[0029] Please see Figure 1As a specific implementation of the distribution network traveling wave fault early warning and location analysis method provided in this application, in S2, the initial distance for double-end positioning is L1=(v×Δt1) / 2, where Δt1 is the time difference between the arrival of the electrical traveling waves at both ends; the auxiliary distance for single-end positioning is L2=(v×Δt2) / 2, where Δt2 is the time difference between the incident wave and the reflected wave at the fault point; when the fault is located on a branch line, the branch node correction distance is L3=L2-ΔL, where ΔL is the distance from the branch node to the main line; noise suppression is performed on L1, L2, and L3 according to the state equation and observation equation in the Kalman filter equation, and the three-dimensional coordinates of the fault point are output. Based on the high-fidelity traveling wave signal obtained in the first step and the dynamically corrected traveling wave propagation speed v, the initial distance L1 for dual-end positioning is calculated. The arrival time of the fault traveling wave is synchronously captured by the monitoring terminals at both ends of the line, and the time difference Δt1 is obtained. Substituting this into the formula L1=(v×Δt1) / 2, the initial distance from the fault point to the reference point is obtained. This method relies on the time difference between the two ends for positioning, and its core advantage lies in its strong stability in trunk line scenarios. Subsequently, single-end positioning auxiliary calibration is initiated. The arrival time of the fault incident wave and the fault point reflected wave are captured by the same terminal, and the time difference Δt2 is calculated. The auxiliary distance L2=(v×Δt2) / 2 is obtained by utilizing the round-trip propagation characteristics of the traveling wave, providing cross-validation basis for the dual-end positioning results and making up for the limitations of dual-end positioning in scenarios with insufficient terminal deployment. When the fault occurs in a branch line, the distance ΔL from the branch node to the main line reference point is obtained by combining the distribution network GIS topology data. The single-end positioning results are corrected by the formula L3=L2-ΔL to solve the positioning offset problem caused by the reflection of the traveling wave at the branch node, and adapt to the complex line structure of trunk and branch lines.

[0030] Finally, the Kalman filter algorithm is introduced for data optimization. A state equation is constructed to describe the dynamic changes in the fault distance. An observation equation is established to link the three positioning results (L1, L2, and L3) with the actual fault location. Through prediction and update iteration, interference caused by traveling wave signal noise and small time synchronization deviations is effectively suppressed. The fused three-dimensional coordinates of the fault point are output. At the same time, coordinate calibration is completed by combining the line terrain and tower location data to ensure that the positioning results accurately match the actual geographical location.

[0031] The multi-positioning method collaboratively covers different topological scenarios such as main lines and branch lines, improving the positioning coverage rate. Noise suppression using Kalman filtering reduces the impact of interference signals, and combined with dynamically corrected wave velocity v, the final positioning accuracy can be controlled within the meter range. The branch node correction mechanism reduces the fault location error of branch lines, and the 3D coordinate output directly interfaces with the operation and maintenance GIS system, accurately marking the specific location of the fault point relative to the tower and terrain, significantly reducing the intensity of manual line inspection.

[0032] Please see Figure 1As a specific implementation of the distribution network traveling wave fault early warning and location analysis method provided in this application, when constructing the dynamic wave velocity correction model, real-time environmental factor parameters and line operating status parameters are acquired, and these parameters are collected every minute by the distribution network online monitoring device. Experiments are conducted using the controlled variable method to determine the influence of single-factor changes on wave velocity, and then the wave velocity correction coefficient under the coupling effect of multiple factors is quantified through orthogonal experiments. When the change of any factor exceeds a preset threshold, model correction is immediately initiated; when there is no significant change, updates are performed periodically at a 5-minute cycle. The corrected wave velocity value is compared with the measured wave velocity value under the same historical conditions; if the deviation exceeds 3%, the coupling coefficient is readjusted. The parameter acquisition range and frequency are clearly defined. Real-time environmental factor parameters (temperature, humidity, air pressure) and line operating status parameters (load rate, insulation resistance, conductor temperature) are synchronously acquired by the distribution network online monitoring device, with the acquisition frequency set to 1 minute / time to ensure timely capture of dynamic parameter changes and avoid wave velocity calculation errors caused by parameter lag. Subsequently, systematic experiments were conducted to quantify the influence patterns. First, the independent effects of single factors (temperature change ±10℃, load rate fluctuation ±20%) on wave velocity were studied using the controlled variable method. Wave velocity change data corresponding to different parameter values ​​were recorded, and wave velocity influence curves of single factors were fitted. Then, through orthogonal experimental design, typical level combinations of each parameter were selected to carry out multi-factor coupling experiments to eliminate mutual interference between factors. Based on the experimental data, multiple regression analysis was used to quantify the wave velocity correction coefficients (including the main effect coefficients and interaction effect coefficients of each factor) under the coupling effect of multiple factors such as temperature, humidity, and load rate, providing a quantitative basis for model construction.

[0033] Next, a model correction trigger mechanism is set up to compare the newly collected parameters with the parameter values ​​from the last correction in real time. If the change of any factor exceeds the preset threshold (temperature change ≥5℃, load rate fluctuation ≥15%), the model correction is immediately initiated, and the wave velocity is recalculated by substituting the latest parameters and coupling correction coefficients. If the change of all factors does not exceed the threshold, the wave velocity value is updated periodically at a 5-minute cycle to balance correction accuracy and computational efficiency. Finally, a closed-loop verification mechanism is established to compare the wave velocity value after each correction with the measured wave velocity value under the same historical conditions (obtained by back-calculating the actual fault location through on-site verification after the fault). If the deviation exceeds 3%, the orthogonal experimental data is re-analyzed, and the weight ratio of the coupling coefficient is adjusted to ensure the accuracy and stability of the model in long-term operation.

[0034] By leveraging high-frequency parameter acquisition, the limitations of traditional low-frequency acquisition are overcome, enabling precise capture of dynamic fluctuations in parameters during distribution network operation and providing a data foundation for real-time wave velocity correction. The combination of the controlled variable method and orthogonal experiments clarifies both the independent influence of single factors and quantifies the coupling effect of multiple factors. The dual-mode correction mechanism of real-time triggering and periodic updates avoids frequent corrections caused by small parameter fluctuations and prevents delayed wave velocity updates when parameters change significantly, ensuring a high degree of matching between wave velocity and the actual operating state of the line. The closed-loop verification mechanism effectively avoids long-term accuracy degradation caused by coupling coefficient drift, keeping the deviation of wave velocity correction consistently within 3%.

[0035] Please see Figure 1 As a specific implementation of the distribution network traveling wave fault early warning and location analysis method provided in this application, in S3, after feature extraction is completed, the extraction results of the same feature in different acquisition channels are compared in real time, and the feature consistency coefficient between channels is calculated. If the consistency coefficient is ≥80%, the feature is determined to be valid; if 60%≤consistency coefficient<80%, the weighted average of the features of each channel is taken as the final result; if the consistency coefficient<60%, the signal re-acquisition and feature re-extraction of the corresponding channel are triggered in real time. After the extraction of basic parameters and coupled correlation features in the time domain, frequency domain, and polarity domain is completed in the third step, for each feature (peak value, main frequency, correlation coefficient, etc.), its extraction results in the voltage traveling wave acquisition channel, current traveling wave acquisition channel, and fiber optic vibration traveling wave acquisition channel (or redundant acquisition channels of the same type) are retrieved synchronously. The feature consistency coefficient between channels is calculated by quantization algorithm: first, the average value of each channel feature value is used as the benchmark to calculate the relative deviation between the feature value of a single channel and the average value, and then the cosine similarity of the feature values ​​between channels is combined to obtain the comprehensive consistency coefficient (the weight is allocated as 60% for relative deviation and 40% for cosine similarity), so as to achieve accurate quantification of feature consistency.

[0036] Subsequently, a tiered processing is performed based on the consistency coefficient results. If the coefficient is ≥80%, it indicates that the feature extraction results of each channel are highly consistent, with no obvious interference or error, and the feature is directly determined to be valid and the original extraction result is retained. If 60% ≤ consistency coefficient < 80%, it indicates that there are slight differences between channels but they are within a reasonable range. Weights are assigned according to the acquisition accuracy level of each channel (the fiber optic vibration channel is weighted at 0.4 due to its strong anti-interference ability, and the voltage and current traveling wave channels are each weighted at 0.3), and the weighted average is calculated as the final result of the feature to balance the channel differences. If the consistency coefficient is <60%, it is determined that there are problems such as signal interference or abnormal acquisition equipment in the corresponding channel, and the signal re-acquisition process of the channel is triggered in real time (the backup acquisition link is used first to ensure efficiency). Feature extraction is re-executed after the new signal acquisition is completed until the consistency coefficient is ≥60%.

[0037] By cross-verifying data from different channels, feature distortion caused by factors such as electromagnetic radiation and equipment drift is effectively avoided, resolving the issue of questionable reliability of single-channel extraction results. The hierarchical processing mechanism not only ensures the efficiency of direct application of highly consistent features but also reasonably resolves minor differences through weighted averaging, avoiding data waste caused by overly stringent standards. At the same time, real-time re-acquisition is initiated for low-consistency features, forming a closed-loop control of extraction, verification, and correction, thereby improving feature effectiveness. High-quality feature data provides a solid foundation for the subsequent fourth step of fault mechanism model adaptation and the fifth step of parameter conflict verification, reducing misjudgment of fault types caused by feature errors and further improving the accuracy of fault identification.

[0038] Please see Figure 1 As a specific implementation of the distribution network traveling wave fault early warning and location analysis method provided in this application, in the feature extraction process of S3, basic parameters and coupled correlation features are extracted in real time, and the correlation with the data in S2 is verified. If the correlation between the feature value and the wave velocity / fault energy is <0.3, the feature is determined to be an abnormal feature. On the one hand, the basic parameters in the time domain (peak value, signal duration), frequency domain (dominant frequency, proportion of high-frequency components), and polarity domain (number of polarity reversals) obtained in the third step are extracted, as well as coupled correlation features such as the correlation coefficient between the peak value and the dominant frequency, and the linkage feature between the rising edge slope and the number of polarity reversals. On the other hand, the key data in the second step are retrieved, including the real-time wave velocity value output by the dynamic wave velocity correction model, and the fault energy value calculated based on the voltage and current amplitude and duration of the high-fidelity traveling wave signal. These two parameters directly reflect the propagation characteristics and severity of the fault traveling wave and are the core indicators related to the physical nature of the fault. Subsequently, the Pearson correlation coefficient method was used to calculate the correlation degree: for each extracted feature, binary data sets of feature value and wave velocity, and feature value and fault energy were constructed respectively. The correlation coefficient of the two dimensions was obtained by calculating the ratio of covariance to standard deviation. The mean of the two was taken as the final correlation degree to realize the quantitative determination of the correlation degree.

[0039] Finally, the feature validity screening rules are executed. If the correlation degree is ≥0.3, it means that the feature has a reasonable correlation with the core physical process of the fault (traveling wave propagation, energy release), and is judged as a valid feature and retained in the feature set. If the correlation degree is <0.3, it means that the feature may be affected by irrelevant interference or signal distortion and is out of touch with the essence of the fault. It is judged as an abnormal feature and directly removed. At the same time, the abnormal feature type and correlation degree data are recorded to provide a basis for subsequent interference source tracing.

[0040] This approach constructs a linked verification logic for physical data and feature parameters, ensuring that the retained features all conform to the essential laws of faults. Furthermore, through the dual correlation screening of the two core physical quantities, wave velocity and fault energy, it significantly improves the effectiveness and relevance of the feature set, reduces the interference of invalid features on the subsequent fourth step of model adaptation, improves the computational efficiency of fault type determination, and reduces the risk of misjudgment caused by abnormal features.

[0041] Please see Figure 1 As a specific implementation of the distribution network traveling wave fault early warning and location analysis method provided in this application, in S4, the fault type membership degree is solved by the parameter and mechanism mapping equation. When the membership degree is greater than the set threshold, the preliminary fault type is determined. The line insulation resistance value is collected in real time by the online monitoring device. When the insulation resistance value is ≥ 80% of the initial value, the standard mechanism model parameters are adopted. When 60% ≤ insulation resistance value < 80%, the energy threshold of partial discharge and insulation breakdown fault models is reduced by 20% in real time. When the insulation resistance value < 60%, the adaptation priority of the insulation-related fault model is raised to the highest level, and the adaptation tolerance of the corresponding features is expanded. The basic parameters and coupling characteristics (peak value correlation coefficient with dominant frequency, rising edge slope and polarity reversal frequency linkage characteristics, etc.) extracted in the third step are substituted into the preset parameter and mechanism mapping equation. The membership degree of each fault type (single-phase grounding, lightning strike, partial discharge, insulation breakdown, etc.) is obtained by solving the equation. The membership degree value directly reflects the degree of matching between the feature and the corresponding fault mechanism. When the membership degree of a certain fault type is greater than the set threshold (usually set to 0.75, determined based on calibration of a large number of fault samples), the type is initially determined to be a candidate fault type. At the same time, the insulation resistance value of the line is collected in real time by the distribution network online monitoring device (the collection frequency is synchronized with the parameter collection in the second step to ensure data timeliness). The insulation resistance value is divided into three adjustment intervals according to the ratio of the insulation resistance value to the initial insulation resistance value: when the ratio is ≥80%, it indicates that the line insulation resistance value is within the range of 100%. When the insulation condition is good and the probability of insulation-related faults is low, standard mechanism model parameters are used for adaptation to avoid over-adjustment leading to misjudgment of non-insulation faults. When the ratio is 60% ≤ Ratio < 80%, it indicates that the line insulation has slightly aged, and the risk of faults such as partial discharge and insulation breakdown has increased. The energy threshold of this type of fault model is reduced by 20% in real time, making it easier for the model to capture the weak fault characteristics in the early stage of insulation degradation and reduce the missed detection of early insulation faults. When the ratio is < 60%, the line insulation condition is severely deteriorated, and insulation-related faults are high-probability events. On the one hand, the adaptation priority of insulation-related fault models such as insulation breakdown and partial discharge is raised to the highest level to ensure that this type of fault is matched first. On the other hand, the adaptation tolerance of corresponding features is expanded (the matching range of features such as peak value and main frequency is widened by 15%) to adapt to the irregular fluctuations of fault characteristics when the insulation is severely deteriorated.

[0042] By solving the membership degree through parameter-mechanism mapping equations, quantitative judgment of fault type matching is achieved, avoiding the subjective bias of traditional qualitative judgment and making the initial fault type determination more objective. Furthermore, it innovatively dynamically binds the line insulation state with the fault mechanism model parameters, solving the problem that traditional fixed models cannot adapt to dynamic changes such as line aging and insulation attenuation, significantly improving the accuracy of insulation-related fault identification. At the same time, the model adjustment strategy between zones takes into account the differences in fault characteristics under different insulation states, ensuring the accuracy of identifying non-insulation faults when insulation is good, while also strengthening the ability to capture high-risk faults when insulation deteriorates, improving the accuracy of initial fault type matching.

[0043] Please see Figure 1 As a specific implementation of the distribution network traveling wave fault early warning and location analysis method provided in this application, real-time acquisition of line load rate data is performed. When the load rate is ≤50%, the adaptation sensitivity of the transient fault model is increased; when the load rate is 50%-80%, a balanced adaptation mode is adopted; when the load rate is >80%, the weight of the load-related fault model is increased in real time, while the mechanism adaptation calculation cycle of the corresponding features is shortened. Based on the distribution network online monitoring device, line load rate data is collected in real time (the acquisition frequency is synchronized with the traveling wave signal and insulation resistance acquisition to ensure data timeliness). The load rate is calculated based on the line rated current, and the specific value is determined by the ratio of the real-time operating current to the rated current. Subsequently, three adaptation intervals are divided according to the load rate, and corresponding strategies are executed: when the load rate is ≤50%, the system is in a low-load operation state, and at this time, line faults are mostly transient faults such as lightning strikes and tree branch contact (these types of faults account for more than 70% of the total faults during low-load periods). By increasing the adaptation sensitivity of the transient fault model (increasing the weight of the corresponding features...), the system... The matching threshold is reduced by 15%-20%, which enhances the ability to capture weak transient fault characteristics and avoids missed detection due to insufficient sensitivity. When the load rate is in the medium load range of 50%-80%, the distribution of line fault types is relatively balanced, and the probabilities of transient faults, permanent faults and equipment inherent faults are similar. Therefore, a balanced adaptation mode is adopted to keep the weights and sensitivity of various fault models consistent, ensuring that different types of faults can be accurately identified. When the load rate is >80%, the line is in a high load operation state, and the risk of load-related faults such as equipment overload, poor contact, and insulation thermal aging increases significantly.

[0044] On the one hand, the weight of load-related fault models such as overload short circuit and equipment thermal failure is increased by more than 30% in real time to prioritize matching such fault characteristics. On the other hand, the calculation cycle for the mechanism adaptation of the corresponding characteristics is shortened from the standard 200ms to less than 100ms, thus accelerating the model adaptation speed.

[0045] Please see Figure 1As a specific implementation of the distribution network traveling wave fault early warning and location analysis method provided in this application, in S5, parameter conflicts include time domain and frequency domain conflicts, polarity and mechanism conflicts, and multi-dimensional cross-conflicts. The feature weights are adjusted according to the conflict type, and the membership degree is recalculated. When the parameter conflict is a time domain and frequency domain conflict, the frequency domain feature weight is increased by 20%; when the parameter conflict is a polarity and mechanism conflict, the coupling and correlation feature weight is increased by 30%; when the parameter conflict is a multi-dimensional cross-conflict, the vibration traveling wave correlation feature weight is increased by 40%. In the fifth step of parameter conflict verification, three typical conflicts are accurately identified: time-domain and frequency-domain conflicts refer to time-domain features (peak value, signal duration) and frequency-domain features (dominant frequency, proportion of high-frequency components) pointing to different fault types. For example, the time-domain peak value conforms to the characteristics of a single-phase grounding fault, but the frequency-domain dominant frequency falls into the typical frequency band of lightning strikes (1MHz-5MHz); polarity and mechanism conflicts refer to the mismatch between polarity features such as the number of polarity reversals and polarity direction of traveling waves and the inherent laws of the fault mechanism model. For example, the mechanism model of metallic grounding faults requires positive polarity, but the actual collected polarity features are negative; multi-dimensional cross-conflicts are the most complex type of conflict, where contradictions exist simultaneously in the time domain, frequency domain, polarity domain, and coupled correlation features.

[0046] Based on the conflict type, targeted weight adjustments are performed. For time-domain and frequency-domain conflicts, considering that time-domain characteristics are easily distorted by line dispersion and electromagnetic interference during traveling wave propagation, while the frequency domain has high concentration and strong anti-interference capability (main frequency band energy ratio ≥70%), the weight of frequency-domain features is increased by 20% to strengthen the decision-making role of stability features. For polarity and mechanism conflicts, since the coupling correlation feature integrates multi-parameter linkage relationships (correlation between peak value and main frequency), it is more in line with the physical essence of the fault and can avoid the one-sidedness of single polarity or mechanism judgment, so its weight is increased by 30%. For multi-dimensional cross-conflicts, given that the vibration traveling wave signal is less affected by interference and has strong characteristic stability in scenarios such as complex topology and high-resistance grounding, it can effectively break the deadlock of multi-dimensional contradictions, so the weight of vibration traveling wave correlation features is increased by 40%. After the weight adjustment, all features are re-substituted into the parameter and mechanism mapping equation to calculate the fault type membership degree. If the new membership degree still does not meet the set threshold or there are still conflicts, the verification and adjustment process is repeated until a unique fault type with the required confidence level is obtained.

[0047] The targeted empowerment of frequency domain features, coupling correlation features, and vibration traveling wave correlation features all rely on their core technological advantages. Frequency domain features have strong anti-distortion capabilities, coupling correlation features closely match fault mechanisms, and vibration traveling wave features have high stability, ensuring the scientific nature of weight adjustment and reducing the misclassification rate of fault types. Iterative membership calculation forms a closed-loop verification, effectively resolving multi-dimensional feature contradictions in complex fault scenarios. Accurate fault type determination provides maintenance personnel with precise troubleshooting directions, avoiding ineffective line patrols due to misjudgments, combined with accurate fault location results.

[0048] Not shown in the figure, this application embodiment also provides a distribution network traveling wave fault early warning and location analysis system, which includes any of the above-mentioned distribution network traveling wave fault early warning and location analysis methods.

[0049] The power distribution network traveling wave fault early warning and location analysis system includes distributed traveling wave monitoring instruments and a data control center. It achieves accurate location by capturing high-frequency transient traveling wave signals generated by faults and combining this with time difference calculations. The instruments are distributed along the line to collect traveling wave current and voltage waveform data in real time; the data control center receives the uploaded traveling wave signals and ensures time consistency through high-precision clock synchronization technology (GPS / BeiDou time service).

[0050] The distribution network traveling wave fault early warning and location analysis system provided in this application adopts the above-mentioned distribution network traveling wave fault early warning and location analysis method. The signal acquisition dimension is expanded from a single electrical signal to three-dimensional electrical and vibration signals, which solves the problem of incomplete fault coverage and significantly reduces the missed detection rate of mechanical faults. The combination of dynamic wave velocity correction model and three-dimensional positioning algorithm further reduces the positioning error. With the addition of a dynamically adjusted fault mechanism model library, the accuracy of fault type identification is improved, and different fault types with similar peak values ​​can be accurately distinguished.

[0051] The above are merely preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for early warning and location analysis of traveling wave faults in distribution networks, characterized in that, Includes S1: Acquire voltage traveling wave, current traveling wave and optical fiber vibration traveling wave signals of the distribution network line, and filter out interference to obtain high-fidelity traveling wave signals; S2: Construct a basic wave velocity model, derive a dynamic wave velocity correction model by combining environmental parameters and line transient load fluctuations, calculate the fault point coordinates using a three-dimensional positioning algorithm with dual-end, single-end, and branch nodes, and trigger a graded early warning; S3: Extract basic parameters in the time domain, frequency domain, and polarity domain from the high-fidelity traveling wave signal, and obtain the coupling correlation characteristics between the basic parameters, including the correlation coefficient between peak value and main frequency, the linkage characteristics between rising edge slope and polarity reversal number, and the mapping relationship between signal duration and high-frequency component proportion; S4: Establish a fault mechanism model library that is dynamically adjusted according to the line operation status. Substitute the basic parameters and coupled correlation features into the model library for adaptation to determine the initial fault type. S5: Perform parameter conflict verification on the initial fault type. If there are contradictions, dynamically adjust and re-adapt the model by adjusting the weights of the coupled correlation features.

2. The distribution network traveling wave fault early warning and location analysis method as described in claim 1, characterized in that, In S1, the traveling wave signal of the distribution network line is acquired by combining a distributed optical fiber sensor and an electrical traveling wave acquisition device. A timing module and a clock synchronization calibration unit are arranged in the combined acquisition device. The timing module provides a high-precision reference time signal, and the clock synchronization calibration unit compares the acquisition timestamps of the three types of traveling wave signals in real time, corrects the offset of the acquisition data with time deviation, and ensures that the acquisition times of the three types of signals are completely aligned before storing them in a unified data frame. The high-fidelity traveling wave signal is obtained by fusing the three types of signals using a dynamic weighted average method.

3. The distribution network traveling wave fault early warning and location analysis method as described in claim 1, characterized in that, In S2, the initial distance for dual-end positioning is L1 = (v × Δt1) / 2, where Δt1 is the time difference between the arrival of the electrical traveling waves at both ends; the auxiliary distance for single-end positioning is L2 = (v × Δt2) / 2, where Δt2 is the time difference between the incident wave and the reflected wave at the fault point; when the fault is located on a branch line, the branch node correction distance is L3 = L2 - ΔL, where ΔL is the distance from the branch node to the main line; noise suppression is performed on L1, L2, and L3 according to the state equation and observation equation in the Kalman filter equation, and the three-dimensional coordinates of the fault point are output.

4. The distribution network traveling wave fault early warning and location analysis method as described in claim 3, characterized in that, When constructing the dynamic wave velocity correction model, real-time environmental factor parameters and line operation status parameters are acquired, and these parameters are collected every minute by the distribution network online monitoring device. Experiments are conducted using the controlled variable method to determine the influence of single factor changes on wave velocity, and then the wave velocity correction coefficient under the coupling effect of multiple factors is quantified through orthogonal experiments. When the change of any factor exceeds the preset threshold, the model correction is immediately initiated. When there is no significant change, the model is updated periodically at a 5-minute cycle. The corrected wave velocity value is compared with the measured wave velocity value under the same historical conditions. If the deviation exceeds 3%, the coupling coefficient is readjusted.

5. The distribution network traveling wave fault early warning and location analysis method as described in claim 1, characterized in that, In S3, after feature extraction is completed, the extraction results of the same feature in different acquisition channels are compared in real time, and the feature consistency coefficient between channels is calculated. If the consistency coefficient is ≥80%, the feature is deemed valid. If 60%≤consistency coefficient<80%, the weighted average of the features of each channel is taken as the final result. If the consistency coefficient<60%, the signal re-acquisition and feature re-extraction of the corresponding channel are triggered in real time.

6. The distribution network traveling wave fault early warning and location analysis method as described in claim 5, characterized in that, In the feature extraction process in S3, the extracted basic parameters and coupled features are linked in real time, and the correlation is verified with the data in S2. If the correlation between the feature value and the wave velocity / fault energy is <0.3, the feature is determined to be an abnormal feature.

7. The distribution network traveling wave fault early warning and location analysis method as described in claim 1, characterized in that, In S4, the membership degree of the fault type is solved by the parameter-mechanism mapping equation. When the membership degree is greater than the set threshold, the preliminary fault type is determined. The insulation resistance value of the line is collected in real time by the online monitoring device. When the insulation resistance value is ≥ 80% of the initial value, the standard mechanism model parameters are adopted. When 60% ≤ insulation resistance value < 80%, the energy threshold of partial discharge and insulation breakdown fault models is reduced by 20% in real time. When the insulation resistance value < 60%, the adaptation priority of the insulation-related fault model is raised to the highest level, and the adaptation tolerance of the corresponding features is expanded.

8. The distribution network traveling wave fault early warning and location analysis method as described in claim 7, characterized in that, Real-time acquisition of line load rate data; when the load rate is ≤50%, increase the adaptation sensitivity of the transient fault model; when the load rate is 50%-80%, adopt the balanced adaptation mode; when the load rate is >80%, increase the weight of the load-related fault model in real time, and shorten the calculation cycle of the mechanism adaptation of the corresponding features.

9. The distribution network traveling wave fault early warning and location analysis method as described in claim 1, characterized in that, In S5, parameter conflicts include time-domain and frequency-domain conflicts, polarity and mechanism conflicts, and multi-dimensional cross-conflicts. The feature weights are adjusted according to the conflict type, and the membership degree is recalculated. When the parameter conflict is a time-domain and frequency-domain conflict, the frequency-domain feature weight is increased by 20%; when the parameter conflict is a polarity and mechanism conflict, the coupling and correlation feature weight is increased by 30%; and when the parameter conflict is a multi-dimensional cross-conflict, the vibration traveling wave correlation feature weight is increased by 40%.

10. A distribution network traveling wave fault early warning and location analysis system, characterized in that, This includes the distribution network traveling wave fault early warning and location analysis method as described in any one of claims 1-9.

Images