A power distribution network fault detection method and system based on FTU

Abstract

The application discloses a kind of power distribution network fault detection method and system based on FTU, it is related to power distribution network fault detection field, including using wavelet transform technique to analyze electrical parameter, identify the fault line of power distribution network, and utilize traveling wave transmitter to emit traveling wave to the endpoint of fault line, locate the fault position point of power distribution network fault line;Based on the influence parameter extracted according to the matching result of fault reason, the probability coefficient that power distribution network power supply area appears power failure condition in target period is judged, and the power supply reliability of power distribution network is evaluated according to probability coefficient;Based on power supply reliability, utilize fault terminal equipment to control switch equipment to carry out circuit operation, isolate power distribution network fault line, and generate personnel dispatching instruction, implement power distribution network fault line repair processing.The application realizes deep reasoning from fault phenomenon to fault cause by introducing highly fine recording wave data analysis and multidimensional dynamic feature extraction mechanism.

Description

Technical Field

[0001] This invention relates to the field of distribution network fault detection, and in particular to a distribution network fault detection method and system based on FTU. Background Technology

[0002] A distribution network is an important component of a power system. It refers to a network system that reliably transmits medium- and high-voltage electrical energy output from substations to end users (such as factories, residences, and commercial facilities) through distribution lines, distribution transformers, and related control equipment.

[0003] FTU (Feeder Terminal Unit) is a core intelligent device in the distribution network. It is mainly deployed in key nodes such as medium-voltage switch stations, ring main units, transformer substations, and branch boxes in the distribution network to realize functions such as feeder monitoring, data acquisition, event detection, remote control, and intelligent judgment.

[0004] Distribution network fault detection is a core component of ensuring stable power supply and rapid fault response. Its goal is to accurately identify and locate faulty lines, determine the type of fault, and assist in fault isolation and power restoration in the shortest possible time.

[0005] However, existing power distribution network fault detection methods can only identify faulty lines but cannot identify their precise coordinates or geographical locations. This fuzzy positioning method cannot support subsequent rapid repairs and fault isolation. Furthermore, it cannot infer whether there will be a substantial power outage in the power supply area based on the detected fault results. Consequently, it affects the intelligent scheduling and operation and maintenance strategy formulation of the power distribution network, leading to problems such as delayed resource allocation and low power restoration efficiency. Summary of the Invention

[0006] To address the aforementioned problems, this invention proposes a distribution network fault detection method and system based on FTU, aiming to improve the accuracy and timeliness of fault handling.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] In a first aspect, the present invention provides a distribution network fault detection method based on FTU, comprising:

[0009] S1. Collect electrical parameters of the distribution network using faulty terminal equipment, analyze the electrical parameters using wavelet transform technology, calculate transient energy based on the analysis results, and identify faulty lines in the distribution network.

[0010] S2. Use a traveling wave transmitter to transmit a traveling wave to the end of the faulty line, extract the fault characteristic wave based on the characteristics of the reflected traveling wave, and locate the fault location of the faulty line in the distribution network according to the wave velocity of the fault characteristic wave.

[0011] S3. Obtain waveform data at the fault location point, and extract dynamic feature metric values ​​from the waveform data based on intrinsic mode decomposition technology. Use the dynamic feature metric values ​​to match the fault cause of the faulty line in the distribution network.

[0012] S4. Extract the impact parameters based on the fault cause matching results, determine the probability coefficient of power outage in the power supply area of ​​the distribution network during the target time period, and evaluate the power supply reliability of the distribution network based on the probability coefficient.

[0013] S5. Based on power supply reliability, use faulty terminal equipment to control switching equipment to perform circuit breaking operations, isolate faulty lines in the distribution network, and generate personnel dispatch instructions to carry out emergency repairs on faulty lines in the distribution network.

[0014] Preferably, the method involves using a traveling wave transmitter to emit a traveling wave towards the endpoint of the faulty line, extracting the fault characteristic wave based on the characteristics of the reflected traveling wave, and locating the fault location point of the distribution network faulty line according to the wave velocity of the fault characteristic wave, including:

[0015] S21. Select one end of the faulty line as the detection end, and use the detection end as the reference position to transmit a traveling wave to the other end of the faulty line using a traveling wave transmitter. The traveling wave propagates along the faulty line.

[0016] S22. During the propagation of the traveling wave, when the received traveling wave encounters a fault point on the faulty line, the traveling wave reflected to the detection end is used as the fault characteristic wave. Based on the fault characteristic wave, an energy propagation spectrum is constructed to obtain the reflected energy trajectory of the traveling wave.

[0017] S23. Based on the reflected energy trajectory of the traveling wave and the actual length of the faulty line, construct a traveling wave propagation inversion model to determine the round-trip time and wave speed of the fault characteristic wave from the detection end to the fault point, and locate the fault point of the faulty line in the distribution network.

[0018] Preferably, the waveform data at the fault location is acquired, and dynamic feature metrics are extracted from the waveform data based on intrinsic mode decomposition (IMD) technology. The dynamic feature metrics are then used to match the fault causes of the distribution network fault lines, including:

[0019] S31. Obtain the waveform signal data of the fault location point, and select a straight line with the same characteristic properties as the axial generatrix on the surface of the fault location point as the component calculation path to obtain the rate of change of the fault mode component.

[0020] S32. Based on the rate of change of fault mode components, mode decomposition technology is used to decompose and optimize the center frequency of the recorded signal data, and the mode component signal corresponding to the fault location point is generated according to the decomposition and optimization results.

[0021] S33. Analyze the mean square value, standard deviation, sample entropy, permutation entropy and singular value entropy of the modal component signals as morphological dynamic feature measures, establish a mixed domain feature group, and extract the fault features of the fault location point.

[0022] S34. Introduce the fault features into the fault cause feature bag-of-words matching network, compare the similarity between the fault features and the fault cause feature bag-of-words, and match the fault cause of the faulty line in the distribution network based on the similarity results.

[0023] Preferably, the recorded waveform data of the fault location point is acquired, and a straight line with the same characteristic properties as the axial generatrix is ​​selected on the surface of the fault location point as the component calculation path. The rate of change of the fault mode components is obtained, including:

[0024] S311. Collect waveform signal data at the fault location point of the faulty line, analyze the strain mode components corresponding to any time point at the fault location point and the previous time point, and define the signal change rate at the fault location point.

[0025] S312. Based on the signal change rate on the surface at the fault location point, select a straight line with the same characteristic properties as the axial generatrix as the component calculation path, extract the equally spaced displacement mode components, and determine the change rate of the displacement mode components.

[0026] S313. Based on the signal change rate on the surface at the fault location point, select a straight line parallel to the axial generatrix as the identification calculation path, calculate the axial displacement at any time point to obtain the identification quantity, and combine it with the displacement modal component change rate to generate the fault modal component change rate.

[0027] Preferably, the mean square value, standard deviation, sample entropy, permutation entropy, and singular value entropy of the modal component signals are analyzed as morphological dynamic feature measures to establish a mixed domain feature set. The fault features extracted from the fault location points include:

[0028] S331. Calculate the mean square value, standard deviation, sample entropy, permutation entropy and singular value entropy of the modal component signals based on the set of equations, and determine the optimal form of the modal component signals based on the calculation results.

[0029] S332. Based on the optimal form, perform feature measurement on the modal component signals to form a mixed domain feature group of the fault location points of the faulty line, and obtain the component fault feature classification matrix based on the mixed domain feature group.

[0030] S333. Divide the component fault feature classification matrix into intrinsic mode matrices according to the complex wavelet decomposition features, obtain the clustering center vectors of the modal component signals based on the intrinsic mode matrices, and construct the feature component matrix.

[0031] S334. Based on the selection constraints and feature component matrix, perform optimal feature selection processing on the modal component signals to determine the feature extraction relationship of the modal component signals and obtain the fault features of the fault location points.

[0032] Preferably, the process of extracting influencing parameters based on fault cause matching results, determining the probability coefficient of a power outage occurring in the distribution network's supply area during the target time period, and assessing the power supply reliability of the distribution network based on the probability coefficient includes:

[0033] S41. Based on the fault cause matching results, extract the influence parameters of the corresponding fault causes, decompose the influence parameters into quantitative feature vectors, generate topological connectivity motion equations, and construct a connectivity weakness analysis model.

[0034] S42. The secant method is used to iteratively solve the connectivity weakness analysis model to generate discrete state transition equations, and based on the discrete state transition equations, the topological connectivity state transition equations within the target time period are obtained.

[0035] S43. Based on the topological connectivity state transition equation, the cumulative sum of Gaussian white noise in the short time interval is separated as a probabilistic analysis model of topological connectivity weakness, and the power outage probability coefficient is calculated by generating the power outage link path.

[0036] S44. Based on the power outage probability coefficient, construct a reliability change surface diagram to show the evolution trend of the power outage probability coefficient in the future period, and evaluate the power supply reliability of the distribution network according to the evolution trend.

[0037] Preferably, based on the topological connectivity state transition equation, the cumulative sum of Gaussian white noise within a short time interval is separated as a probabilistic analysis model for topological connectivity weakness, and the power outage probability coefficients for generating power-loss link paths are calculated as follows:

[0038] S431. Extract transient electrical waveform sequences within a short time window based on the topological connectivity state transition equation, and use wavelet packet decomposition technology to analyze the transient electrical waveform sequences to construct a signal energy leakage map;

[0039] S432. Strip away Gaussian white noise energy segments below the kurtosis threshold from the signal energy leakage map, and use the Gaussian white noise energy segments as the reference base signal to construct an abnormal energy accumulation and sequence.

[0040] S433. Based on the accumulation and sequence of abnormal energy, analyze the effective connectivity between faulty lines and adjacent distribution network lines in any time period, and construct a probabilistic analysis model to describe the topological connectivity state.

[0041] S434. Using a probability analysis model, starting from the operation of the fault point of the faulty line in the distribution network, simulate the power outage link path of the chain fault and construct a power outage probability coefficient model to calculate the power outage probability coefficient.

[0042] Preferably, a probabilistic analysis model is used to simulate the power outage link path of a chain reaction fault, starting from the operation of the fault point of the faulty line in the distribution network, and a power outage probability coefficient model is constructed to calculate the power outage probability coefficient, including:

[0043] S4341. Use a probabilistic analysis model to determine the effective connectivity between faulty lines and remaining distribution network lines, and define the corresponding path to enter a weak connectivity degradation state when the effective connectivity is lower than a set threshold.

[0044] S4342. Generate a dynamic connectivity matrix based on the weak connectivity state, and introduce a random walk algorithm to simulate the power loss link path of a chain-like fault starting from the fault point operation of the fault line in the distribution network.

[0045] S4343. Based on the power outage link path markers, the nodes that reach the load loss are marked, and a power outage probability coefficient model is constructed to superimpose the power outage probability to form a quantitative prediction of the power outage probability in the distribution network area.

[0046] The preferred expression for the power loss probability coefficient model is:

[0047] ;

[0048] In the formula, L t Indicates in t The probability coefficient of power loss at time 10:00. a Indicates the location of the fault on the faulty line in the distribution network. e Indicates the risk weight of the power outage link path. c This represents an estimated value of the recession intensity parameter. b This represents an estimate of the scaling parameter. A ( t () indicates time t Covariates affecting the probability of power loss f This represents the estimated value of the regression parameter. h Indicates the effective connectivity. v This indicates the location of the node where the load is lost.

[0049] Secondly, the present invention also provides an FTU-based distribution network fault detection system, the system comprising:

[0050] The distribution network fault analysis module is used to analyze electrical parameters using wavelet transform technology, identify faulty lines in the distribution network, and use a traveling wave transmitter to emit traveling waves to the endpoints of the faulty lines to locate the fault location points of the faulty lines in the distribution network.

[0051] The power supply reliability assessment module is used to extract impact parameters based on the fault cause matching results, determine the probability coefficient of power outage in the distribution network supply area within the target time period, and assess the power supply reliability of the distribution network based on the probability coefficient.

[0052] The power distribution network operation control module is used to control switching equipment to perform circuit breaking operations based on power supply reliability, isolate faulty lines in the power distribution network, generate personnel dispatch instructions, and carry out emergency repairs on faulty lines in the power distribution network.

[0053] The beneficial effects of this invention are as follows:

[0054] 1. This invention utilizes an FTU to collect electrical parameters and combines wavelet transform to extract transient energy features, enabling high-resolution anomaly identification and preliminary fault location. This improves sensitivity to sudden disturbances. Furthermore, it introduces a traveling wave transmitter to actively inject high-frequency disturbances and track their reflected characteristic waves to locate the fault point. It also introduces intrinsic mode decomposition technology to dynamically extract features from the fault point waveform data, constructing a power outage probability coefficient model. This shifts the power distribution network from fault response to power outage risk prediction. In addition, it introduces quantitative evaluation of power supply reliability, which helps to form a resilient power distribution network with proactive perception and early warning. Finally, the reliability judgment results drive the field FTU to control the circuit breaker equipment, automatically isolate the faulty section, and simultaneously generate dispatch instructions to guide emergency repairs, thereby improving the accuracy and timeliness of fault handling.

[0055] 2. This invention introduces highly refined waveform data analysis and a multi-dimensional dynamic feature extraction mechanism to achieve in-depth reasoning from fault phenomena to fault causes. Furthermore, by converting the fault cause matching results into structural influence parameters and further quantifying them into feature vectors, it effectively solves the information gap problem between qualitative judgment and quantitative modeling. Combined with the connectivity weakness analysis model constructed by the topological connectivity motion equation, it calculates the probability of the occurrence of power outage link paths, realizing the prediction of power outages under the premise of faulty lines in the distribution network. This provides the dispatch center with a more refined and forward-looking reliability assessment and resource allocation reference. Attached Figure Description

[0056] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0057] Figure 1This is a flowchart of a distribution network fault detection method based on an FTU according to an embodiment of the present invention;

[0058] Figure 2 This is a schematic diagram of a power distribution network fault detection system based on an FTU according to an embodiment of the present invention.

[0059] In the picture:

[0060] 1. Distribution network fault analysis module; 2. Power supply reliability assessment module; 3. Distribution network operation control module. Detailed Implementation

[0061] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0062] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0063] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0064] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0065] Please see Figure 1 This invention provides a distribution network fault detection method based on FTU, comprising:

[0066] S1. Collect electrical parameters of the distribution network using faulty terminal equipment, analyze the electrical parameters using wavelet transform technology, calculate transient energy based on the analysis results, and identify faulty lines in the distribution network.

[0067] In one embodiment, during the identification of faulty lines in a distribution network, the method utilizes FTU devices installed on the main feeders or key nodes of the distribution network to collect three-phase current and voltage signals in real time. When the distribution network is in a normal state, these signals are stable and have uniform energy distribution. When a fault occurs, such as a short circuit, grounding, or open circuit, significant electromagnetic transient disturbances are generated at the fault point. The FTU can sample these abrupt changes at frequencies above 10kHz and record their variation curves. Wavelet transform processing is applied to the collected raw current signals. The wavelet transform synchronously expands the signals in the time and frequency domains through multi-scale analysis, preserving both the abrupt changes and the transient signals. The variable point information can be used to separate the energy distribution in different frequency bands. By statistically summing the energy of all sub-bands, a transient total energy index is formed, which will suddenly increase in a very short time when a fault occurs. An empirical reference threshold is set, and its average value and standard deviation are statistically analyzed in the normal operation data. Once the transient total energy index is detected to be greater than the empirical reference threshold, it can be determined that the line where the sampling point is located has a fault. In order to enhance the accuracy of identification, the transient total energy index collected by each FTU node can also be spatially compared. That is, if the FTU on line A shows a strong sudden change in the transient total energy index, while the fluctuations in other branches are small, the confidence that line A is a faulty line is further enhanced.

[0068] Taking a 10kV distribution network as an example, FTUs are deployed on feeders F1, F2, and F3, with a sampling frequency of 20kHz. Assuming a single-phase ground fault occurs at 13:46:12 on a certain day, the FTU on line F2 detects a strong abrupt change in the A-phase current. After wavelet decomposition (five levels), the sub-band energies are D1=8.24, D2=10.32, D3=12.88, D4=6.15, and D5=4.91, with a transient total energy index of 42.5. The empirical reference threshold under the reference operating condition is 18.3. At this point... E tot The readings significantly exceeded the threshold, therefore a fault was determined to have occurred in line F2; while the readings collected by F1 and F3... E tot The values ​​are 15.6 and 17.1 respectively, which are within the Eref range, further confirming that F2 is a faulty line.

[0069] S2. Use a traveling wave transmitter to transmit a traveling wave to the endpoint of the faulty line, extract the fault characteristic wave based on the characteristics of the reflected traveling wave, and locate the fault location of the faulty line in the distribution network according to the wave velocity of the fault characteristic wave.

[0070] In one embodiment, a traveling wave transmitter is used to transmit a traveling wave to the endpoint of the faulty line, the fault characteristic wave is extracted based on the characteristics of the reflected traveling wave, and the fault location of the distribution network faulty line is located according to the wave velocity of the fault characteristic wave, including:

[0071] S21. Select one end of the faulty line as the detection end, and use the detection end as the reference position to transmit a traveling wave to the other end of the faulty line using a traveling wave transmitter. The traveling wave propagates along the faulty line.

[0072] S22. During the propagation of the traveling wave, when the received traveling wave encounters a fault point on the faulty line, the traveling wave reflected to the detection end is used as the fault characteristic wave. Based on the fault characteristic wave, an energy propagation spectrum is constructed to obtain the reflected energy trajectory of the traveling wave.

[0073] S23. Based on the reflected energy trajectory of the traveling wave and the actual length of the faulty line, construct a traveling wave propagation inversion model to determine the round-trip time and wave speed of the fault characteristic wave from the detection end to the fault point, and locate the fault point of the faulty line in the distribution network.

[0074] It needs to be explained that when locating the fault location, the end of the line where the fault occurred is selected as the detection end. This is typically chosen near the feeder protection device or where FTUs are concentrated to ensure stable communication and control. A traveling wave transmitter installed at this end actively transmits a high-frequency traveling wave signal of a specific frequency to the other end of the line, such as a 1MHz high-frequency narrow pulse wave. The signal propagates along the distribution line in the form of an electromagnetic wave. When this traveling wave signal reaches the location of the fault in the line (such as a short circuit, grounding, or other points of sudden resistance change), strong reflection occurs due to impedance discontinuity. The reflected wave returns to the detection end along the original path. A high-speed waveform recording device records the energy distribution and time axis changes of the returned wave. By extracting the characteristic peaks in the signal waveform and filtering out non-faulty reflected signals, and constructing an energy propagation spectrum, the characteristic wave reflection trajectory truly caused by the fault point is identified. Using the known length L of the faulty line and the traveling wave propagation speed... n (Usually taken as 0.95 times the speed of light, approximately 2.85 × 10⁻⁶) 8 (m / s), combined with the time difference Δ between the initial traveling wave emitted by the detection end and the received reflected wave. t The traveling wave propagation inversion model is used to calculate the one-way distance from the fault point to the detection end. g =( n ×Δ t ) / 2, thereby accurately locating the fault location.

[0075] Taking a 10kV distribution line as an example, the total line length is... L =3.6km, the traveling wave propagation speed is set to n =2.85×10 8 m / s, at the moment of failure T 0=12:31:42.120 sends an excitation signal, in T If a significant high-energy reflected wave is received at 12:31:42.145, then the reflection time difference Δ... t =25 milliseconds=25×10 -3seconds, substitute into the formula g =(2.85×10 8 m / s×25×10 -3 )÷2=3.5625km, meaning the fault point is 3.5625km away from the detection end. Since the total length of the line is 3.6km, combined with the direction judgment, it can be deduced that the fault occurred only about 37.5 meters away from the other end. Therefore, it can complete the accurate location without relying on multi-point synchronization or topology information, and only one end needs to be detected. It is suitable for application in scenarios with complex structures and frequent topology changes in the middle or end of the distribution network, which significantly improves the fault handling speed, reduces the cost of manual investigation, and provides accurate reference location information for automated isolation and emergency repair scheduling.

[0076] S3. Obtain waveform data at the fault location point, and extract dynamic feature values ​​from the waveform data based on intrinsic mode decomposition technology. Use the dynamic feature values ​​to match the fault cause of the faulty line in the distribution network.

[0077] In one embodiment, waveform data of the fault location is acquired, and dynamic feature metrics of the waveform data are extracted based on intrinsic mode decomposition (IMD) technology. The dynamic feature metrics are then used to match the fault causes of the faulty line in the distribution network, including:

[0078] S31. Obtain the waveform signal data of the fault location point, and select a straight line with the same characteristic properties as the axial generatrix on the surface of the fault location point as the component calculation path to obtain the rate of change of the fault mode component.

[0079] S32. Based on the rate of change of fault mode components, mode decomposition technology is used to decompose and optimize the center frequency of the recorded signal data, and the mode component signal corresponding to the fault location point is generated according to the decomposition and optimization results.

[0080] S33. Analyze the mean square value, standard deviation, sample entropy, permutation entropy and singular value entropy of the modal component signals as morphological dynamic feature measures, establish a mixed domain feature group, and extract the fault features of the fault location point.

[0081] S34. Introduce the fault features into the fault cause feature bag-of-words matching network, compare the similarity between the fault features and the fault cause feature bag-of-words, and match the fault cause of the faulty line in the distribution network based on the similarity results.

[0082] This includes acquiring waveform signal data at the fault location point, selecting a straight line with the same characteristic properties as the axial generatrix on the surface of the fault location point as the component calculation path, and obtaining the rate of change of fault mode components, including:

[0083] S311. Collect waveform signal data at the fault location point of the faulty line, analyze the strain mode components corresponding to any time point at the fault location point and the previous time point, and define the signal change rate at the fault location point.

[0084] S312. Based on the signal change rate on the surface at the fault location point, select a straight line with the same characteristic properties as the axial generatrix as the component calculation path, extract the equally spaced displacement mode components, and determine the change rate of the displacement mode components.

[0085] S313. Based on the signal change rate on the surface at the fault location point, select a straight line parallel to the axial generatrix as the identification calculation path, calculate the axial displacement at any time point to obtain the identification quantity, and combine it with the displacement modal component change rate to generate the fault modal component change rate.

[0086] It needs to be explained that by extracting multi-dimensional signal features and fusing modal change rates, a precise quantitative description of the dynamic behavior at the fault location can be achieved. Three-phase voltage or current waveforms at the fault location are acquired using high-speed sampling devices (such as electronic transformers or waveform recorders). The sampling frequency is typically no less than 10kHz to ensure the capture of high-frequency transient components. For the acquired discrete signal sequence, the difference between any given time point and the previous time point is calculated, and strain modal components are constructed based on the difference sequence, representing the energy density distribution of signal change per unit time. The signal change rate is defined as reflecting the abrupt change in signal intensity at the fault location. A straight line with the same conductivity and mechanical strength distribution as the axial busbar is selected on the physical surface of the fault location (such as cable insulation or conductor cross-section) as a component meter. The calculation path must meet the same material parameters and electromagnetic field boundary conditions as the busbar. Virtual sensor nodes are arranged along this path at equal intervals (e.g., one sampling point every 5 mm). The displacement modal components (reflecting mechanical deformation or electromagnetic field distortion caused by fault disturbance) of each node are extracted through finite element simulation or measured data. The displacement change gradient between adjacent nodes is calculated, and the displacement modal component change rate of the entire path is obtained. This index is used to quantify the degree of local distortion of the physical field caused by the fault. At the same time, another straight line is selected on the surface of the fault point along a direction parallel to the busbar as the calculation path. This path is used to capture the axial propagation characteristics caused by the fault, calculate the fluctuation of axial displacement at each time point (e.g., by measuring magnetic field offset through Hall sensor or inferring deformation through infrared thermography), and calculate its standard deviation. σ As a marker, the signal change rate will ultimately be... R signal Displacement modal change rate R displacement With axial markings δ The rate of change of comprehensive fault mode components is generated by weighted fusion. R total = α · R signal + β · R displacement +γ· σ · δ The weighting coefficient α , β , γ Adaptive adjustment based on fault type (e.g., emphasizing short-circuit faults). α Mechanical fracture focuses β By fusing electrical-mechanical multi-physics field characteristics, it overcomes the limitations of single signal analysis and significantly improves the sensitivity and diagnostic robustness for complex faults.

[0087] Specifically, the mean square value, standard deviation, sample entropy, permutation entropy, and singular value entropy of the modal component signals are analyzed as morphological dynamic feature measures. A mixed domain feature set is established, and fault features at the fault location points are extracted, including:

[0088] S331. Calculate the mean square value, standard deviation, sample entropy, permutation entropy and singular value entropy of the modal component signals based on the set of equations, and determine the optimal form of the modal component signals based on the calculation results.

[0089] S332. Based on the optimal form, perform feature measurement on the modal component signals to form a mixed domain feature group of the fault location points of the faulty line, and obtain the component fault feature classification matrix based on the mixed domain feature group.

[0090] S333. Divide the component fault feature classification matrix into intrinsic mode matrices according to the complex wavelet decomposition features, obtain the clustering center vectors of the modal component signals based on the intrinsic mode matrices, and construct the feature component matrix.

[0091] S334. Based on the selection constraints and feature component matrix, perform optimal feature selection processing on the modal component signals to determine the feature extraction relationship of the modal component signals and obtain the fault features of the fault location points.

[0092] It needs to be explained that the process of matching the fault cause of a faulty line in the distribution network is based on Eigenmode Decomposition (EMD), combined with multi-scale feature extraction and modal feature clustering of the recorded signal, to achieve intelligent identification from "fault phenomenon" to "fault cause". Specifically, high sampling rate voltage or current signals are collected from the location of the faulty line in the distribution network over a continuous period of time, and the strain mode components corresponding to any time point and its previous time are analyzed to construct a signal change rate sequence. Then, on the surface of that point, along a direction with the same electrical and structural properties as the axial bus, a straight line is selected as the calculation path, and displacement mode components are extracted at equal intervals, and their change rate is calculated. Further, another straight line is selected on the same surface along a direction parallel to the bus as an identifier calculation path, axial displacement is extracted and an identifier is generated, which is combined with the aforementioned displacement mode change rate to form a composite fault mode component change rate. Based on this change rate, the sensitive frequency band of the mode component is determined, and Eigenmode Decomposition is performed on the original recorded signal to decompose it into... Several intrinsic mode functions (IMFs) with local frequency characteristics are identified. The center frequency of each IMF is adaptively optimized and filtered to eliminate low-energy or redundant components, constructing a set of modal components that best represent the dynamic behavior of faults. The mean square value, standard deviation, sample entropy, permutation entropy, and singular value entropy of the selected modal components are calculated. These metrics reflect the energy intensity, volatility, complexity, time series structure, and singular structure of the signal, respectively. After fusion, they form a hybrid domain feature group. The modal response morphology corresponding to each type of fault is further extracted based on complex wavelet decomposition. The distribution law of different types of faults in the feature space is established through the component fault feature classification matrix. The proposed feature vectors are introduced into a fault cause feature word bag matching network. The network is pre-set with feature word vectors of various typical faults (such as single-phase grounding, metallic wire breakage, insulation breakdown, etc.). The matching degree between the current feature group and the historical word bag library is compared through similarity functions such as cosine similarity or Mahalanobis distance, thereby realizing intelligent matching of fault causes.

[0093] Taking a ground fault on a 10kV line as an example, a phase A current waveform signal was collected at the fault point with a sampling frequency of 20kHz and a total duration of 200ms. Time series differencing was performed on the signal to construct a discrete signal sequence. It was found that... t A significant jump occurs between 68ms and 70ms, with a rate of change reaching 0.42A / ms. Six points are evenly spaced along the busbar direction on the insulation surface at the fault point. The calculated displacement modal change rate reaches its maximum at point 3, at 2.1mm / ms. Simultaneously, the axial displacement is calculated using line B parallel to the busbar. t=68.7ms, the value is 0.38mm / ms, and the two are combined to generate a modal component change rate sequence; the signal is decomposed into 7 IMF components by EMD, and IMF3~IMF5 are retained as effective modal components, with center frequencies of 410Hz, 210Hz and 110Hz respectively. The statistical characteristics of these 3 components are calculated, and the specific results are shown in Table 1 below:

[0094] Table 1: Results of Statistical Characteristic Calculation

[0095] Modal Mean square value Standard deviation Sample Entropy Permutation Entropy Singular value entropy IMF3 0.84 0.29 0.16 0.19 0.31 IMF4 0.66 0.22 0.13 0.15 0.28 IMF5 0.52 0.17 0.11 0.12 0.25

[0096] The above features are fused into a feature vector of length 15 and a hybrid domain feature group is constructed. Based on waveform trends and prior rules, the most representative feature triplet (standard deviation, permutation entropy, singular value entropy) is selected and the similarity is calculated to be 0.97. Therefore, the cause of this fault is finally matched as "single-phase grounding + leakage contact caused by insulation aging". Furthermore, the dynamic behavioral features extracted from the waveform data are used to establish a cognitive association of fault knowledge, which helps to form a closed-loop perception-interpretation-attribution mechanism and realize intelligent fault cognition of the distribution network.

[0097] S4. Extract the influencing parameters based on the fault cause matching results, determine the probability coefficient of power outage in the distribution network supply area during the target time period, and evaluate the power supply reliability of the distribution network based on the probability coefficient.

[0098] In one embodiment, extracting impact parameters based on fault cause matching results, determining the probability coefficient of a power outage occurring in the distribution network supply area within a target time period, and assessing the power supply reliability of the distribution network based on the probability coefficient includes:

[0099] S41. Based on the fault cause matching results, extract the influence parameters of the corresponding fault causes, decompose the influence parameters into quantitative feature vectors, generate topological connectivity motion equations, and construct a connectivity weakness analysis model.

[0100] S42. The secant method is used to iteratively solve the connectivity weakness analysis model to generate discrete state transition equations, and based on the discrete state transition equations, the topological connectivity state transition equations within the target time period are obtained.

[0101] S43. Based on the topological connectivity state transition equation, the cumulative sum of Gaussian white noise in the short time interval is separated as a probabilistic analysis model of topological connectivity weakness, and the power outage probability coefficient is calculated by generating the power outage link path.

[0102] S44. Based on the power outage probability coefficient, construct a reliability change surface diagram to show the evolution trend of the power outage probability coefficient in the future period, and evaluate the power supply reliability of the distribution network according to the evolution trend.

[0103] Specifically, in the process of obtaining the topological connectivity state transition equation within the target duration, the fault causes identified in the previous stage are transformed into structured influencing factors, and these influencing factors are mapped into quantitative feature parameters to model the connectivity weakness model. Specifically, based on the matching results (e.g., the fault cause is insulation degradation + metallic grounding), key influencing parameters are extracted, including but not limited to: fault duration, peak fault current, line load rate, node connectivity level (e.g., trunk, branch, terminal), historical fault statistical weights, etc. For example, the larger the fault current, the longer the duration, and the lower the node's level, the greater its impact on the overall network topology connectivity. The parameters are constructed into a quantitative feature vector and introduced into the topological connectivity motion equation, the mathematical form of which can be: B ( i )= C ( t )⋅ R -G⋅F ,in C ( t ) represents the time-dependent power grid structure function. G This represents the matrix of influencing weight factors. F The eigenvector represents the degree of weakening of electrical connectivity between different nodes due to fault disturbances. This equation is used to describe the degree of weakening of electrical connectivity between different nodes, and ultimately forms the initial connectivity weakness analysis model.

[0104] The connectivity weakness analysis model is solved numerically using the secant method instead of Newton's method. Because it does not rely on derivative functions, it is suitable for problems with complex topological state functions or local discontinuities. Specifically, at each time step, a linear approximation is constructed using the weakness function values ​​from the previous two steps, and the connectivity state evolution data at different times are obtained. This leads to a discrete state transition equation set, which represents the trajectory of the connectivity between topological nodes over time during the target time period.

[0105] Furthermore, it captures the dynamic evolution of the distribution network from "normal connectivity" to "weak connectivity" or even "power outage," which is beneficial for identifying the risks of "potential regional islands" or "chain-like power outages" in advance. It is suitable for assessing phenomena such as multiple nodes, weak loops, and degraded support capabilities in complex distribution network environments. In addition, the use of the secant method avoids the difficulty of differentiating high-dimensional complex functions, making the model more universal and feasible for practical engineering.

[0106] Specifically, based on the topological connectivity state transition equation, the cumulative sum of Gaussian white noise within a short time interval is separated as a probabilistic analysis model for topological connectivity weakness. This model generates power outage link paths and calculates power outage probability coefficients, including:

[0107] S431. Extract transient electrical waveform sequences within a short time window based on the topological connectivity state transition equation, and use wavelet packet decomposition technology to analyze the transient electrical waveform sequences to construct a signal energy leakage map;

[0108] S432. Strip away Gaussian white noise energy segments below the kurtosis threshold from the signal energy leakage map, and use the Gaussian white noise energy segments as the reference base signal to construct an abnormal energy accumulation and sequence.

[0109] S433. Based on the accumulation and sequence of abnormal energy, analyze the effective connectivity between faulty lines and adjacent distribution network lines in any time period, and construct a probabilistic analysis model to describe the topological connectivity state.

[0110] S434. Using a probability analysis model, starting from the operation of the fault point of the faulty line in the distribution network, simulate the power outage link path of the chain fault and construct a power outage probability coefficient model to calculate the power outage probability coefficient.

[0111] Among them, the probability analysis model is used to simulate the power outage link path of a chain reaction fault, starting from the operation of the fault point of the faulty line in the distribution network, and a power outage probability coefficient model is constructed to calculate the power outage probability coefficient, including:

[0112] S4341. Use a probabilistic analysis model to determine the effective connectivity between faulty lines and remaining distribution network lines, and define the corresponding path to enter a weak connectivity degradation state when the effective connectivity is lower than a set threshold.

[0113] S4342. Generate a dynamic connectivity matrix based on the weak connectivity state, and introduce a random walk algorithm to simulate the power loss link path of a chain-like fault starting from the fault point operation of the fault line in the distribution network.

[0114] S4343. Based on the power outage link path markers, the nodes that reach the load loss are marked, and a power outage probability coefficient model is constructed to superimpose the power outage probability to form a quantitative prediction of the power outage probability in the distribution network area.

[0115] The expression for the power loss probability coefficient model is as follows:

[0116] ;

[0117] In the formula, L t Indicates in t The probability coefficient of power loss at time 10:00. a Indicates the location of the fault on the faulty line in the distribution network. e Indicates the risk weight of the power outage link path. c This represents an estimated value of the recession intensity parameter. b This represents an estimate of the scaling parameter. A (t () indicates time t Covariates affecting the probability of power loss f This represents the estimated value of the regression parameter. h Indicates the effective connectivity. v This indicates the location of the node where the load is lost.

[0118] It needs to be explained that, based on the fault type obtained from the previous stage of fault identification, such as "single-phase grounding + insulation degradation," corresponding influencing parameters are extracted, such as fault duration, phase-to-phase short-circuit current amplitude, line insulation level, historical fault frequency, and the topological position weight of the node. These multidimensional parameters are mapped into quantified feature vectors, and then a topological connectivity equation is constructed to reflect the current flow response capability between the fault point and other nodes. A model based on connectivity weakness is then used to measure the degradation trend of electrical connectivity under fault conditions. The secant method is used to perform nonlinear iterative solution of the topological weakness analysis model, and the connectivity attenuation of each node under fault conditions is calculated piecewise using a line segment approximation method, constructing... Discrete state transition equations are used to obtain the dynamic evolution trajectory of the distribution network connectivity state within a target time period (e.g., within 1 hour after a fault occurs). Simultaneously, perturbation modeling is performed on this state transition trajectory. Transient voltage and current waveforms collected within a short time window are decomposed into multiple energy subbands by wavelet packets to construct a signal energy leakage map, which is used to identify possible transmission obstacles of electrical energy in the topology. Subsequently, Gaussian white noise segments below a set kurtosis threshold (e.g., 3.2) are stripped out and used as the basis reference signal. The accumulated abnormally high energy portion is used as the "connectivity degradation feature". Based on these accumulated quantities and connection edge weights, the electrical connectivity between the faulty line and adjacent lines is evaluated, an effective connectivity matrix is ​​generated, and a probabilistic analysis model is constructed.

[0119] A chain-like power outage path is constructed starting from the fault point using a probabilistic analysis model. A random walk algorithm simulates fault propagation. When the connectivity is lower than a set threshold (e.g., 0.65), the corresponding path is marked as entering a weak connectivity degradation state. Furthermore, potential power outage links are identified through a dynamic connectivity matrix. Then, the set of nodes experiencing load loss within a certain period is statistically analyzed. Based on the probability of each link and the load weight of each node, the regional power outage probability coefficient is calculated. Based on this probability coefficient, a three-dimensional reliability change surface map is constructed to show the reliability evolution trend under different times, node locations, and grid conditions, providing decision support for early warning and resource allocation in the dispatch center.

[0120] Meanwhile, assuming a 10kV distribution network in a certain area as an example, the current identified cause of the fault on line L2 is "metallic single-phase grounding + aging short circuit". The influencing parameters are extracted as follows: short circuit current amplitude. I s =480A, duration T s=0.18s, the historical monthly average failure frequency of node K3 is 0.7 times / month, L2 belongs to the level 3 branch in the topology, and the topology transmission coefficient is... o =0.62, forming the initial feature vector. V =[480,0.18,0.7,3,0.62], and then construct the connectivity weakness function. p ( t )= ow -kt ,in k =0.03, w Representing a constant, using the secant method... t =0 to t =60 minutes of piecewise approximation to obtain the state transition equation; after wavelet packet decomposition, it was found that the energy leakage in the frequency band (200-400Hz) between 10-20ms of L2 node was significant, with a peak value of 6.3kJ, corresponding to a Gaussian white noise reference basis of only 1.4kJ and a kurtosis of 3.4. An abnormal energy accumulation sequence was constructed, and it was found that the connectivity of nodes such as K3, K4, and K6 decreased to 0.54, 0.47, and 0.43, respectively, entering a weakly degenerate state. Subsequently, the link path L2→K3→K6 was constructed. Based on the propagation probability of each path and the node load, p1=0.63, p2=0.57, w1=1.2MW, and w2=0.8MW respectively, the power outage probability coefficient was calculated to be 1.364. Finally, the reliability change surface plot showed that the power outage risk of L2 area continued to be higher than the set safety line (0.8) in the next 2 hours, so the scheduling strategy warning was automatically triggered, and it was recommended to start the L4 backup branch power supply.

[0121] S5. Based on power supply reliability, use faulty terminal equipment to control switching equipment to perform circuit breaking operations, isolate faulty lines in the distribution network, and generate personnel dispatch instructions to carry out emergency repairs on faulty lines in the distribution network.

[0122] In one embodiment, based on power supply reliability, fault terminal equipment is used to control switching equipment to perform circuit breaking operations, isolate faulty lines in the distribution network, and generate personnel dispatch instructions to carry out emergency repairs on faulty lines in the distribution network. This includes: when power supply reliability analysis determines that the current fault has a high risk of power loss or its cascading effects may lead to a large-scale power outage, a circuit breaking command is immediately triggered by fault terminal equipment (such as feeder terminal unit, switch control unit, or substation remote terminal unit). This command is sent to the target circuit breaker, switch, or sectionalizer through the distribution network automation system (DMS) to achieve electrical isolation of the faulty line, cut off its connection path with the main network or other branches, and prevent the fault from spreading further. After isolation, the spatial coordinates of the faulty section and the unique identifier of the equipment involved are automatically matched with the asset archive based on the GIS geographic information system, and targeted emergency repair instructions are generated in combination with the emergency repair resource scheduling model.

[0123] Please see Figure 2 The present invention also provides an FTU-based distribution network fault detection system, the system comprising:

[0124] The distribution network fault analysis module 1 is used to analyze electrical parameters using wavelet transform technology, identify faulty lines in the distribution network, and use a traveling wave transmitter to transmit traveling waves to the endpoints of the faulty lines to locate the fault location points of the faulty lines in the distribution network.

[0125] The power supply reliability assessment module 2 is used to extract impact parameters based on the fault cause matching results, determine the probability coefficient of power outage in the distribution network power supply area during the target time period, and assess the power supply reliability of the distribution network based on the probability coefficient.

[0126] The distribution network operation control module 3 is used to control the switching equipment to perform circuit breaking operations based on power supply reliability, isolate the faulty lines in the distribution network, generate personnel dispatch instructions, and carry out emergency repairs on the faulty lines in the distribution network.

[0127] The power distribution network fault analysis module 1, the power supply reliability assessment module 2, and the power distribution network operation control module 3 are connected in sequence.

[0128] Those skilled in the art will recognize that the units and algorithm steps described in conjunction with the embodiments herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0129] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for detecting distribution network faults based on FTU, characterized in that, include: S1. The electrical parameters of the distribution network are collected using fault terminal equipment, and the electrical parameters are analyzed using wavelet transform technology. The transient energy is calculated based on the analysis results, and the faulty line of the distribution network is identified. The fault terminal equipment is an FTU. S2. Use a traveling wave transmitter to transmit a traveling wave to the end of the faulty line, extract the fault characteristic wave based on the characteristics of the reflected traveling wave, and locate the fault location of the faulty line in the distribution network according to the wave velocity of the fault characteristic wave. S3. Obtain waveform data at the fault location point, and extract dynamic feature metric values ​​from the waveform data based on intrinsic mode decomposition technology. Use the dynamic feature metric values ​​to match the fault cause of the faulty line in the distribution network. S4. Extract the impact parameters based on the fault cause matching results, determine the probability coefficient of power outage in the power supply area of ​​the distribution network during the target time period, and evaluate the power supply reliability of the distribution network based on the probability coefficient. S5. Based on power supply reliability, use faulty terminal equipment to control switching equipment to perform circuit breaking operations, isolate faulty lines in the distribution network, and generate personnel dispatch instructions to carry out emergency repairs on faulty lines in the distribution network. The process of extracting impact parameters based on fault cause matching results, determining the probability coefficient of a power outage occurring in the distribution network supply area within the target time period, and assessing the power supply reliability of the distribution network based on the probability coefficient includes: S41. Based on the fault cause matching results, extract the influence parameters of the corresponding fault causes, decompose the influence parameters into quantitative feature vectors, generate topological connectivity motion equations, and construct a connectivity weakness analysis model. S42. The secant method is used to iteratively solve the connectivity weakness analysis model to generate discrete state transition equations, and based on the discrete state transition equations, the topological connectivity state transition equations within the target time period are obtained. S43. Based on the topological connectivity state transition equation, the cumulative sum of Gaussian white noise in the short time interval is separated as a probabilistic analysis model of topological connectivity weakness, and the power outage probability coefficient is calculated by generating the power outage link path. S44. Based on the power outage probability coefficient, construct a reliability change surface diagram to show the evolution trend of the power outage probability coefficient in the future period, and evaluate the power supply reliability of the distribution network according to the evolution trend.

2. The method for distribution network fault detection based on FTU according to claim 1, characterized in that, The process of using a traveling wave transmitter to emit a traveling wave towards the endpoint of the faulty line, extracting the fault characteristic wave based on the characteristics of the reflected traveling wave, and locating the fault location point of the distribution network faulty line according to the wave velocity of the fault characteristic wave includes: S21. Select one end of the faulty line as the detection end, and use the detection end as the reference position to transmit a traveling wave to the other end of the faulty line using a traveling wave transmitter. The traveling wave propagates along the faulty line. S22. During the propagation of the traveling wave, when the received traveling wave encounters a fault point on the faulty line, the traveling wave reflected to the detection end is used as the fault characteristic wave. Based on the fault characteristic wave, an energy propagation spectrum is constructed to obtain the reflected energy trajectory of the traveling wave. S23. Based on the reflected energy trajectory of the traveling wave and the actual length of the faulty line, construct a traveling wave propagation inversion model to determine the round-trip time and wave speed of the fault characteristic wave from the detection end to the fault point, and locate the fault point of the faulty line in the distribution network.

3. The method for detecting distribution network faults based on FTU according to claim 1, characterized in that, The process of acquiring waveform data at the fault location point, extracting dynamic feature metrics from the waveform data based on intrinsic mode decomposition (IMD) technology, and using these dynamic feature metrics to match the fault causes of the distribution network fault lines includes: S31. Obtain the waveform signal data of the fault location point, and select a straight line with the same characteristic properties as the axial generatrix on the surface of the fault location point as the component calculation path to obtain the rate of change of the fault mode component. S32. Based on the rate of change of fault mode components, mode decomposition technology is used to decompose and optimize the center frequency of the recorded signal data, and the mode component signal corresponding to the fault location point is generated according to the decomposition and optimization results. S33. Analyze the mean square value, standard deviation, sample entropy, permutation entropy and singular value entropy of the modal component signals as morphological dynamic feature measures, establish a mixed domain feature group, and extract the fault features of the fault location point. S34. Introduce the fault features into the fault cause feature bag-of-words matching network, compare the similarity between the fault features and the fault cause feature bag-of-words, and match the fault cause of the faulty line in the distribution network based on the similarity results.

4. The method for distribution network fault detection based on FTU according to claim 3, characterized in that, The process of acquiring the waveform signal data at the fault location point and selecting a straight line with the same characteristic properties as the axial generatrix on the surface of the fault location point as the component calculation path to obtain the rate of change of the fault mode components includes: S311. Collect waveform signal data at the fault location point of the faulty line, analyze the strain mode components corresponding to any time point at the fault location point and the previous time point, and define the signal change rate at the fault location point. S312. Based on the signal change rate on the surface at the fault location point, select a straight line with the same characteristic properties as the axial generatrix as the component calculation path, extract the equally spaced displacement mode components, and determine the change rate of the displacement mode components. S313. Based on the signal change rate on the surface at the fault location point, select a straight line parallel to the axial generatrix as the identification calculation path, calculate the axial displacement at any time point to obtain the identification quantity, and combine it with the displacement modal component change rate to generate the fault modal component change rate.

5. The method for detecting distribution network faults based on FTU according to claim 4, characterized in that, The mean square value, standard deviation, sample entropy, permutation entropy, and singular value entropy of the analyzed modal component signals are used as morphological dynamic feature measures to establish a mixed domain feature set. The fault features extracted from the fault location points include: S331. Calculate the mean square value, standard deviation, sample entropy, permutation entropy and singular value entropy of the modal component signals based on the set of equations, and determine the optimal form of the modal component signals based on the calculation results. S332. Based on the optimal form, perform feature measurement on the modal component signals to form a mixed domain feature group of the fault location points of the faulty line, and obtain the component fault feature classification matrix based on the mixed domain feature group. S333. Divide the component fault feature classification matrix into intrinsic mode matrices according to the complex wavelet decomposition features, obtain the clustering center vectors of the modal component signals based on the intrinsic mode matrices, and construct the feature component matrix. S334. Based on the selection constraints and feature component matrix, perform optimal feature selection processing on the modal component signals to determine the feature extraction relationship of the modal component signals and obtain the fault features of the fault location points.

6. The method for detecting distribution network faults based on FTU according to claim 5, characterized in that, The probabilistic analysis model based on the topological connectivity state transition equation, which separates the cumulative sum of Gaussian white noise within a short time interval as the topological connectivity weakness, generates power outage probability coefficients for power-loss link paths, including: S431. Extract transient electrical waveform sequences within a short time window based on the topological connectivity state transition equation, and use wavelet packet decomposition technology to analyze the transient electrical waveform sequences to construct a signal energy leakage map; S432. Strip away Gaussian white noise energy segments below the kurtosis threshold from the signal energy leakage map, and use the Gaussian white noise energy segments as the reference base signal to construct an abnormal energy accumulation and sequence. S433. Based on the accumulation and sequence of abnormal energy, analyze the effective connectivity between faulty lines and adjacent distribution network lines in any time period, and construct a probabilistic analysis model to describe the topological connectivity state. S434. Using a probability analysis model, starting from the operation of the fault point of the faulty line in the distribution network, simulate the power outage link path of the chain fault and construct a power outage probability coefficient model to calculate the power outage probability coefficient.

7. The method for distribution network fault detection based on FTU according to claim 6, characterized in that, The method of using a probabilistic analysis model to simulate the power outage link path of a chain reaction fault, starting from the fault point operation of the faulty line in the distribution network, and constructing a power outage probability coefficient model to calculate the power outage probability coefficient includes: S4341. Use a probabilistic analysis model to determine the effective connectivity between faulty lines and remaining distribution network lines, and define the corresponding path to enter a weak connectivity degradation state when the effective connectivity is lower than a set threshold. S4342. Generate a dynamic connectivity matrix based on the weak connectivity state, and introduce a random walk algorithm to simulate the power loss link path of a chain-like fault starting from the fault point operation of the fault line in the distribution network. S4343. Based on the power outage link path markers, the nodes that reach the load loss are marked, and a power outage probability coefficient model is constructed to superimpose the power outage probability to form a quantitative prediction of the power outage probability in the distribution network area.

8. The method for detecting distribution network faults based on FTU according to claim 7, characterized in that, The expression for the power loss probability coefficient model is as follows: In the formula, L t Indicates in t The probability coefficient of power loss at time 10:

00. a Indicates the location of the fault on the faulty line in the distribution network. e Indicates the risk weight of the power outage link path. c This represents an estimated value of the recession intensity parameter. b This represents an estimate of the scaling parameter. A ( t () indicates time t Covariates affecting the probability of power loss f This represents the estimated value of the regression parameter. h Indicates the effective connectivity. v This indicates the location of the node where the load is lost.

9. A distribution network fault detection system based on an FTU, used to implement the distribution network fault detection method based on an FTU as described in any one of claims 1-8, characterized in that, The system includes: The distribution network fault analysis module is used to analyze electrical parameters using wavelet transform technology, identify faulty lines in the distribution network, and use a traveling wave transmitter to emit traveling waves to the endpoints of the faulty lines to locate the fault location points of the faulty lines in the distribution network. The power supply reliability assessment module is used to extract impact parameters based on the fault cause matching results, determine the probability coefficient of power outage in the distribution network supply area within the target time period, and assess the power supply reliability of the distribution network based on the probability coefficient. The power distribution network operation control module is used to control switching equipment to perform circuit breaking operations based on power supply reliability, isolate faulty lines in the power distribution network, generate personnel dispatch instructions, and carry out emergency repairs on faulty lines in the power distribution network.

Images