Fault monitoring and locating method for high voltage drop-out fuse

Abstract

The application discloses a fault monitoring and positioning method for a high-voltage drop-out fuse, and relates to the technical field of power equipment fault monitoring.The method comprises the following steps: acquiring original monitoring sequences corresponding to current and voltage sensors arranged at the upper and lower ports of a target fuse; activating a working mode identifier to perform working mode identification of the target fuse, establishing a global mode label, performing adaptive matching of a mode fingerprint library, and establishing a standard comparison fingerprint with a trust label; establishing a residual spectrum feature; establishing a fault discrimination vector, using the fault discrimination vector to perform spatial fault positioning, and generating a fault positioning result.The application solves the technical problems of insufficient accuracy, lack of visual presentation capability, and low monitoring and positioning efficiency of the prior art in the fault monitoring and positioning of high-voltage drop-out fuses, achieves precise positioning and visual presentation of the faults of high-voltage drop-out fuses, and improves the accuracy and efficiency of fault monitoring and positioning.

Description

Technical Field

[0001] This invention relates to the field of power equipment fault monitoring technology, and specifically to a fault monitoring and location method for high-voltage drop-out fuses. Background Technology

[0002] High-voltage drop-out fuses, as critical protection devices in distribution networks, are widely used in 10kV and below line branches and distribution transformers. Their operational status is crucial to the reliability of power supply. Currently, fault monitoring of these devices mostly relies on manual inspections or traditional single-parameter monitoring methods. Manual inspections are limited by terrain and inspection cycles, making it difficult to detect hidden faults such as poor contact and insulation aging in a timely manner. Furthermore, fault location depends on the experience of maintenance personnel, which can easily lead to misjudgments. Traditional monitoring technologies only collect single parameters such as current or voltage, lacking dynamic identification of operating modes and failing to distinguish between normal load fluctuations and fault signals, resulting in a high rate of false alarms and missed alarms. In addition, existing technologies struggle to combine equipment parameters with operational scenario adaptation standards, lacking accurate basis for fault diagnosis, and cannot be visualized by associating with the power grid GIS topology. This makes it difficult for maintenance personnel to promptly grasp the location and impact of faults, thus prolonging fault handling time and affecting the stability of power supply.

[0003] Existing technologies suffer from insufficient accuracy in fault monitoring and location of high-voltage drop-out fuses, lack of visualization capabilities, and low monitoring and location efficiency. Summary of the Invention

[0004] This application provides a fault monitoring and location method for high-voltage drop-out fuses, which addresses the technical problems of insufficient accuracy, lack of visualization capabilities, and low monitoring and location efficiency in the prior art for fault monitoring and location of high-voltage drop-out fuses.

[0005] In view of the above problems, this application provides a fault monitoring and location method for high-voltage drop-out fuses.

[0006] A first aspect of this application provides a fault monitoring and location method for high-voltage drop-out fuses, the method comprising:

[0007] Obtain the original monitoring sequences corresponding to the current and voltage sensors deployed at the upper and lower ports of the target fuse; activate the working mode recognizer to perform working mode recognition of the target fuse, establish a global mode label, perform adaptation and matching of the mode fingerprint library according to the device parameters of the target fuse and the global mode label, and establish a standard comparison fingerprint with a trust identifier; use the standard comparison fingerprint to perform dynamic analysis and comparison of the original monitoring sequences, and establish residual spectrum features; establish a fault discrimination vector according to the residual spectrum features and the trust identifier, use the fault discrimination vector to perform spatial fault location, and generate fault location results based on power grid GIS topology data.

[0008] A second aspect of this application provides a fault monitoring and location system for high-voltage drop-out fuses, the system comprising:

[0009] The module for establishing the original monitoring sequence is used to acquire the original monitoring sequences corresponding to the current and voltage sensors deployed at the upper and lower ports of the target fuse. The module for establishing the global mode label is used to activate the working mode recognizer to perform working mode recognition of the target fuse, establish a global mode label, and perform adaptation and matching of the mode fingerprint library according to the device parameters of the target fuse and the global mode label to establish a standard comparison fingerprint with a trust identifier. The module for establishing the residual spectrum feature is used to perform dynamic analysis and comparison of the original monitoring sequence using the standard comparison fingerprint to establish residual spectrum features. The module for generating the fault location result is used to establish a fault discrimination vector according to the residual spectrum feature and the trust identifier, perform spatial fault location using the fault discrimination vector, and generate a fault location result based on the power grid GIS topology data.

[0010] A third aspect of this application provides an electronic device comprising: a processor; and a memory for storing processor-executable instructions; wherein the processor is configured to execute the fault monitoring and location method for high-voltage drop-out fuses provided in this application.

[0011] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0012] The process involves acquiring the original monitoring sequence; activating the operating mode recognizer to identify the target fuse's operating mode, establishing a global mode label, and performing adaptation and matching with the mode fingerprint database based on the target fuse's equipment parameters and the global mode label to establish a standard comparison fingerprint with a trust identifier; using the standard comparison fingerprint to perform dynamic analysis and comparison of the original monitoring sequence to establish residual spectrum features; and performing spatial fault location, generating fault location results based on power grid GIS topology data. This achieves the technical effect of accurately locating and visualizing high-voltage drop-out fuse faults, improving the accuracy and efficiency of fault monitoring and location. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a schematic flowchart of a fault monitoring and location method for high-voltage drop-out fuses provided in an embodiment of this application.

[0015] Figure 2 This is a schematic diagram of the fault monitoring and location system for high-voltage drop-out fuses provided in an embodiment of this application.

[0016] Figure 3 This is a schematic diagram of the structure of an electronic device provided in this application.

[0017] Figure labeling: Original monitoring sequence establishment module 10, global pattern label establishment module 20, residual spectrum feature establishment module 30, fault location result generation module 40, processor 21, memory 22, input device 23, output device 24. Detailed Implementation

[0018] This application provides a fault monitoring and location method for high-voltage drop-out fuses, which addresses the technical problems of insufficient accuracy, lack of visualization capabilities, and low monitoring and location efficiency in the prior art for fault monitoring and location of high-voltage drop-out fuses.

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

[0020] Example 1, as Figure 1 As shown, this application provides a fault monitoring and location method for high-voltage drop-out fuses, the method comprising:

[0021] Step S100: Obtain the original monitoring sequences corresponding to the current and voltage sensors deployed at the upper and lower ports of the target fuse.

[0022] Specifically, monitoring components are deployed at key operational nodes of high-voltage drop-out fuses to collect operational behavior data under different operating conditions. These conditions cover scenarios such as normal power supply operation, operation under grid load fluctuations, and operation under minor current and voltage anomalies. The collected data includes real-time current change curves, dynamic voltage fluctuation trends, time series of fuse switching actions, and power changes under different loads. The collected raw operational behavior data is then preprocessed. First, data cleaning techniques are used to remove duplicate and invalid data. Then, filtering algorithms are used to remove high-frequency interference from the grid and noise generated by equipment vibration. Next, the preprocessed data is standardized to unify the data format and dimensions. Finally, according to the correspondence between operating condition type and operating parameter characteristics, the standardized operational behavior data is categorized and integrated. For example, current and voltage stability data under normal operating conditions are grouped into one category, and parameter change data under load fluctuation conditions are grouped into another category, forming a set of operating modes containing multiple typical operating states. For the established set of operating modes, dynamic current-voltage characteristics reflecting the core operating state of the fuse are extracted for each mode. These characteristics include peak current, valley current, stable voltage operating range, phase difference between current and voltage, magnitude and duration of current surges under different operating conditions, and trends and amplitudes of voltage dips or rises. The extracted dynamic current-voltage characteristics are then quantified and encoded, transforming each feature into a calculable and comparable numerical vector, ensuring each operating mode corresponds to a unique numerical vector as its exclusive mode fingerprint. Finally, according to the type of operating mode, such as normal operating mode, load fluctuation mode, and minor abnormality mode, all mode fingerprints are categorized, organized, and archived to construct a well-structured and characteristic mode fingerprint library. This library will provide core data support for subsequent target fuse operating mode identification and standard comparison fingerprint matching.

[0023] High-precision current and voltage sensors are deployed at the upper and lower ports of the target high-voltage drop-out fuse to ensure that the sampling frequency and range of the sensors are adapted to the rated operating parameters of the target fuse, so as to accurately capture the current and voltage signals during operation. After the sensors are deployed, the current and voltage monitoring signals they collect are read in real time. First, noise reduction processing is performed on both types of signals to filter out irrelevant signals such as electromagnetic interference and equipment vibration noise in the power grid environment. Then, the noise-reduced current and voltage signals are time-aligned using timestamp synchronization technology to ensure that the current and voltage data at the same moment are accurately matched. Subsequently, local adaptive signal anomaly authentication is performed on the time-aligned monitoring signals to identify and eliminate false anomaly signals caused by temporary sensor failures, instantaneous electromagnetic pulses, etc. Finally, the effective monitoring signals after noise reduction, time alignment, and anomaly authentication are integrated in chronological order to establish an original monitoring sequence that can truly and continuously reflect the operating status of the target fuse.

[0024] Step S200: Activate the working mode recognizer to perform working mode recognition of the target fuse, establish a global mode label, perform adaptation matching of the mode fingerprint library according to the device parameters of the target fuse and the global mode label, and establish a standard comparison fingerprint with a trust identifier.

[0025] Specifically, the built-in operating mode recognizer is activated, and the established original monitoring sequence is input into the recognizer. The recognizer determines the current actual operating mode of the target fuse by calculating the feature similarity between the original monitoring sequence and each operating mode in the established operating mode set, such as the matching degree of current change trend and voltage stability range, and generates a unique corresponding global mode label to clarify the current operating status category. Next, the equipment parameters of the target fuse are retrieved. These parameters include core configuration information such as the fuse's rated current, rated voltage, insulation class, and factory calibration benchmark value. These equipment parameters and the generated global mode label are used together as search conditions to perform an adaptation matching operation in the constructed mode fingerprint library, and the mode fingerprint that best matches the current equipment configuration and operating mode is selected. Finally, based on the feature overlap calculated during the adaptation and matching process, such as the matching ratio of key current-voltage dynamic features, trust labels are added to the selected pattern fingerprints. For example, feature overlap of 95% or above is labeled as high trust, 80% to 95% as medium trust, and below 80% as low trust. This ultimately forms a standard comparison fingerprint with trust labels, providing a benchmark for the dynamic analysis and comparison of the original monitoring sequences.

[0026] Step S300: Use the standard comparison fingerprint to perform dynamic analysis and comparison of the original monitoring sequence to establish residual spectrum features.

[0027] Specifically, using the obtained standard comparison fingerprint with trust identifier as a benchmark, a multi-dimensional dynamic analysis and comparison are performed on the established original monitoring sequence. First, within a preset sliding window, the voltage drop amplitude difference and current instantaneous slope difference between the original monitoring sequence and the standard comparison fingerprint are calculated to construct a time-domain residual vector that reflects the time-domain differences. Next, Fast Fourier Transform is performed on the original monitoring sequence and the standard comparison fingerprint respectively to extract their fundamental frequency and odd harmonic amplitudes. The corresponding amplitude difference is calculated at each frequency point to form a frequency-domain residual vector that reflects the frequency-domain differences. Subsequently, the energy distribution of the original monitoring sequence and the standard comparison fingerprint within each time window is obtained through Short-Time Fourier Transform, and the energy distribution difference between the two is calculated to generate a time-frequency residual vector that characterizes the time-frequency domain differences. After constructing the three types of residual vectors, they are first normalized to unify the data dimensions. Then, the standard deviation of each residual dimension is calculated in combination with historical monitoring data. The normalized residual vectors are then standardized and weighted according to the standard deviation. Finally, the weighted time-domain, frequency-domain, and time-frequency residual vectors are concatenated to establish residual spectral features that can comprehensively reflect the differences between the original monitoring sequence and the standard fingerprint.

[0028] Step S400: Establish a fault discrimination vector based on the residual spectrum features and the trust identifier, use the fault discrimination vector to locate spatial faults, and generate fault location results based on power grid GIS topology data.

[0029] Specifically, the obtained trust identifiers are transformed into specific trust factors, such as high trust identifiers corresponding to high trust factors, medium trust identifiers corresponding to medium trust factors, and low trust identifiers corresponding to low trust factors. These trust factors are then used to weight and fuse the established residual spectrum features, strengthening the influence of key residual dimensions under high trust identifiers and weakening residual information that may contain errors under low trust identifiers, thereby constructing a weighted residual vector. This weighted residual vector is then input into a preset fault discriminator. Based on built-in fault feature thresholds, such as the residual exceedance ratio and parameter mutation amplitude, the fault discriminator outputs a fault discrimination vector containing the probability of fault occurrence and a preliminary judgment of the fault type. Next, the current and voltage difference information of other fuses within the neighborhood of the target fuse is obtained, such as the current difference and voltage difference between neighboring fuses and the target fuse. This is combined with the fault discrimination vector to construct a spatial fault matrix. Matrix analysis is used to determine the spatial correlation of fault features. Finally, the spatial fault matrix is ​​used to perform spatial consistency verification, verifying whether the fault features conform to a reasonable propagation path between devices in the power grid topology, eliminating isolated abnormal signal interference, and completing the spatial fault location. Next, a corresponding verification window is configured according to the fault level. Additional operational data of the target fuse is collected within the window, such as current and voltage fluctuations and equipment temperature during additional periods. The preliminary fault location results are verified using the additional dataset, verification feedback is generated, and the fault location results are updated accordingly to ensure location accuracy. Finally, the final fault location results are mapped to the power grid GIS topology data to generate a distribution map that includes the visualized fault location, fault type (such as fuse blowout, poor contact, insulation breakdown, etc.), fault confidence, maintenance priority, and fault propagation hazard indicators. This distribution map is then pushed to the power grid monitoring system according to preset rules to implement fault early warning and dispatch management, providing maintenance personnel with a clear basis for fault handling.

[0030] In one possible implementation, step S300 further includes:

[0031] Step S310: Calculate the voltage drop amplitude difference and current instantaneous slope difference between the original monitoring sequence and the standard fingerprint within a preset sliding window, and establish a time-domain residual vector.

[0032] Step S320: Perform a fast Fourier transform on the original monitoring sequence and the standard comparison fingerprint to extract the fundamental frequency and odd harmonic amplitude.

[0033] Step S330: Calculate the amplitude difference between the original monitoring sequence and the standard comparison fingerprint at each frequency point, and establish the frequency domain residual vector.

[0034] Step S340: Perform short-time Fourier transform on the original monitoring sequence and the standard comparison fingerprint, calculate the energy distribution difference of each time window, and generate a time-frequency residual vector.

[0035] Step S350: Perform vector concatenation of the time-domain residual vector, frequency-domain residual vector, and time-frequency residual vector to establish residual spectrum features.

[0036] Specifically, based on the fluctuation characteristics of the current and voltage signals of high-voltage drop-out fuses, a suitable preset sliding window is set. The window duration must balance the accuracy of signal change capture and computational efficiency to ensure coverage of short-term signal fluctuation cycles. Then, within this sliding window, voltage and current data for corresponding time periods in the original monitoring sequence and the standard comparison fingerprint are simultaneously captured. For the captured voltage data, the voltage difference between the original monitoring sequence and the standard comparison fingerprint at the same time point is calculated, and the voltage drop amplitude difference reflecting voltage anomalies is extracted. For the captured current data, the instantaneous current slope is obtained by calculating the ratio of current changes at adjacent time points. The instantaneous current slope difference is then compared with that of the original monitoring sequence and the standard comparison fingerprint to obtain the instantaneous current slope difference. Finally, the calculated voltage drop amplitude difference and instantaneous current slope difference are arranged chronologically to form a time-domain residual vector that intuitively reflects the time-domain differences between the original monitoring sequence and the standard comparison fingerprint.

[0037] Complete current-voltage signal data from the established original monitoring sequence and the obtained standard comparison fingerprint were retrieved separately. Fast Fourier Transform (FFT) processing was performed on both types of signals to convert the continuous signals originally in the time domain into frequency domain signals, achieving a conversion from the time-amplitude dimension to the frequency-amplitude dimension. In the converted frequency domain signal, the fundamental frequency (typically 50Hz), consistent with the grid's rated operating frequency, was first identified and extracted. This amplitude directly reflects the signal's energy intensity at the main frequency. Subsequently, odd harmonic components with frequencies 3, 5, and 7 times the fundamental frequency were further screened, and their corresponding amplitudes were extracted. These odd harmonic amplitudes reflect the frequency distortion characteristics caused by equipment anomalies, such as poor contact or component aging. Finally, the fundamental frequency and odd harmonic amplitudes of the original monitoring sequence and the standard comparison fingerprint were extracted, providing data support for subsequent frequency domain difference analysis.

[0038] First, the fundamental frequency and odd harmonic amplitude data of the extracted original monitoring sequence and standard comparison fingerprint are processed, identifying all frequency points corresponding to the two types of data, including the fundamental frequency and each odd harmonic frequency point, ensuring a one-to-one correspondence between the frequency points of the two types of data. Then, for each frequency point, the difference between the amplitude of the original monitoring sequence at that frequency point and the amplitude of the standard comparison fingerprint at the same frequency point is calculated. If the amplitude of the original monitoring sequence is greater than that of the standard comparison fingerprint, the value is positive; otherwise, it is negative. The absolute value of the difference reflects the degree of amplitude difference at that frequency point. Finally, the amplitude differences of all frequency points are arranged in ascending order of frequency, forming a frequency domain residual vector that systematically reflects the differences between the original monitoring sequence and the standard comparison fingerprint at the frequency domain level. Each element in this vector corresponds to the amplitude difference at a single frequency point.

[0039] Short-time Fourier transform (SFT) processing was performed on the complete current-voltage signals from the established original monitoring sequence and the obtained standard comparison fingerprint. A fixed-length sliding time window was set, its length adapted to the short-term variation characteristics of the signal, ensuring both the capture of instantaneous signal fluctuations and frequency resolution. The two types of long-time domain signals were divided into multiple continuous short-time signal segments. A Fourier transform was then performed on each short-time signal segment individually, yielding energy distribution matrices of the original monitoring sequence and the standard comparison fingerprint at different time windows and frequencies. Each element in the matrix represents the signal energy value at the corresponding time window and frequency. Subsequently, for each identical time window and frequency, the difference between the corresponding element in the energy distribution matrix of the original monitoring sequence and the corresponding element in the energy distribution matrix of the standard comparison fingerprint was calculated, obtaining the energy distribution difference at different frequencies within each time window. Finally, the energy distribution differences of all time windows were integrated according to the time window order-frequency order rule to generate a time-frequency residual vector that simultaneously reflects the differences in the signal in both the time and frequency dimensions. This vector accurately reflects the abnormal energy changes of the original monitoring sequence relative to the standard comparison fingerprint in the time and frequency domains.

[0040] First, the obtained time-domain residual vector, frequency-domain residual vector, and time-frequency residual vector are normalized by mapping the element values ​​of each vector to the same numerical range, such as 0-1, to eliminate the numerical range deviation caused by the difference in physical dimensions between different vectors. Then, combined with historical fault monitoring data, the correlation between each dimension element of the three types of residual vectors—that is, the correlation between individual residual values ​​and fault characteristics—is analyzed. Elements with strong correlations are assigned higher weights, and elements with weak correlations are assigned lower weights, completing the weighted optimization of the vectors. Finally, according to a preset splicing order, such as first the time-domain residual vector, then the frequency-domain residual vector, and finally the time-frequency residual vector, the three types of weighted residual vectors are concatenated end-to-end to form a complete vector. This vector integrates the difference information of the time domain, frequency domain, and time-frequency domain, together forming a residual spectrum feature that comprehensively reflects the difference between the original monitoring sequence and the standard comparison fingerprint, providing a comprehensive feature basis for subsequent fault identification.

[0041] In one possible implementation, step S350 further includes:

[0042] Step S351: Normalize the time-domain residual vector, frequency-domain residual vector, and time-frequency residual vector.

[0043] Step S352: Calculate the standard deviation of each residual dimension based on historical monitoring data, perform standardized weighting based on the standard deviation, and generate residual spectrum features.

[0044] Specifically, normalization is performed on the obtained time-domain residual vector, frequency-domain residual vector, and time-frequency residual vector. By calculating the maximum and minimum values ​​of the elements in each vector, each element value is mapped proportionally to a preset standard numerical range, such as 0-1. This eliminates the inconsistency in the dimensions and numerical ranges caused by differences in physical meaning, such as amplitude difference in the time domain and frequency amplitude difference in the frequency domain, ensuring that the elements of the three types of residual vectors are comparable.

[0045] First, historical monitoring data of the same model and operating conditions as the target fuse are retrieved from the historical database, and the corresponding time-domain, frequency-domain, and time-frequency residual vectors are extracted as a sample set. For each residual dimension in the sample set, that is, the dimension corresponding to all individual residual values ​​in the three types of vectors, the standard deviation of each residual dimension is obtained by calculating the square root of the average of the squared differences between the residual values ​​of all samples in that dimension and the mean. Then, the reciprocal of the standard deviation is used as the weight coefficient. The smaller the standard deviation, the larger the weight coefficient. The elements of the corresponding dimensions in the normalized time-domain residual vector, frequency-domain residual vector, and time-frequency residual vector are weighted respectively. Finally, the weighted three types of residual vectors are concatenated in a fixed order of time-domain-frequency-time-frequency to form a residual spectrum feature that integrates multi-dimensional difference information and highlights the characteristics of highly sensitive faults.

[0046] In one possible implementation, step S400 further includes:

[0047] Step S410: After converting the trust identifier into a trust factor, perform weighted fusion of the corresponding residual spectrum features to establish a weighted residual vector.

[0048] Step S420: Input the weighted residual vector into the fault discriminator to establish a fault discrimination vector.

[0049] Step S430: Obtain the current-voltage differential information of the neighboring fuses, and establish a spatial fault matrix based on the current-voltage differential information and the fault discrimination vector.

[0050] Step S440: Use the spatial fault matrix to perform spatial consistency verification and complete spatial fault location.

[0051] Specifically, the generated trust identifiers are converted into corresponding trust factors according to preset quantification rules. For example, high trust identifiers correspond to trust factors of 0.9-1.0, medium trust identifiers correspond to trust factors of 0.6-0.8, and low trust identifiers correspond to trust factors of 0.3-0.5, realizing the transformation of trust identifiers from qualitative descriptions to quantitative values. Then, using the trust factor as a weighting coefficient, the residual values ​​of each dimension in the generated residual spectrum feature are weighted and calculated. That is, the value of each residual dimension is multiplied by the trust factor, thereby strengthening the reliability weight of the residual spectrum feature under high trust identifiers and weakening the influence of residual information that may have errors under low trust identifiers. Finally, all the weighted residual dimension values ​​are integrated in their original order to form a weighted residual vector that can reflect the difference between the original monitoring sequence and the standard comparison fingerprint and incorporate trust information.

[0052] The constructed weighted residual vector is used as input data and fed into a pre-defined neural network fault discriminator. This neural network has been trained with a large number of historical fault samples, covering weighted residual vector data of typical faults such as fuse blowout, poor contact, and insulation breakdown, as well as normal operating conditions. The network structure includes an input layer, hidden layers, and an output layer. After receiving the data of each dimension of the weighted residual vector, the input layer performs nonlinear transformation and fusion of features through the activation function of the hidden layer, such as the ReLU function, to gradually filter out key features strongly correlated with the fault. The output layer outputs multiple sets of quantitative results based on the feature mapping relationship formed during training, including the confidence level of the fault occurrence (a value in the range of 0-1, with the closer to 1 indicating a higher probability of the fault), the probability distribution of various typical faults (e.g., 82% probability of fuse blowout, 15% probability of poor contact, and 3% probability of no fault), and the quantitative value of the fault severity (1-5 levels corresponding to the residual deviation, with higher levels indicating a more significant fault impact). These output results are integrated in a fixed dimensional order of fault confidence level - fault type probability - severity to finally establish a fault discrimination vector that can comprehensively characterize the fault state of the target fuse.

[0053] First, the real-time current and voltage data of neighboring fuses within the distribution topology of the target fuse are obtained through the real-time power grid monitoring network. For example, the real-time current and voltage data of adjacent level 3-5 fuses on the same branch line are obtained. The current difference between the neighboring fuses and the target fuse at the same time node is calculated, that is, the absolute difference between the current value of the neighboring fuse and the current value of the target fuse, and the voltage difference, that is, the absolute difference between the voltage value of the neighboring fuse and the voltage value of the target fuse, are formed to form current and voltage difference information that includes the differences in electrical quantities between each neighboring device and the target device. Subsequently, using the parameters of each dimension of the fault discrimination vector, such as fault confidence, probability of various types of faults, and severity, as the matrix row index, and the device numbers of the target fuse and each neighboring fuse as the matrix column index, the corresponding parameter values ​​of the fault discrimination vector are filled into the target fuse column, and the current and voltage difference information is filled into the corresponding columns according to the neighboring devices. A two-dimensional spatial fault matrix containing the fault characteristics of the target device and the electrical relationship with the neighboring devices is constructed. Each element in the matrix reflects both the fault attributes of the target device and the electrical quantity differences between the neighboring devices and the target device.

[0054] Based on the power grid topology, the physical connection relationships between each neighboring fuse and the target fuse in the spatial fault matrix are determined, and electrical quantity difference thresholds and fault feature propagation coefficients are set. Subsequently, a spatial consistency check is performed on the spatial fault matrix: by comparing the current and voltage difference information between each neighboring fuse and the target fuse in the matrix, it is determined whether the electrical quantity differences conform to the spatial attenuation characteristics of fault propagation; simultaneously, it is checked whether the spatial distribution of fault features in the fault discrimination vector is consistent with the device association logic in the power grid topology. Abnormal data that does not conform to consistency is marked as isolated interference and removed; for features that conform to consistency, the location of the device with the most significant fault features and the most consistent propagation logic is located by combining the fault association strength of each device in the matrix, ultimately completing the spatial fault location and determining the specific fuse number and line section where the fault occurred.

[0055] In one possible implementation, step S400 further includes:

[0056] Step S450: Map the spatial consistency verification results to the power grid GIS topology to generate a visual fault location and fault type distribution map, wherein the fault type distribution map includes fault type, confidence level, maintenance priority and fault propagation hazard indicator.

[0057] Step S460: Perform fault early warning and management on the visualized fault location and fault type distribution map.

[0058] Specifically, the power grid GIS topology data is retrieved first. This data includes spatial information such as the route of distribution lines, the installation coordinates of fuses, equipment model labels, and line connection relationships. Then, the obtained spatial consistency verification results, including the specific fuse number, the line section where the fault is located, and the fault characteristic parameters, are matched with the GIS topology data. The fault location is then accurately mapped to the actual geographic coordinates on the GIS map using a spatial mapping algorithm, generating a visual fault location with obvious visual markers, such as a red flashing icon. Simultaneously, based on the core information in the fault discrimination vector, a fault type distribution map is generated by overlaying it next to the visualized fault location. The map clearly marks the fault type, such as specific fault forms like blown fuse, poor contact, and insulation breakdown. The fault confidence level is displayed as a percentage, reflecting the reliability of the location results. Based on the severity of the fault, such as whether it affects the power supply of the main line and the importance of the equipment, and whether it is a critical node fuse, the maintenance priority is determined and marked, divided into four levels: emergency, high, medium, and low. The map also uses dynamic diffusion graphics of different colors to mark the fault propagation hazard indicators, such as red for high-risk propagation areas and yellow for medium-risk propagation areas. Finally, a visualized chart that combines spatial location and fault attributes is formed, intuitively presenting comprehensive information about the fault.

[0059] Based on the maintenance priorities in the fault type distribution map, the visualization results are graded and processed. For high-priority faults, such as severe fuse failures, an immediate early warning mechanism is triggered. Alarm information including the visualized fault location, fault type, and urgency level is pushed to the operation and maintenance terminal through the power grid monitoring system, and a maintenance dispatch order is automatically generated. For medium- and low-priority faults, fault reports are compiled according to a preset cycle and updated synchronously to the power grid operation and maintenance management platform. At the same time, the time of the early warning, the receiving terminal, and the processing status are recorded to establish a closed-loop management mechanism for early warning, ensuring that fault information can be responded to and handled in a timely manner, thereby improving fault handling efficiency.

[0060] In one possible implementation, step S400 further includes:

[0061] Step S470: Configure the verification window according to the fault level, perform additional data acquisition of the target fuse in the verification window, and establish an additional dataset.

[0062] Step S480: Use the additional dataset to verify the fault location results and establish verification feedback.

[0063] Step S490: Update the fault location result based on the verification feedback.

[0064] Specifically, the system first configures a verification window of corresponding duration based on the fault level determined in the fault discrimination vector, such as emergency, high, medium, and low. The verification window duration for emergency and high-level faults is set shorter, such as 1-3 minutes, to quickly acquire secondary verification data. The verification window duration for medium and low-level faults can be appropriately extended, such as 5-10 minutes, to balance resource consumption while ensuring verification accuracy. After the verification window is activated, additional real-time operating data is collected during this period through the multi-dimensional monitoring modules already deployed on the target fuse, including current and voltage acquisition units, temperature sensors, and partial discharge detection components. This data specifically covers information such as instantaneous peak current changes, voltage fluctuation cycles, temperature curves of the equipment surface and terminals, and partial discharge signal strength. These newly added multi-dimensional monitoring data are then sorted and integrated according to timestamps to establish a structurally standardized and information-complete supplementary dataset, providing supplementary data support for subsequent verification of fault location results.

[0065] First, the current time-series data, voltage sampling sequence, temperature change curve, and partial discharge signal data from the supplementary dataset are retrieved, while fault type feature parameters and anomaly judgment thresholds are extracted from the initial fault location results. Then, a dynamic time warping algorithm is used to calculate the similarity between the current curve in the supplementary data and the current anomaly curve corresponding to the initial fault location, obtaining the current feature matching value. The Pearson correlation coefficient algorithm is used to analyze the correlation between the voltage fluctuation trend of the supplementary data and the fault voltage feature trend, generating a voltage correlation degree. A sliding window mean algorithm is used to process the temperature data, comparing the temperature change rate with the typical temperature change rate of the corresponding fault type to obtain the temperature conformity degree. A fast Fourier transform algorithm is used to perform spectral analysis on the partial discharge signal, extracting characteristic frequency components and comparing them with a fault characteristic frequency database to obtain the discharge signal matching rate. The current feature matching value, voltage correlation degree, temperature conformity degree, and discharge signal matching rate are substituted into a weighted summation formula according to preset weights to calculate the comprehensive verification index. Finally, verification feedback is generated based on the comprehensive verification index: when the index is greater than or equal to 0.8, the fault location result is highly matched with the supplementary data, and the location credibility is high; when the index is between 0.5 and 0.8, the matching details of each dimension and slight deviation items are provided, indicating that the location result is basically reliable, but the deviation dimension needs to be paid attention to; when the index is less than 0.5, the feedback indicates that the fault location result and the supplementary data have a low degree of matching and there are obvious abnormal differences, and the matching values ​​of each dimension and the difference data fragments are attached.

[0066] First, analyze the comprehensive verification index, matching details of each dimension, and location credibility description in the verification feedback to determine the direction and extent of adjustment to the fault location results. If the verification feedback shows that the comprehensive verification index is greater than or equal to 0.8 and the location credibility is high, maintain the fault type and location information in the initial fault location results, while increasing the fault confidence value by a preset ratio and adding a verification pass label to the results; if the feedback indicates that the comprehensive verification index is between 0.5 and 0.8 and there are slight deviations, combine the dimension data corresponding to the deviation, such as slightly lower current feature matching values ​​and fluctuating temperature compliance, adjust the fault severity level, add the specific data range of the deviation period, and correct the description of the fault impact range; if the feedback shows that the comprehensive verification index is less than 0.5 and the matching degree is low, trigger the relocation process, integrate the abnormal data fragments in the supplementary dataset into the spatial fault matrix construction stage, recalculate the current and voltage difference information of the neighboring fuses, perform spatial consistency verification, update the fault type, fault location, and confidence based on the new verification results, generate the corrected fault location results, and record the basis and data source of this update to ensure the results are traceable.

[0067] In one possible implementation, step S100 further includes:

[0068] Step S110: Read the monitoring signal from the current and voltage sensors.

[0069] Step S120: Perform denoising processing on the monitoring signal and perform timing alignment.

[0070] Step S130: Perform local adaptive signal anomaly authentication on the time-aligned monitoring signal, and establish the original monitoring sequence based on the local adaptive signal anomaly authentication result.

[0071] Specifically, the current and voltage sensors installed on the target high-voltage drop-out fuse are connected via a preset data acquisition interface. Data is collected at a set sampling frequency, such as 500 to 1000 data points per second, and the monitoring signals output by the sensors are read in real time. The read signals include the instantaneous values ​​of the three-phase currents, the effective values ​​of the line voltages between each phase, the precise timestamp of the signal acquisition time, and the sensor's own operating status indicator. This data is aggregated to the data processing unit via wired or wireless transmission, forming an initial, unprocessed monitoring signal stream.

[0072] The acquired monitoring signals are denoised using a wavelet threshold denoising algorithm to filter out high-frequency interference and impulse noise while retaining valid signal components. Simultaneously, the denoised current and voltage signals are time-aligned based on timestamp information to ensure accurate matching of current and voltage data at the same time point and eliminate timing deviations caused by sensor sampling delays.

[0073] First, a fixed-length sliding window technique is used, with a window duration of 100ms and a step size of 50ms, to segment the time-aligned current and voltage monitoring signals into continuous local signal segments. This ensures that each segment contains sufficient data while capturing short-term signal changes. For each local segment, the mean and standard deviation of the signal within the segment are calculated, and an anomaly threshold is dynamically generated based on the 3σ principle: upper threshold = mean + 3 × standard deviation, lower threshold = mean - 3 × standard deviation. This allows the threshold to be adaptively adjusted according to the local signal characteristics. Then, the signal sampling points within each segment are traversed, and the sampled values ​​are compared with the corresponding upper and lower thresholds of the segment. If the sampled value exceeds the threshold range, it is marked as an anomaly and the deviation magnitude is recorded; otherwise, it is marked as a normal point. After anomaly authentication of all segments is completed, the signal sampling points of each segment, along with their normal / anomaly markings and deviation magnitudes, are integrated in chronological order. Simultaneously, the acquisition timestamp and signal type information for each data point are added to construct a complete original monitoring sequence containing anomaly identifiers.

[0074] Example 2, based on the same inventive concept as the fault monitoring and location method for high-voltage drop-out fuses in the foregoing examples, such as... Figure 2 As shown, this application provides a fault monitoring and location system for high-voltage drop-out fuses. The system and method embodiments in this application are based on the same inventive concept. The system includes:

[0075] The original monitoring sequence establishment module 10 is used to obtain the original monitoring sequence corresponding to the current and voltage sensors deployed at the upper and lower ports of the target fuse.

[0076] The global pattern label establishment module 20 is used to activate the working mode recognizer to perform working mode recognition of the target fuse, establish a global pattern label, perform pattern fingerprint database adaptation matching according to the device parameters of the target fuse and the global pattern label, and establish a standard comparison fingerprint with a trust identifier.

[0077] The residual spectrum feature establishment module 30 is used to perform dynamic analysis and comparison of the original monitoring sequence using the standard comparison fingerprint to establish residual spectrum features.

[0078] The fault location result generation module 40 is used to establish a fault discrimination vector based on the residual spectrum features and the trust identifier, perform spatial fault location using the fault discrimination vector, and generate fault location results based on power grid GIS topology data.

[0079] Furthermore, the system is also used to implement the following functions:

[0080] The system comprises the following components: a time-domain residual vector unit, used to calculate the voltage drop amplitude difference and current instantaneous slope difference between the original monitoring sequence and the standard comparison fingerprint within a preset sliding window, and establish a time-domain residual vector; a fast Fourier transform unit, used to perform a fast Fourier transform on the original monitoring sequence and the standard comparison fingerprint to extract the fundamental frequency and odd harmonic amplitudes; a frequency-domain residual vector establishment unit, used to calculate the amplitude difference between the original monitoring sequence and the standard comparison fingerprint at each frequency point, and establish a frequency-domain residual vector; a time-frequency residual vector generation unit, used to perform a short-time Fourier transform on the original monitoring sequence and the standard comparison fingerprint, calculate the energy distribution difference at each time window, and generate a time-frequency residual vector; and a residual spectrum feature establishment unit, used to perform vector concatenation of the time-domain residual vector, the frequency-domain residual vector, and the time-frequency residual vector to establish a residual spectrum feature.

[0081] Furthermore, the system is also used to implement the following functions:

[0082] The normalization processing subunit is used to normalize the time-domain residual vector, frequency-domain residual vector, and time-frequency residual vector; the standardization weighting subunit is used to calculate the standard deviation of each residual dimension based on historical monitoring data, perform standardization weighting based on the standard deviation, and generate residual spectrum features.

[0083] Furthermore, the system is also used to implement the following functions:

[0084] The weighted residual vector establishment unit is used to convert the trust identifier into a trust factor and then perform weighted fusion of the corresponding residual spectrum features to establish a weighted residual vector; the fault discrimination vector establishment unit is used to input the weighted residual vector into the fault discriminator to establish a fault discrimination vector; the fault matrix establishment unit is used to obtain the current and voltage differential information of the neighboring fuses and establish a spatial fault matrix based on the current and voltage differential information and the fault discrimination vector; the spatial fault location completion unit is used to perform spatial consistency verification using the spatial fault matrix to complete the spatial fault location.

[0085] Furthermore, the system is also used to implement the following functions:

[0086] The fault type distribution map generation unit is used to map the spatial consistency verification results to the power grid GIS topology and generate a visual fault location and fault type distribution map. The fault type distribution map includes fault type, confidence level, maintenance priority and fault propagation hazard indicator. The fault early warning unit is used to manage the fault early warning of the visual fault location and fault type distribution map.

[0087] Furthermore, the system is also used to implement the following functions:

[0088] The additional dataset establishment unit is used to configure a verification window according to the fault level, perform additional data acquisition of the target fuse in the verification window, and establish an additional dataset; the verification feedback establishment unit is used to use the additional dataset to verify the fault location result and establish verification feedback; the fault location result update unit is used to update the fault location result according to the verification feedback.

[0089] Furthermore, the system is also used to implement the following functions:

[0090] The monitoring signal reading unit is used to read the monitoring signals from the current and voltage sensors; the timing alignment unit is used to perform noise reduction processing on the monitoring signals and to perform timing alignment; the anomaly authentication unit is used to perform local adaptive signal anomaly authentication on the timing aligned monitoring signals and to establish the original monitoring sequence based on the local adaptive signal anomaly authentication results.

[0091] Example 3, Figure 3 This is a schematic diagram of the electronic device provided by the present invention for the fault monitoring and location method of high voltage drop-out fuses, showing an exemplary electronic device suitable for implementing the embodiments of the present invention. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality or scope of the embodiments of the present invention. Figure 3 As shown, the electronic device includes a processor 21, a memory 22, an input device 23, and an output device 24; the number of processors 21 in the electronic device can be one or more. Figure 3 Taking a processor 21 as an example, the processor 21, memory 22, input device 23, and output device 24 in an electronic device can be connected via a bus or other means. Figure 3 Taking the example of a connection between China and Israel via a bus.

[0092] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0093] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0094] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A fault monitoring and locating method for a high voltage drop-out fuse, characterized in that, The method includes: Obtain the original monitoring sequences corresponding to the current and voltage sensors deployed at the upper and lower ports of the target fuse; The working mode recognizer is activated to perform working mode recognition of the target fuse, establish a global mode label, perform mode fingerprint database adaptation matching based on the device parameters of the target fuse and the global mode label, and establish a standard comparison fingerprint with a trust identifier. The original monitoring sequence is dynamically analyzed and compared using the standard comparison fingerprint to establish residual spectrum characteristics; A fault discrimination vector is established based on the residual spectrum features and the trust identifier. Spatial fault location is performed using the fault discrimination vector, and fault location results are generated based on power grid GIS topology data. Using the aforementioned standard comparison fingerprint, dynamic analysis and comparison of the original monitoring sequence are performed to establish residual spectrum features, including: Within a preset sliding window, calculate the voltage drop amplitude difference and current instantaneous slope difference between the original monitoring sequence and the standard fingerprint, and establish a time-domain residual vector; Perform a Fast Fourier Transform on the original monitoring sequence and the standard comparison fingerprint to extract the fundamental frequency and odd harmonic amplitude; The amplitude difference between the original monitoring sequence and the standard comparison fingerprint is calculated at each frequency point to establish a frequency domain residual vector; A short-time Fourier transform is performed on the original monitoring sequence and the standard comparison fingerprint to calculate the energy distribution difference of each time window and generate a time-frequency residual vector; The time-domain residual vector, frequency-domain residual vector, and time-frequency residual vector are concatenated to establish residual spectrum features; A fault discrimination vector is established based on the residual spectrum features and the trust identifier. Spatial fault localization is performed using the fault discrimination vector, including: After converting the trust identifier into a trust factor, a weighted fusion of the corresponding residual spectrum features is performed to establish a weighted residual vector; The weighted residual vector is input into the fault discriminator to establish the fault discrimination vector; Obtain the current-voltage differential information of neighboring fuses, and establish a spatial fault matrix based on the current-voltage differential information and the fault discrimination vector; Spatial consistency verification is performed using the aforementioned spatial fault matrix to complete spatial fault location. Fault location results are generated based on power grid GIS topology data, including: Configure the verification window according to the fault level, perform additional data acquisition of the target fuse in the verification window, and establish an additional dataset. The additional dataset is used to verify the fault location results and establish verification feedback. The fault location result is updated based on the verification feedback.

2. The fault monitoring and location method for high-voltage drop-out fuses as described in claim 1, characterized in that, The time-domain residual vector, frequency-domain residual vector, and time-frequency residual vector are concatenated to establish residual spectral features, including: The time-domain residual vector, frequency-domain residual vector, and time-frequency residual vector are normalized. The standard deviation of each residual dimension is calculated based on historical monitoring data, and standardized weighting is performed based on the standard deviation to generate residual spectral features.

3. The fault monitoring and location method for high-voltage drop-out fuses as described in claim 1, characterized in that, Fault location results are generated based on power grid GIS topology data, including: The spatial consistency verification results are mapped to the power grid GIS topology to generate a visual fault location and fault type distribution map. The fault type distribution map includes fault type, confidence level, maintenance priority, and fault propagation hazard indicator. The visualized fault location and fault type distribution map is used for fault early warning and dispatch management.

4. The fault monitoring and location method for high-voltage drop-out fuses as described in claim 1, characterized in that, Obtain the original monitoring sequences corresponding to the current and voltage sensors deployed at the upper and lower ports of the target fuse, including: Read the monitoring signals from the current and voltage sensors; Perform denoising processing on the monitored signal and perform timing alignment; The time-aligned monitoring signals are subjected to local adaptive signal anomaly authentication, and the original monitoring sequence is established based on the results of the local adaptive signal anomaly authentication.

5. A fault monitoring and location system for high-voltage drop-out fuses, characterized in that, include: The original monitoring sequence establishment module is used to obtain the original monitoring sequences corresponding to the current and voltage sensors deployed at the upper and lower ports of the target fuse; The global mode tag establishment module is used to activate the working mode recognizer to perform working mode recognition of the target fuse, establish a global mode tag, perform mode fingerprint database adaptation matching according to the device parameters of the target fuse and the global mode tag, and establish a standard comparison fingerprint with a trust identifier. The residual spectrum feature establishment module is used to perform dynamic analysis and comparison of the original monitoring sequence using the standard comparison fingerprint to establish residual spectrum features; The fault location result generation module is used to establish a fault discrimination vector based on the residual spectrum features and the trust identifier, perform spatial fault location using the fault discrimination vector, and generate fault location results based on power grid GIS topology data. The residual spectrum feature establishment module also includes: The time-domain residual vector unit is used to calculate the voltage drop amplitude difference and current instantaneous slope difference between the original monitoring sequence and the standard comparison fingerprint within a preset sliding window, and to establish a time-domain residual vector. The Fast Fourier Transform (FFT) unit is used to perform FFT on the original monitoring sequence and the standard comparison fingerprint to extract the fundamental frequency and odd harmonic amplitude. The frequency domain residual vector establishment unit is used to calculate the amplitude difference between the original monitoring sequence and the standard comparison fingerprint at each frequency point and establish the frequency domain residual vector. The time-frequency residual vector generation unit is used to perform short-time Fourier transform on the original monitoring sequence and the standard comparison fingerprint, calculate the energy distribution difference of each time window, and generate a time-frequency residual vector; The residual spectrum feature establishment unit is used to perform vector concatenation of the time-domain residual vector, frequency-domain residual vector, and time-frequency residual vector to establish residual spectrum features; The fault location result generation module also includes: The weighted residual vector establishment unit is used to convert the trust identifier into a trust factor, and then perform weighted fusion of the corresponding residual spectrum features to establish a weighted residual vector. The fault discrimination vector establishment unit is used to input the weighted residual vector into the fault discriminator to establish the fault discrimination vector; The fault matrix establishment unit is used to obtain the current-voltage differential information of the neighboring fuses and establish a spatial fault matrix based on the current-voltage differential information and the fault discrimination vector. The spatial fault location completion unit is used to perform spatial consistency verification using the spatial fault matrix to complete spatial fault location. The fault location result generation module also includes: The additional dataset creation unit is used to configure the verification window according to the fault level, perform additional data acquisition of the target fuse in the verification window, and create an additional dataset. The verification feedback establishment unit is used to verify the fault location results using the additional dataset and establish verification feedback. The fault location result update unit is used to update the fault location result based on the verification feedback.

6. The fault monitoring and location system for high-voltage drop-out fuses as described in claim 5, characterized in that, The residual spectrum feature establishment unit also includes: The normalization processing subunit is used to normalize the time-domain residual vector, frequency-domain residual vector, and time-frequency residual vector. The standardized weighting subunit is used to calculate the standard deviation of each residual dimension based on historical monitoring data, and to perform standardized weighting based on the standard deviation to generate residual spectral features.

7. The fault monitoring and location system for high-voltage drop-out fuses as described in claim 5, characterized in that, The fault location result generation module also includes: The fault type distribution map generation unit is used to map the spatial consistency verification results to the power grid GIS topology and generate a visual fault location and fault type distribution map. The fault type distribution map includes fault type, confidence level, maintenance priority and fault propagation hazard indicator. The fault early warning unit is used to manage the early warning of the visualized fault location and fault type distribution map.

8. The fault monitoring and location system for high-voltage drop-out fuses as described in claim 5, characterized in that, The original monitoring sequence establishment module also includes: The monitoring signal reading unit is used to read the monitoring signals from the current and voltage sensors; A timing alignment unit is used to perform noise reduction processing on the monitoring signal and to perform timing alignment. The anomaly authentication unit is used to perform local adaptive signal anomaly authentication on the time-aligned monitoring signal and to establish the original monitoring sequence based on the local adaptive signal anomaly authentication result.

9. An electronic device, characterized in that, The electronic device includes: processor; Memory used to store the processor's executable instructions; The processor is used to execute the fault monitoring and location method for high-voltage drop-out fuses as described in any one of claims 1 to 4.

Images