Dynamic evolution monitoring system and method for electric leakage fault of overhead line

Abstract

The invention discloses a dynamic evolution monitoring system and method for an electric leakage fault of an overhead line, and the method comprises the steps: transmitting an instruction to a grounding wire controller through a wireless communication module according to the coordinate information of a potential electric leakage region, and obtaining a feedback signal of the on-off state of a grounding wire; for the abnormal mode set, grouping multi-point data by adopting a k-means clustering method, and determining a fault feature vector represented by a clustering center; if the matching degree of the fault feature vector and an electric leakage mode stored in a historical database is higher than a threshold value, judging that an electric leakage point is confirmed, and generating a positioning coordinate sequence; transmitting the positioning coordinate sequence to a central terminal through a wireless network to obtain an integrated line state map; and adjusting the scanning path of the unmanned aerial vehicle by adopting a real-time updating mechanism according to the line state map to obtain an optimized monitoring data cycle so as to continuously track fault evolution.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of power system maintenance, in particular to a dynamic evolution monitoring system and method for overhead line leakage fault. BACKGROUND

[0002] With the continuous growth of global energy demand and the continuous improvement of power grid infrastructure, the power network is increasingly large and complex, and the power transmission line is like the blood vessels of the city, spreading throughout the urban and rural areas, and delivering light and power to millions of households. However, with the expansion of the power grid and the increase of aging equipment, the leakage problem has gradually become a major challenge to the safe and stable operation of the power system. Leakage not only causes unnecessary energy loss, but also can cause fires, electric shocks and other serious safety accidents, posing a threat to personnel life and property safety.

[0003] In this context, the leakage fault positioning technology for 0.4kV transformer user side distribution line is particularly important. As a common voltage level in low-voltage distribution systems, 0.4kV is widely used in residential areas, commercial buildings, industrial plants and other places. Its line is complex and variable, and the operating environment is different. Once leakage occurs, quickly and accurately locating the fault point becomes the key to ensuring the safety of the power system, reducing power outage time and improving power supply reliability.

[0004] In leakage fault detection, the core technical difficulty lies in how to realize real-time monitoring and remote control of line current. The collection of current parameters of overhead lines needs to be carried out in high-altitude environment. The traditional method relies on fixed installation of monitoring equipment, which is difficult to adjust position flexibly, especially when temporary detection of specific line sections is needed. In addition, real-time collection and remote control of grounding current also face challenges. Since the on-off state of the grounding line needs to respond quickly to the field situation, the traditional manual operation method is not only inefficient, but also may cause the fault to expand due to operation delay.

[0005] Therefore, how to realize flexible deployment of high-altitude current monitoring in complex environment, while ensuring remote control of grounding line on-off and real-time and efficient transmission of multi-point data, becomes a key problem of power line leakage fault tracing. SUMMARY

[0006] The purpose of the present application is to solve the above problems and provide a dynamic evolution monitoring system and method for overhead line leakage fault.

[0007] The technical solution adopted by the present application to solve its technical problems is: Further, a dynamic evolution monitoring system for overhead line leakage fault includes a grounding line device and a current clamp. The grounding wire device comprises a shell, and the shell 1 is sequentially provided with a gear switch, a current-limiting capacitor, a relay and a current transformer from left to right, and two current clamps are connected to both ends of the shell to facilitate the construction of the circuit.

[0008] Further, the gear switch has four gears: Gear 0 represents disconnection, and no subsequent line is connected; Gear 1 is connected to the relay through the No. 1 connecting terminal of the switch, and no current-limiting capacitor is added to this line, and the current can reach more than 100 A; Gear 2 is connected to a 300 UF capacitor through the No. 2 connecting terminal of the switch, and then connected to the relay, and for the 220 V alternating current connected to the device, the current is limited to within 20 A; Gear 3 is connected to a 200 UF capacitor through the No. 3 connecting terminal of the switch, and then connected to the relay, and for the 220 V alternating current connected to the device, the current is limited to within 14 A.

[0009] Further, a dynamic evolution monitoring method for overhead line leakage fault, using the device, comprising the following steps: S101, detecting the overhead line by the unmanned aerial vehicle carrying the current clamp, obtaining the current parameter data stream at the high-altitude position, and determining the potential leakage area if the current value in the data stream exceeds the preset threshold; S102, according to the coordinate information of the potential leakage area, using a wireless communication module to send instructions to the grounding wire controller, and obtaining the feedback signal of the grounding wire on-off state; S103, after obtaining the feedback signal, classifying the current parameter data stream and the grounding wire state by using the support vector machine algorithm, and obtaining the classified abnormal mode set; S104, for the abnormal mode set, grouping the multi-point data by using the k-means clustering method, and determining the fault feature vector represented by the clustering center; S105, if the fault feature vector matches the leakage mode stored in the historical database with a matching degree higher than a threshold, it is determined that the leakage point is confirmed, and a positioning coordinate sequence is generated; S106, transmitting the positioning coordinate sequence to the central terminal through the wireless network, and obtaining the integrated line state atlas; S107, according to the line state atlas, using a real-time updating mechanism to adjust the scanning path of the unmanned aerial vehicle, and obtaining the optimized monitoring data cycle, so as to continuously track the fault evolution.

[0010] Further, step S101 comprises: flying along the overhead line by the unmanned aerial vehicle carrying the current clamp, obtaining the current data stream at the high-altitude position, and obtaining the continuous current data sequence; According to the obtained current data sequence, the data stream is processed by using the sliding window method, the average value of the current in each window is calculated, and the smoothed current feature sequence is obtained; If the average value of any window in the smoothed current feature sequence exceeds the preset threshold, the potential leakage point is determined by the anomaly detection algorithm, and a preliminary leakage position set is obtained.

[0011] Further, step S102 comprises: The wireless communication module is used to establish a connection with the grounding line controller, send a command containing coordinate information, and obtain a feedback signal. The on-off state of the grounding line is parsed from the feedback signal to obtain state data. If the state data indicates that the grounding line is disconnected, the potential leakage area is located by the coordinate information, and the leakage area coordinates are determined. According to the leakage area coordinates, a region positioning algorithm is used to generate a region boundary, and boundary data is obtained. The boundary data is optimized by signal processing technology to obtain optimized boundary information. If the optimized boundary information exceeds the preset threshold, an adjustment instruction is sent to the grounding line controller using the communication protocol to obtain an adjusted on-off state.

[0012] Further, step S103 comprises: The feedback signal is obtained by collecting a current parameter data stream using a sensor to obtain an original data set. The original data set is denoised and standardized using a preprocessing technique to obtain a processed data set. The processed data set and the grounding line state are classified and trained using a support vector machine algorithm to obtain a classification model. If the accuracy of the classification model reaches a preset threshold, the newly collected current parameter data stream is predicted to obtain a classification result. According to the classification result, the correspondence between the abnormal mode and the grounding line state is analyzed to determine an abnormal mode set.

[0013] Further, step S104 comprises: The multi-point data in the abnormal mode set is obtained, and noise data is removed by preprocessing to obtain a cleaned data set. The cleaned data set is grouped using a K-means clustering method to determine the center points of each clustering group, and a clustering center set is obtained. The vector representation of each clustering center is extracted from the clustering center set to obtain an initial fault feature vector set. If the vector in the initial fault feature vector set has a similarity higher than a threshold to a preset fault template library, it is determined as a feature vector of the corresponding fault type, and a matched fault feature vector is obtained.

[0014] Further, step S105 comprises: The principal component analysis algorithm is used to perform dimension reduction processing on the fault feature vectors, and a dimension-reduced feature vector is obtained. The electric leakage mode corresponding to the dimension-reduced feature vector is retrieved from the historical database, and the matching degree is calculated. If the matching degree is higher than a preset threshold, the electric leakage point is determined, and a preliminary determination result is generated. According to the preliminary determination result, the electric leakage point positioning coordinates are calculated in combination with the sensor position information. The clustering algorithm is used to optimize the positioning coordinates, and a final coordinate sequence is generated.

[0015] Further, step S106 includes: The positioning coordinate sequence is obtained through a wireless network, encoded by using a data transmission protocol, and a standardized coordinate data stream is obtained. If the received coordinate data stream meets the preset format requirement, the central terminal processing module is used for decoding to obtain the parsed positioning coordinate sequence. According to the parsed positioning coordinate sequence, a data synchronization mechanism is used for timestamp alignment to determine a synchronized coordinate data set. The synchronized coordinate data set is subjected to clustering analysis by using the terminal computing capacity, and the K-means algorithm is used to obtain the classification result of the line state. If the classification result of the line state meets the preset update condition, the integrated line state graph is generated by using the graph visualization module, and the visualized state output is obtained.

[0016] Further, step S107 includes: The line state graph is obtained, the initial fault distribution data is generated by preprocessing, and the fault area range is determined. According to the initial fault distribution data, the path optimization algorithm is used to generate the initial scanning path of the unmanned aerial vehicle, and the scanning coverage area is obtained. The unmanned aerial vehicle performs the scanning task, obtains the real-time environment data and fault evolution data, and judges the environment change trend. If the environment change trend exceeds the preset threshold, the real-time update mechanism is used to adjust the scanning path of the unmanned aerial vehicle, and the dynamically adjusted path is obtained.

[0017] The beneficial effects of the present application are: 1、The application obtains current parameter data stream by carrying current clamp on unmanned aerial vehicle to scan overhead line, automatically identifies potential leakage area when data stream current value exceeds preset threshold, and sends wireless instruction to grounding line controller according to coordinates to obtain on-off feedback signal, then the application uses support vector machine algorithm to classify current data stream and grounding line state to obtain abnormal mode set, adopts k-means clustering method to group multi-point data to extract fault feature vector, confirms leakage point and generates positioning coordinate sequence if the vector matches historical database leakage mode higher than threshold, transmits to central terminal through wireless network to form integrated line state atlas, the application optimizes unmanned aerial vehicle scanning path by real-time updating mechanism to realize monitoring data closed loop circulation, thereby solving the fusion service problem from potential identification to dynamic tracking of overhead line leakage fault, ensuring accurate positioning and continuous monitoring in whole process of fault evolution, and finally realizing real-time visualization and efficient maintenance of line safety state. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 It is a schematic diagram of the grounding line device of the application; Figure 2 It is a schematic diagram of the current clamp of the application; Figure 3 It is a flow chart of the application. DETAILED DESCRIPTION

[0019] As shown in Figure 1 and Figure 2 , a dynamic evolution monitoring system for overhead line leakage fault, comprising a grounding line device and a current clamp, compares grounding line current and overhead current; the grounding line device comprises a shell 1, the shell 1 is provided with a gear switch 2, a current-limiting capacitor 3, a relay 4 and a current transformer 5 from left to right inside, two current clamps are connected at both ends of the shell to facilitate the construction of the circuit; there are also a battery, a main control chip, a small relay, a power switch and the like which are not marked in the figure, the whole device is powered by a 12V lithium battery, 12V directly meets the power demand of the relay. 5V voltage is sent out by 12V to 5V voltage stabilizing chip to meet the power demand of the main control chip.

[0020] The current-limiting capacitor is not a kind of independent "capacitor type", but a kind of working mode which uses the capacitive reactance to naturally limit the current in the alternating current loop. Its essence is that under 50 Hz / 60 Hz mains frequency, the capacitor presents a fixed and calculable impedance Xc, as long as the load voltage is much lower than the power supply voltage, the loop current is "fixed" by this Xc, thereby playing a current-limiting role.

[0021] The relay is a kind of electric control switch: low voltage and small current are used to control the on-off of high voltage and large current, and electrical isolation is realized at the same time. It is essentially "electromagnet + contact", simple in structure but extremely common.

[0022] Current Transformer, CT for short, is a "measuring transformer" that converts large current into small current according to accurate proportion, and is connected in series in the main loop, and the secondary is connected to the instrument / protection / energy meter to realize isolation + sampling + metering.

[0023] Assuming that the leftmost side is the incoming line end, the current enters the switch from the incoming line end, the gear switch 2 has 4 gears: 0 gear indicates disconnection, and no subsequent line is connected; 1 gear is connected to the relay through the 1# connection terminal of the switch, and no current limiting capacitor is added to this line, and the current can reach more than 100A; 2 gears are connected to a 300UF capacitor through the 2# connection terminal of the switch, and then connected to the relay, and for the 220V AC power connected to this device, the current is limited to within 20A; 3 gears are connected to a 200UF capacitor through the 3# connection terminal of the switch, and then connected to the relay, and for the 220V AC power connected to this device, the current is limited to within 14A.

[0024] As shown in Figure 3 A dynamic evolution monitoring method for overhead line leakage fault, the device comprises the following steps: S101, detect the overhead line by a drone carrying a current clamp to obtain a current parameter data stream at a high-altitude position, and if the current value in the data stream exceeds a preset threshold, determine a potential leakage area; S102, according to the coordinate information of the potential leakage area, use a wireless communication module to send an instruction to a grounding line controller to obtain a feedback signal of the grounding line on-off state; S103, after obtaining the feedback signal, classify the current parameter data stream and the grounding line state by a support vector machine algorithm to obtain a classified abnormal pattern set; S104, for the abnormal pattern set, use a k-means clustering method to group multi-point data to determine a fault feature vector represented by a clustering center; S105, if the fault feature vector matches a leakage pattern stored in a historical database with a matching degree higher than a threshold, determine a confirmed leakage point and generate a positioning coordinate sequence; S106, transmit the positioning coordinate sequence to a central terminal through a wireless network to obtain an integrated line state atlas; S107, according to the line state atlas, use a real-time updating mechanism to adjust the scanning path of the drone to obtain optimized monitoring data cycles, thereby continuously tracking the fault evolution.

[0025] Step S101 includes: flying along the overhead line by the unmanned aerial vehicle carrying the current clamp, using a fixed path scanning method, obtaining the current data stream at high altitude, and obtaining a continuous current data sequence. According to the obtained current data sequence, the data stream is processed by using a sliding window method, the average value of the current in each window is calculated, and a smoothed current feature sequence is obtained. If the average value of any window in the smoothed current feature sequence exceeds a preset threshold, such as 60 amperes, it is judged as a potential leakage point by an anomaly detection algorithm, and a preliminary leakage position set is obtained. For the preliminary leakage position set, the GPS coordinates of the corresponding position of the unmanned aerial vehicle are obtained, the physical position on the overhead line is determined through coordinate mapping, and the geographical information of the leakage area is obtained.

[0026] The unmanned aerial vehicle collects data at a frequency of 10 times per second, and obtains a current data sequence, such as 1000 current values collected in 100 seconds, with units of amperes. The current clamp captures the line current change through the principle of electromagnetic induction, and generates sequence data containing time stamps. This method ensures high spatiotemporal resolution of data, which helps the accuracy of subsequent analysis.

[0027] When the sliding window method is used to process the current data stream, the window size is set to 10 seconds, and the step size is 1 second. The average value of the current in each window is calculated. The window is sequentially slid to obtain a smoothed current feature sequence. This method can effectively filter out transient noise, highlight the trend change of the current, and improve the reliability of the leakage detection.

[0028] Based on the statistical method, it is judged whether the data in the window deviates from the normal range or not, and is marked as a potential leakage point. Assuming that 5 window average values are detected to exceed the threshold in 1000 data points, a preliminary leakage position set is generated. The anomaly detection algorithm reduces the false positive rate by analyzing the data distribution characteristics, and improves the identification accuracy of the leakage point.

[0029] For the preliminary leakage position set, the physical position is mapped by using the GPS coordinates of the unmanned aerial vehicle. Assuming that the GPS coordinates of a leakage point are north latitude 30.1234 degrees and east longitude 114.5678 degrees, the specific tower position on the line is mapped through the line geographic information system, such as “near the 25th tower of line A”. This method converts abstract data into operable geographical information, which is convenient for on-site investigation.

[0030] Step S102 includes: establishing a connection with the grounding line controller using a wireless communication module, sending instructions containing coordinate information, obtaining feedback signals; parsing the grounding line on-off state from the feedback signals to obtain state data; if the state data indicates that the grounding line is disconnected, locate the potential electric leakage area through the coordinate information, determine the electric leakage area coordinates; according to the electric leakage area coordinates, generate the area boundary using the area positioning algorithm to obtain boundary data; optimize the boundary data through signal processing technology to obtain optimized boundary information; if the optimized boundary information exceeds the preset threshold, send adjustment instructions to the grounding line controller using the communication protocol to obtain the adjusted on-off state.

[0031] When establishing a connection with the grounding line controller using a wireless communication module, the LoRa-based low-power wide-area network technology can be used. The LoRa module is suitable for long-distance signal transmission in overhead line scenarios due to its low power consumption and high penetration characteristics. The LoRa module can establish a connection with the controller through a gateway and send instruction data packets containing GPS coordinates. The data packet format includes coordinates, timestamps, and instruction types.

[0032] When parsing the grounding line on-off state from the feedback signal, the state data can be extracted through a signal decoder. Assuming that the feedback signal is in JSON format and contains the fields "status:0" and "timestamp:2025-10-1422:00:00". After parsing, if the state is "0", it indicates that the grounding line is disconnected, and the system immediately calls the stored coordinate information, such as longitude and latitude (116.391, 39.904), to locate the potential electric leakage area. This method ensures the accuracy of the state data by real-time parsing and reduces false positives.

[0033] The area positioning algorithm generates the area boundary by inputting the coordinate point set and setting the radius parameter such as 50 meters to cluster the boundary point set based on the DBSCAN density clustering algorithm.

[0034] When optimizing the boundary data, signal processing techniques such as Gaussian filter can be used to smooth the boundary curve. This optimization can improve the reliability of the boundary data and facilitate the controller's accurate identification of the target area.

[0035] If the optimized boundary information exceeds the preset threshold, such as the boundary area exceeding 500 square meters, the system sends adjustment instructions to the controller through the MQTT communication protocol.

[0036] Step S103 comprises: acquiring a feedback signal, collecting a current parameter data stream through a sensor to obtain an original data set; adopting a preprocessing technique to denoise and standardize the original data set to obtain a processed data set; classifying and training the processed data set and the grounding wire state through a support vector machine algorithm to obtain a classification model; if the accuracy of the classification model reaches a preset threshold, predicting a newly collected current parameter data stream to obtain a classification result; and analyzing a corresponding relationship between an abnormal mode and the grounding wire state according to the classification result to determine an abnormal mode set. The denoising processing can adopt a sliding average method, and a window size of 5 is set to smooth data fluctuations. The standardization normalizes the current value to the range of 0-1, facilitating subsequent analysis. The processed data set is more suitable for machine learning algorithm application.

[0037] The support vector machine algorithm is used for classification training. The classification target is the grounding wire on-off state, which is divided into two categories: "on" and "off". The processed data set is input into the model, and the kernel function can be set as a radial basis function during training. The classification accuracy needs to reach more than 90%.

[0038] If 950 groups of data are correctly classified out of 1000 groups of data, the accuracy is 95%, meeting the threshold requirement. When predicting the newly collected data, the sensor detects that the current intensity abnormally decreases to 0.1 ampere, and the model predicts "off", indicating that there may be a risk of electric leakage.

[0039] Specifically, when analyzing the corresponding relationship between the abnormal mode and the grounding wire state, the abnormal frequency and duration can be counted to determine the abnormal mode set. Assuming that 10 current abnormalities are recorded in a substation within a week, each lasting for 2 seconds, and being classified as a "short intermittent" mode. The K-means algorithm is used for clustering analysis, and the abnormal modes are divided into three categories: short intermittent, long-term disconnection, and periodic fluctuation.

[0040] Step S104 comprises: acquiring multi-point data in the abnormal mode set, removing noise data through preprocessing to obtain a cleaned data set; grouping the cleaned data set using the K-means clustering method to determine the center points of each clustering group and obtain a clustering center set; extracting the vector representation of each clustering center from the clustering center set to obtain an initial fault feature vector set; if the similarity between the vector in the initial fault feature vector set and a preset fault template library is higher than a threshold, it is determined as a feature vector of the corresponding fault type to obtain a matched fault feature vector.

[0041] In the field of grounding wire state monitoring of power systems, when acquiring multi-point data in the abnormal mode set, current, voltage and other parameters can be collected in real time through multiple sensors to form a multi-dimensional data set. During preprocessing, median filtering can be used to remove peak noise and retain true signal characteristics. The cleaned data set contains smooth current waveforms, reflecting the changes in the grounding wire state.

[0042] When grouping the cleaned dataset using the K-means clustering method, the number of clusters can be set to 3, representing normal, slight abnormal, and severe abnormal states respectively. Taking 1000 groups of data collected at a certain time as an example, after clustering, 3 groups are obtained, and each group center point represents a typical feature, such as the center point of the normal state being a stable current value of 20A, the slight abnormal being a fluctuation value of 25A, and the severe abnormal being a mutation value of 40A. The cluster center set clearly reflects the fault feature distribution.

[0043] When extracting the vector representation from the cluster center set, the current and voltage values of the center point can be combined into a vector, such as [20A, 220V]. The initial fault feature vector set is compared with the fault template library, which contains known fault modes, such as [50A, 200V] corresponding to a short circuit. Through cosine similarity calculation, if the similarity is higher than 0.9, it is determined as a short circuit feature vector. This process ensures accurate matching of fault types.

[0044] Step S105 includes: using principal component analysis algorithm to reduce the dimension of the fault feature vector, and obtaining a reduced dimension feature vector, such as the original vector dimension being 10, and through the algorithm, two principal components with a contribution rate of 95% are retained, and after dimension reduction, the vector is simplified to [2.1, 0.9], which is convenient for subsequent efficient matching; retrieving the leakage mode corresponding to the reduced dimension feature vector from the historical database, calculating the matching degree of the two; if the matching degree is higher than a preset threshold, it is determined as a leakage point, and a preliminary determination result is generated; according to the preliminary determination result, combining the sensor position information, calculating the leakage point positioning coordinates; using a clustering algorithm to optimize the positioning coordinates, and generating a final coordinate sequence.

[0045] The database stores historical leakage modes such as [2.0, 1.0], and the cosine similarity of the current vector [2.1, 0.9] is 0.98, which is higher than the threshold 0.85, that is, it is determined as a leakage, and the preliminary result "line A section leakage" is output. According to the preliminary determination result, combining the sensor position information, calculating the leakage point positioning coordinates. The sensor position is coordinate (100m, 50m), and the signal propagation delay is 0.2ms, and the leakage point coordinate (120m, 55m) is calculated.

[0046] The clustering algorithm is used to optimize the positioning coordinates, and a final coordinate sequence is generated. Input multiple preliminary coordinates such as (118m, 54m), (122m, 56m), and after K-means clustering k=3, the center sequence is [(115m, 52m), (120m, 55m), (125m, 58m)], and the noise points are removed to improve the accuracy.

[0047] Step S106 includes: obtaining the positioning coordinate sequence through the wireless network, encoding with the data transmission protocol to obtain the standardized coordinate data stream; if the received coordinate data stream meets the preset format requirement, such as checksum matching, decoding through the central terminal processing module to parse the complete sequence, the data loss rate is lower than 0.1%, obtaining the parsed positioning coordinate sequence; according to the parsed positioning coordinate sequence, the parsed positioning coordinate sequence contains 5 groups of point positions, using the data synchronization mechanism to align the time stamp, aligning the collection time from 13:05:12 to the unified reference, the error is controlled within 50ms, determining the synchronized coordinate data set; through the terminal computing capacity, the synchronized coordinate data set is analyzed by clustering, using the K-means algorithm, obtaining the classification result of the line state; if the classification result of the line state meets the preset update condition, generating the integrated line state graph through the graph visualization module, obtaining the visualized state output. According to the visualized state output, using the state monitoring analysis module for real-time comparison, judging the change trend of the line state through the change trend, updating the line state database of the central terminal, and generating the dynamically updated line state graph.

[0048] The process of obtaining the positioning coordinate sequence through the wireless network can utilize the ZigBee protocol to transmit data streams between power line sensors. When encoding with the data transmission protocol, the JSON format is selected to standardize the coordinate data stream, and the original sequence is compressed to {“x”:10.5,“y”:20.3}, reducing the volume by 30% for easy network transmission.

[0049] After determining the synchronized coordinate data set, the data set is analyzed by clustering through the terminal computing capacity, using the K-means algorithm, setting K=3, and classifying into normal, warning, and abnormal clusters, with cluster centers respectively being (12.0, 18.0) and the like.

[0050] If the line state classification result meets the preset update condition, such as the abnormal cluster ratio exceeding 20%, the integrated line state graph is generated through the graph visualization module, using SVG to render the color node graph, and red dots mark the abnormal area.

[0051] The visualized state output displays 3 red dots in the 10km line, and the state monitoring analysis module is used for real-time comparison, and the difference calculation with the previous graph identifies the change trend, such as the number of abnormal points increasing from 2 to 3.

[0052] In one embodiment, the central terminal line state database is updated through the change trend, the new trend value such as the growth rate 15% is stored in the SQL table, and the dynamically updated line state graph is generated, and the graph is refreshed every 5 minutes. This process is closely related to the leakage point positioning coordinate sequence, and the matched coordinates in the historical database are used to ensure that the visualized graph directly reflects the confirmed leakage position, such as mapping the point position with a matching degree of 85% to the red dot in the graph.

[0053] Encoding and decoding ensure data integrity, supporting clustering accuracy of 95%; in the extension scheme, adding timestamp alignment can optionally optimize synchronization accuracy to 20ms, further improving the reliability of classification results.

[0054] With multi-side support, K-means clustering combined with coordinate sequence history patterns verifies the matching of abnormal clusters and leakage vectors, and the accuracy of trend judgment is improved by 25%, which is beneficial to real-time monitoring and reduces the response time of line fault to minutes.

[0055] The visualization module integrates database updates to form a closed loop, such as displaying coordinate sequence details when clicking the red dot on the atlas, which helps operation and maintenance personnel quickly locate the source of the leakage and avoid the spread of losses.

[0056] Step S107 includes: obtaining a line state atlas, generating initial fault distribution data through preprocessing, and determining the fault area range; according to the initial fault distribution data, generating an initial scanning path of the unmanned aerial vehicle using a path optimization algorithm to obtain a scanning coverage area; through the unmanned aerial vehicle executing the scanning task, obtaining real-time environmental data and fault evolution data, and judging the environmental change trend; if the environmental change trend exceeds the preset threshold, then using a real-time updating mechanism to adjust the unmanned aerial vehicle scanning path to obtain a dynamically adjusted path.

[0057] The initial fault distribution data is generated by preprocessing, which means that the threshold segmentation is performed on the atlas pixel points to identify abnormal areas, thereby determining the fault area range. For example, the atlas collected by the line inspection unmanned aerial vehicle shows that the hot spot area of a certain section of high-voltage line reaches 50 square meters, and after preprocessing, the initial fault distribution data marks 3 hot spot clusters, with the range limited to coordinates (100, 200) to (150, 250), which ensures that the subsequent analysis focuses on the key area and avoids irrelevant interference.

[0058] According to the initial fault distribution data, the process of analyzing the line state atlas to obtain the line state atlas starts from preprocessing to generate the initial fault distribution data, which helps to identify potential problem areas. The path optimization algorithm generates an initial scanning path of the unmanned aerial vehicle, which means that the A* algorithm is used to calculate the minimum energy path to obtain the scanning coverage area.

[0059] Through the unmanned aerial vehicle executing the scanning task, real-time environmental data and fault evolution data are obtained, which means that the infrared sensor is carried to collect temperature and humidity every 10 seconds, and the environmental change trend is judged. In the scanning, the temperature of cluster 2 rises from 65°C to 72°C, and the humidity drops to 40%, with a trend value of 7% increase, which reflects the intensification of fault heat accumulation, facilitating early warning.

[0060] If the environmental change trend exceeds the preset threshold of 5%, then a real-time updating mechanism is used to adjust the unmanned aerial vehicle scanning path to obtain a dynamically adjusted path.

Claims

1. A dynamic evolution monitoring system for leakage fault of overhead line, characterized in that, The ground wire device comprises a shell (1) in which a gear switch (2), a current-limiting capacitor (3), a relay (4) and a current transformer (5) are sequentially arranged from left to right, and two current clamps are connected to the two ends of the shell to facilitate the construction of the circuit. The gear switch (2) has four gears:

2. The dynamic evolution monitoring system for the leakage fault of overhead line according to claim 1, characterized in that, Gear 0 indicates disconnection, and no subsequent line is connected; Gear 1 is connected to the relay through the No. 1 connecting terminal of the switch, and no current-limiting capacitor is added to this line, and the current can reach more than 100A; Gear 2 is connected to a 300UF capacitor through the No. 2 connecting terminal of the switch, and then connected to the relay, and for the 220V AC power connected to this device, the current is limited to within 20A; Gear 3 is connected to a 200UF capacitor through the No. 3 connecting terminal of the switch, and then connected to the relay, and for the 220V AC power connected to this device, the current is limited to within 14A. The method comprises the following steps:

3. A method for dynamic evolution monitoring of an overhead line leakage fault, using the device of claims 1 to 2, characterized in that, S101, detecting an overhead line by a current clamp carried by a UAV to obtain a current parameter data stream at a high-altitude position, and determining a potential electric leakage area if a current value in the data stream exceeds a preset threshold value; S102, sending an instruction to a ground wire controller according to coordinate information of the potential electric leakage area by using a wireless communication module to obtain a feedback signal of a ground wire on-off state; S103, after obtaining the feedback signal, classifying and processing the current parameter data stream and the ground wire state by using a support vector machine algorithm to obtain an abnormal mode set after classification; S104, grouping multi-point data by using a k-means clustering method for the abnormal mode set to determine a fault feature vector represented by a clustering center; S105, if a fault feature vector matches an electric leakage mode stored in a historical database with a matching degree higher than a threshold value, judging that an electric leakage point is confirmed, and generating a positioning coordinate sequence; S106, transmitting the positioning coordinate sequence to a central terminal through a wireless network to obtain an integrated line state atlas; and S107, adjusting a UAV scanning path according to the line state atlas by using a real-time updating mechanism to obtain optimized monitoring data circulation, thereby continuously tracking fault evolution. Step S101 comprises:

4. The method for dynamic evolution monitoring of overhead line leakage fault according to claim 3, characterized in that, flying along an overhead line by a current clamp carried by a UAV to obtain a current data stream at a high-altitude position, and obtaining a continuous current data sequence; processing the data stream by using a sliding window method according to the obtained current data sequence, calculating an average current value in each window, and obtaining a smoothed current feature sequence; if the average value of any window in the smoothed current feature sequence exceeds a preset threshold value, judging that it is a potential electric leakage point by using an abnormality detection algorithm, and obtaining a preliminary electric leakage position set. Step S102 comprises:

5. The method for dynamic evolution monitoring of overhead line leakage fault according to claim 3, characterized in that, establishing a connection with a ground wire controller by using a wireless communication module, sending an instruction containing coordinate information, and obtaining a feedback signal; analyzing a ground wire on-off state from the feedback signal to obtain state data; if the state data indicates that the ground wire is disconnected, positioning a potential electric leakage area by using the coordinate information to determine an electric leakage area coordinate; generating a region boundary by using a region positioning algorithm according to the electric leakage area coordinate to obtain boundary data; optimizing the boundary data by using a signal processing technology to obtain optimized boundary information; ​ If the optimized boundary information exceeds the preset threshold, an adjustment instruction is sent to the grounding line controller through a communication protocol to obtain an adjusted on-off state.

6. A method for dynamic evolution monitoring of leakage fault of overhead line according to claim 3, characterized in that, Step S103 comprises: Obtain the feedback signal, collect the current parameter data stream through the sensor, and obtain the original data set; Use preprocessing technology to denoise and standardize the original data set to obtain the processed data set; Classify and train the processed data set and the grounding line state through the support vector machine algorithm to obtain a classification model; If the accuracy of the classification model reaches the preset threshold, predict the newly collected current parameter data stream to obtain a classification result; According to the classification result, analyze the correspondence between the abnormal mode and the grounding line state to determine the abnormal mode set.

7. The method for dynamic evolution monitoring of overhead line leakage fault according to claim 3, characterized in that, Step S104 comprises: Obtain the multi-point data in the abnormal mode set, remove the noise data through preprocessing to obtain the cleaned data set; Use the K-means clustering method to group the cleaned data set, determine the center point of each clustering group, and obtain a clustering center set; Extract the vector representation of each clustering center from the clustering center set to obtain an initial fault feature vector set; If the vector in the initial fault feature vector set has a similarity higher than a threshold with a preset fault template library, it is determined as a feature vector of the corresponding fault type to obtain a matched fault feature vector.

8. The method for dynamic evolution monitoring of overhead line leakage fault according to claim 3, characterized in that, Step S105 comprises: Use principal component analysis algorithm to reduce the dimension of the fault feature vector to obtain a reduced dimension feature vector; Retrieve the electric leakage mode corresponding to the reduced dimension feature vector from the historical database and calculate the matching degree thereof; If the matching degree is higher than a preset threshold, it is determined as an electric leakage point to generate a preliminary determination result; According to the preliminary determination result, combine the sensor position information to calculate the positioning coordinates of the electric leakage point; Use a clustering algorithm to optimize the positioning coordinates to generate a final coordinate sequence.

9. The method for dynamic evolution monitoring of overhead line leakage fault according to claim 3, characterized in that, Step S106 comprises: Obtain the positioning coordinate sequence through a wireless network, encode using a data transmission protocol to obtain a standardized coordinate data stream; If the received coordinate data stream meets the preset format requirement, decode through the central terminal processing module to obtain an analyzed positioning coordinate sequence; According to the analyzed positioning coordinate sequence, use a data synchronization mechanism to align the time stamp to determine a synchronized coordinate data set; Perform clustering analysis on the synchronized coordinate data set through the terminal computing power, and use the K-means algorithm to obtain a classification result of the line state; If the classification result of the line state meets the preset update condition, generate an integrated line state graph through the graph visualization module to obtain a visual state output.

10. The method for dynamic evolution monitoring of leakage fault of overhead line according to claim 3, characterized in that, Step S107 comprises: Obtain the line state graph, generate initial fault distribution data through preprocessing, and determine the fault area range; According to the initial fault distribution data, use a path optimization algorithm to generate an initial scanning path of the unmanned aerial vehicle to obtain a scanning coverage area; Through the unmanned aerial vehicle to perform the scanning task, obtain real-time environment data and fault evolution data, and judge the environment change trend; If the environment change trend exceeds the preset threshold, adjust the unmanned aerial vehicle scanning path using a real-time updating mechanism to obtain a dynamically adjusted path.

Images