A leakage detection method and leakage detection system for power system

Abstract

The present invention discloses a leakage detection method and a leakage detection system for an electric power system. The method extracts features from the collected leakage current data of a line to be detected to obtain leakage characteristic parameters. When it is determined according to the leakage characteristic parameters that the line to be detected has a leakage risk, line information of the line to be detected is obtained. The line information includes line parameters, line voltage levels, and line load currents. Based on a pre-trained fault type diagnosis model, the leakage fault type of the line to be detected is determined according to the leakage characteristic parameters and the line information, thereby realizing the analysis of the leakage cause and providing an effective maintenance direction for the maintenance of the line to be detected, which is beneficial to improving the maintenance efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of power grids, and in particular to a leakage detection method and a leakage detection system for a power system. Background Art

[0002] In a complex power system, the probability of leakage in power lines increases with the increase of service life, so it is necessary to perform leakage detection on power lines to detect leakage in time. In the prior art, leakage current signals are generally collected to determine whether leakage occurs, but the cause of leakage is not analyzed, and effective maintenance direction cannot be provided for the maintenance of power lines, resulting in low maintenance efficiency. Therefore, it is urgent to provide a leakage detection method that can accurately identify the type of leakage fault. Summary of the invention

[0003] Based on this, the present invention provides a leakage detection method and a leakage detection system for an electric power system to solve the current defect that the cause of leakage cannot be identified in time.

[0004] To achieve the above object, an embodiment of the present invention provides a leakage detection method for a power system, comprising:

[0005] Collect leakage current data of the circuit to be tested;

[0006] Extracting leakage characteristic parameters of the leakage current data;

[0007] When it is determined that the line to be detected has a leakage risk according to the leakage characteristic parameter, line information of the line to be detected is obtained; wherein the line information includes line parameters and state parameters, and the state parameters include line voltage level and line load current;

[0008] Based on a pre-trained fault type diagnosis model, the leakage fault type of the line to be detected is determined according to the leakage characteristic parameters and line information of the line to be detected; wherein the fault type diagnosis model is trained based on several historical leakage fault cases, and the historical leakage fault cases include historical leakage characteristics, historical line information and fault type labels.

[0009] To achieve the above object, an embodiment of the present invention further provides a leakage detection system, comprising:

[0010] A data acquisition module, used to collect leakage current data of the circuit to be detected;

[0011] A feature extraction module, used to extract leakage characteristic parameters of the leakage current data;

[0012] A line information acquisition module, used for acquiring line information of the line to be detected when it is determined that the line to be detected has a leakage risk according to the leakage characteristic parameter; wherein the line information includes line parameters and state parameters, and the state parameters include line voltage level and line load current;

[0013] A fault classification module is used to determine the leakage fault type of the line to be detected based on a pre-trained fault type diagnosis model, according to the leakage characteristic parameters and line information of the line to be detected; wherein the fault type diagnosis model is trained based on a number of historical leakage fault cases, and the historical leakage fault cases include historical leakage characteristics, historical line information and fault type labels.

[0014] Compared with the prior art, the leakage detection method and leakage detection system of the power system disclosed in the embodiment of the present invention first collects the leakage current data of the line to be detected; then, by extracting the leakage characteristic parameters of the leakage current data, it is determined whether the line to be detected has a leakage risk; then, when it is determined that the line to be detected has a leakage risk according to the leakage characteristic parameters, the line information of the line to be detected is obtained; wherein the line information includes line parameters and state parameters, and the state parameters include line voltage level and line load current; finally, based on the pre-trained fault type diagnosis model, the leakage fault type of the line to be detected is determined according to the leakage characteristic parameters and line information of the line to be detected; wherein the fault type diagnosis model is trained according to several historical leakage fault cases, and the historical leakage fault cases include historical leakage characteristics, historical line information and fault type labels. It can be seen that the embodiment of the present invention determines the leakage fault type by combining the line information of the line to be detected when determining that the line to be detected has a leakage risk according to the leakage current data of the line to be detected, thereby realizing the analysis of the leakage cause, providing an effective maintenance direction for the maintenance of the line to be detected, and being conducive to improving the maintenance efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solution of the present invention, the drawings required for use in the implementation mode will be briefly introduced below. Obviously, the drawings described below are only some implementation modes of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 It is a flow chart of a leakage detection method for a power system provided by an embodiment of the present invention;

[0017] Figure 2 It is a structural schematic diagram of a leakage detection system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] See also Figure 1 , is a flow chart of a method for detecting leakage of an electric power system provided by an embodiment of the present invention. Specifically, the method for detecting leakage of an electric power system comprises steps S1 to S4:

[0020] S1. Collect leakage current data of the circuit to be tested.

[0021] Specifically, a current sensor is configured for the line to be detected in the power system to collect leakage current data. A current sensor is a detection device that can sense the current information being measured. Current sensors are divided into shunts, electromagnetic current transformers, and electronic current transformers according to different measurement principles. The specific type of current sensor can be selected according to actual needs and is not limited here.

[0022] S2. Extract leakage characteristic parameters from the leakage current data.

[0023] Specifically, after obtaining the leakage current data, the leakage current data is analyzed to extract leakage characteristic parameters, such as leakage current amplitude, leakage current frequency, leakage duration, etc. The extracted leakage characteristic parameters are used to determine whether there is a leakage risk in the circuit to be detected. If there is a leakage risk, continue to perform steps S3 and S4 to determine the type of leakage fault; if there is no leakage risk, there is no need to analyze the cause of the leakage.

[0024] S3. When it is determined that there is a leakage risk in the line to be detected according to the leakage characteristic parameters, obtain line information of the line to be detected; wherein the line information includes line parameters and state parameters, and the state parameters include line voltage level and line load current.

[0025] S4. Based on a pre-trained fault type diagnosis model, determine the leakage fault type of the line to be detected according to the leakage characteristic parameters and line information of the line to be detected; wherein the fault type diagnosis model is trained based on several historical leakage fault cases, and the historical leakage fault cases include historical leakage characteristics, historical line information and fault type labels.

[0026] Specifically, when it is determined that there is a leakage risk in the line to be detected, it is necessary to analyze the cause of the leakage by determining the leakage fault type. The leakage fault type is determined by inputting the leakage characteristic parameters of the line to be detected and the line information of the type to be detected into a pre-trained fault type diagnosis model, wherein the line information includes line parameters and state parameters, and the line parameters of the line to be detected are a set of circuit equivalent parameters that characterize the electromagnetic characteristics of the line to be detected, including but not limited to resistance, reactance, conductance, and susceptance.

[0027] Compared with the prior art, the embodiment of the present invention determines the leakage fault type by combining the line information of the line to be detected when determining that there is a leakage risk in the line to be detected based on the leakage current data of the line to be detected, thereby realizing the analysis of the cause of leakage. Maintenance personnel can select corresponding maintenance measures according to the leakage fault type, which is conducive to improving maintenance efficiency.

[0028] In one implementation, based on steps S1 to S4, the fault type diagnosis model is pre-trained in the following manner:

[0029] Acquire a number of historical leakage fault cases; wherein the historical leakage fault cases include the historical leakage characteristics, the historical line information and the fault type label, and the fault type label is insulation aging, equipment damage or poor grounding;

[0030] According to the historical leakage fault cases, the support vector machine was trained using the LIBSVM tool, and the optimal model hyperparameters were selected through the cross-validation method to obtain the initial fault classification model;

[0031] Selecting a plurality of first leakage fault cases from a historical maintenance record database, wherein the first leakage fault cases are historical leakage fault cases that are most similar to line information of the line to be detected;

[0032] The initial fault classification model is optimized according to the first leakage fault case to obtain the fault type diagnosis model.

[0033] Specifically, a historical maintenance record database stores multiple historical leakage fault cases. The historical leakage fault samples include historical leakage features, historical line features and corresponding fault type labels. These cases are used as training samples, and the LIBSVM tool is used to train the support vector machine (SVM). The optimal model hyperparameters are selected through the cross-validation method to obtain the initial fault classification model, wherein LIBSVM is a simple, easy-to-use, fast and effective SVM pattern recognition and regression software package. After determining the initial fault classification model, in order to make the model more suitable for fault diagnosis of the line to be detected, the K-nearest neighbor algorithm (KNN) is used to extract multiple historical leakage fault cases that are most similar to the line information of the line to be detected from the historical maintenance record database. For example, the historical leakage fault case with a line information similarity of more than 80% (this value can be set according to actual conditions and is not limited here) extracted from the historical maintenance record database by the K-nearest neighbor algorithm is used as the first leakage fault case, as the supplementary training data of the model, to further improve the classification performance of the model and obtain the final fault type diagnosis model.

[0034] Furthermore, the method of determining the leakage fault type of the line to be detected based on the pre-trained fault type diagnosis model and the leakage characteristic parameters of the line to be detected and the line information includes:

[0035] Inputting the leakage characteristic parameters and line information of the line to be detected into the fault type diagnosis model to obtain the initial fault type of the line to be detected;

[0036] Extracting a first set number of second leakage fault cases from the historical maintenance record database; wherein the second leakage fault case is a historical leakage fault case that is most similar to the current leakage fault case, and the current leakage fault case includes leakage characteristic parameters, line information, and initial fault type of the line to be detected;

[0037] When the number of target leakage fault cases is greater than a second set number, the initial fault type of the line to be detected is used as the leakage fault type of the line to be detected, and the current leakage fault case is recorded in the historical maintenance record database; wherein the second set number is greater than zero and less than the first set number; the target leakage fault case is the second leakage fault case whose fault type label is the same as the initial fault type.

[0038] Specifically, after step S1, the leakage current data is preprocessed, including data cleaning operations such as removing abnormal values ​​and missing values, and then the leakage characteristic parameters and line information of the line to be detected are input into the trained fault type diagnosis model, and the initial fault type of the line to be detected is output, and the initial fault type is insulation aging, equipment damage or poor grounding, etc. Then, the first set number of historical leakage fault cases that are most similar to the current leakage fault case are extracted as the second leakage fault case, for example, the historical leakage fault cases with a similarity greater than 85% (this value can be set according to actual conditions) are extracted as the second leakage fault case, and the fault type labels of these cases are analyzed to verify the initial fault type. For example, when the number of the second leakage fault cases with the same fault type label as the initial fault type of the line to be detected is greater than the second set number, the initial fault type of the line to be detected is used as the leakage fault type of the line to be detected, and the current leakage fault case is recorded in the historical maintenance record database for optimizing the fault type diagnosis model. The first set number and the second set number can be set according to actual conditions, for example, the value of the second set number is one, two or three.

[0039] In one embodiment, on the basis of steps S1 to S4, the leakage detection method is applied to a leakage finder, which includes a power module, a current sensor and a communication module;

[0040] Periodically triggering the self-test program of the leakage finder;

[0041] After the self-test program is triggered, the operating parameters of the various components of the leakage finder are collected;

[0042] When the working parameter of the first component exceeds the set normal working parameter range, the self-test result of the first component is abnormal, and the first component is an abnormal component; wherein the first component is any one of all components of the leakage finder;

[0043] Controlling the standby component corresponding to the abnormal component to take over the work of the abnormal component;

[0044] Sending an abnormal alarm message carrying a unique identifier of the abnormal component to a management system;

[0045] Recording the working parameters and the self-test results in a historical self-test database; the historical self-test database is used to record the working parameters and self-test results of each self-test;

[0046] Inputting the working parameters collected in real time into a pre-trained abnormal probability prediction model, and outputting the abnormal probability of the component; wherein the abnormal probability prediction model is pre-trained using the historical self-test database;

[0047] When the probability of the component being abnormal is less than a first set probability threshold, setting the trigger period of the self-check program to a first trigger period;

[0048] When the probability of the component abnormality is greater than a second set probability threshold, setting the trigger period of the self-check program to a second trigger period; wherein the first set probability threshold is less than the second set probability threshold, and the first trigger period is greater than the second trigger period;

[0049] When the probability of component abnormality is greater than or equal to the first set probability threshold and less than or equal to the second set probability threshold, the trigger period of the self-test program is set to a third trigger period; wherein the third trigger period is less than the first trigger period and greater than the second trigger period.

[0050] Specifically, the leakage detection method of the power system is executed on the leakage finder, which is composed of multiple components, including a power module for providing power to the leakage finder, a sensor (including a current sensor, a temperature sensor, etc.) and a communication module. According to the preset time interval, the self-test program of the leakage finder is triggered regularly, and a comprehensive test is performed on each key component in the leakage finder. The specific detection method is as follows: 1. By collecting the working parameters of each component and comparing them with the preset normal working parameter range, it is judged whether the working state of each component is normal. 2. If it is detected that the working parameters of a component exceed the normal working parameter range, it is determined that the component is abnormal and the switching operation is triggered. 3. According to the pre-configured backup plan, the backup component corresponding to the abnormal component is selected, and the work of the abnormal component is switched to the backup component, so that the backup component takes over the work of the abnormal component to ensure that the function of the leakage finder is not affected. 4. After switching to the backup component, an abnormal alarm message carrying the unique identifier of the abnormal component is sent to the management system at the same time to prompt relevant personnel to perform inspection and maintenance. 5. The self-test program will record the result data of each self-test (including working parameters and self-test results) into the historical self-test database. 6. The SVM algorithm is used to analyze and train the data in the historical self-test database, and an SVM-based abnormal probability prediction model is established for each component. By inputting the working parameters of each component collected in real time into the abnormal probability prediction model, the probability of abnormality of the component in the future (i.e., the component abnormality probability) is predicted. 7. According to the component abnormality probability given by the abnormal probability prediction model, the self-test cycle and the configuration of spare parts are dynamically adjusted. When the probability of component abnormality is high, the trigger cycle of the self-test program is shortened and the self-test frequency is increased; at the same time, the corresponding spare parts are prepared in advance to shorten the switching time. When the probability of component abnormality is low, the self-test cycle is appropriately extended, the self-test frequency is reduced, and equipment resources are saved. By adaptively adjusting the self-test cycle and spare part configuration, combined with real-time self-test and abnormal prediction, the continuous and stable operation capability of the leakage finder is improved, the equipment downtime caused by component abnormality is minimized, and the reliability and safety of the equipment are guaranteed.

[0051] Exemplarily, the self-test program of the leakage finder can be set to automatically trigger once every 24 hours. During the self-test process, for example, the output voltage and current of the power module will be collected in real time and compared with the rated value. If the deviation exceeds 5%, the power module is judged to be abnormal. The data transmission rate and packet loss rate of the communication module will also be monitored. If the transmission rate is lower than 10Mbps or the packet loss rate is higher than 5%, the communication module is judged to be abnormal. Once an abnormality is found, the system will automatically switch to pre-configured backup components, such as backup power supplies, backup sensors, and backup communication links, so as to ensure the continuous operation of the leakage finder. At the same time, the system will send abnormal alarm information to the management system through the wireless network to notify the operation and maintenance personnel to carry out maintenance. In addition, the system will also use the SVM algorithm to perform machine learning training on the collected massive self-test data, establish an abnormal probability prediction model, and analyze the trend of changes in the working parameters of the components over time to warn of possible component abnormality risks in advance (such as 3 days or 7 days, etc.). The self-test cycle is dynamically adjusted according to the probability of component abnormality. For example, when the probability of component abnormality is high, the trigger cycle of the self-test program is shortened from 24 hours to 2 hours, and the corresponding spare parts are activated in advance to deal with possible component abnormalities, thereby maximizing the stable operation of the leakage finder.

[0052] It is worth noting that the first set probability threshold, the second set probability threshold, the first trigger period, the second trigger period and the third trigger period can be set according to actual conditions and are not limited here.

[0053] In one embodiment, based on steps S1 to S4, the method further includes:

[0054] Acquiring topological structure information of the line to be detected;

[0055] Building a line model according to the topological structure information and line information of the line to be detected;

[0056] In the line model, the fault point location is adjusted multiple times;

[0057] After adjusting the position of the fault point each time, calculating the estimated impedance information of the line to be detected in the line model until the estimated impedance information is the same as the actual impedance information of the line to be detected obtained by actual measurement;

[0058] Outputting a predicted fault position; wherein the predicted fault position is a fault point position where the predicted impedance information is the same as the actual impedance information.

[0059] Specifically, by acquiring the topological structure information of the line to be detected and combining it with the line information of the line to be detected to perform line simulation, a line model is obtained. In the line model, the position of the fault point is continuously adjusted, and the estimated impedance position of the line to be detected corresponding to each fault point position is calculated, which is used for comparison with the actual impedance information of the line to be detected obtained by actual measurement. Finally, the position where the estimated impedance information is the same as the actual impedance information is found as the estimated fault position, thereby completing the fault location.

[0060] Preferably, after outputting the estimated fault location, the method further comprises:

[0061] Taking the estimated fault location as the positioning starting point, a handheld leakage locator is used to perform fault location search to find the actual fault location;

[0062] According to the deviation between the actual fault position and the predicted fault position, the line parameters in the line model are corrected until the deviation between the predicted fault position recalculated using the corrected line model and the actual fault position is less than a set deviation threshold.

[0063] Specifically, in order to make the fault location more accurate, after determining the estimated fault location, the estimated fault location is used as the starting point, and a handheld leakage locator is used to measure the change in the electromagnetic signal strength of the line to be detected, combined with parameters such as the cable burial depth, to calculate the precise location of the leakage fault (i.e., the actual fault location), and then calculate the deviation between the actual fault location and the estimated fault location, and correct the line parameters of the line model so that the line model can more accurately reflect the actual line conditions, thereby improving the accuracy of subsequent leakage fault location. After the line parameters are corrected, the estimated fault location is calculated again, and the actual fault location and the estimated fault location are compared again, and the line parameters are corrected, until the deviation between the recalculated estimated fault location and the actual fault location is less than the set deviation threshold. It is worth noting that the specific value of the deviation threshold can be set according to the actual situation and is not limited here.

[0064] Preferably, the method further comprises:

[0065] Obtaining a leakage time node at which the circuit to be detected has the leakage risk;

[0066] Generate a leakage fault diagnosis report according to the leakage fault type of the to-be-detected line, the estimated fault location and the leakage time node, and perform packet processing on the leakage fault diagnosis report and compress it using a Huffman coding algorithm;

[0067] The compressed leakage fault diagnosis report is uploaded to the cloud platform through the wireless communication module, so that the cloud platform performs the following steps:

[0068] Generate a fault work order based on the leakage fault diagnosis report and send the fault work order to the terminal of the operation and maintenance personnel; generate a fault isolation instruction based on the leakage fault diagnosis report and send the fault isolation instruction to the intelligent power distribution terminal, so that the intelligent power distribution terminal performs a trip isolation operation on the section where the fault estimated location is located according to the fault isolation instruction; receive the on-site maintenance situation and emergency repair measures record uploaded by the operation and maintenance personnel for the fault work order, and based on the correspondence between the preset maintenance status and the maintenance operation guidance, determine the next maintenance operation suggestion according to the on-site maintenance situation and send the next maintenance operation suggestion to the terminal of the operation and maintenance personnel; when receiving the fault repair information uploaded by the operation and maintenance personnel, change the status of the fault work order to completed, and count the completion time of the fault work order long; when the status of the fault work order is completed, sending a power supply restoration instruction to the intelligent distribution terminal to enable the intelligent distribution terminal to restore the power supply of the section where the fault estimated location is located; storing the leakage fault diagnosis report, the emergency repair measures record and the completion time of the fault work order in the emergency repair statistical database; using the data in the emergency repair statistical database to analyze, and optimize the correspondence between the maintenance status and the maintenance operation instructions; performing statistical analysis on all the leakage fault diagnosis reports in the emergency repair statistical database to obtain leakage time distribution information and leakage area distribution information, and presenting the leakage time distribution information and the leakage area distribution information in the form of visual charts to provide data support and decision-making reference for the distribution network operation and maintenance management of the power system.

[0069] Specifically, after determining the type of leakage fault and the estimated fault location, the relevant leakage information is packaged and sent to the cloud platform, and a fault work order is generated and dispatched on the cloud platform. The specific process is as follows: 1. Obtain the leakage time node where the line to be detected has a leakage risk, and generate a leakage fault diagnosis report based on the leakage fault type, estimated fault location and corresponding leakage time node of the line to be detected. In order to facilitate data upload, the leakage fault diagnosis report is subpackaged and compressed, and uploaded to the cloud platform with the communication module (such as wireless communication module). 2. After receiving the uploaded leakage fault diagnosis report, the cloud platform analyzes the report, enters the leakage fault type, estimated fault location and leakage time node of the line to be detected into the fault work order, and issues a fault isolation instruction to the intelligent distribution terminal at the same time. The intelligent distribution terminal performs a trip isolation operation on the section where the estimated fault location is located, cuts off the power supply to the section where the estimated fault location is located, and avoids the expansion of the leakage fault. 3. While performing the tripping isolation operation, the cloud platform sends the fault work order to the terminal of the operation and maintenance personnel, reminding the operation and maintenance personnel to go to the site for emergency repair as soon as possible. After receiving the fault work order, the operation and maintenance personnel navigate to the site according to the estimated fault location provided by the fault work order, and carry the corresponding emergency repair tools and spare parts according to the type of leakage fault, such as insulation resistance tester, infrared thermal imager, etc., to troubleshoot and repair the fault point. During the fault repair process, the operation and maintenance personnel can communicate with the cloud platform in real time through the handheld terminal, and upload the on-site maintenance situation and emergency repair measures to the cloud platform. 4. After receiving the maintenance feedback, the cloud platform analyzes the on-site maintenance situation based on the correspondence between the preset maintenance status and the maintenance operation guidance, gives the next maintenance operation suggestion, and pushes the suggestion information to the terminal of the operation and maintenance personnel to guide the operation and maintenance personnel to troubleshoot and repair the fault, thereby improving the efficiency of emergency repair. 5. After the fault is repaired, the operation and maintenance personnel upload the emergency repair results to the cloud platform through the handheld terminal, and the cloud platform updates the fault work order status to completed. At the same time, the cloud platform sends a power restoration instruction to the intelligent distribution terminal to restore the power supply to the section where the fault is estimated to be located. 6. The cloud platform stores the leakage fault diagnosis report, emergency repair process (such as emergency repair measures records and completion time of the fault work order) and other data of this leakage fault in the emergency repair statistics database, and uses the association rule mining algorithm Apriori to analyze the occurrence pattern of leakage faults, emergency repair efficiency, etc., to form optimization suggestions, which are used to improve the correspondence between maintenance status and maintenance operation instructions, and continuously improve the level of leakage fault handling. 7. The cloud platform regularly performs statistical analysis on the leakage fault diagnosis reports in the emergency repair statistics database, and intuitively displays the time distribution, regional distribution and other laws of leakage faults through visual charts such as pie charts and bar charts, providing data support and decision-making reference for the distribution network operation and maintenance management of the power system.

[0070] In one implementation, based on steps S1 to S4, collecting leakage current data of the circuit to be detected includes:

[0071] Obtaining a line voltage level and a line load current of a line to be detected, and determining an optimal sampling frequency and setting an amplitude threshold according to the line voltage level and the line load current of the line to be detected based on a preset mapping relationship table; wherein the mapping relationship table is pre-established based on expert experience;

[0072] Using a current sensor to collect line current data of the line to be detected in real time and store it in a designated data buffer area;

[0073] Obtain the rated voltage level and real-time load current data of the line, and determine the corresponding set amplitude threshold according to the preset mapping relationship table;

[0074] Adjusting the center frequency of the adaptive notch filter according to the optimal sampling frequency, and adjusting the bandwidth and stopband suppression of the adaptive notch filter according to the set amplitude threshold;

[0075] Using the adaptive notch filter to filter the line current data in the data buffer area to obtain current filtering data;

[0076] When the amplitude of the current filtering data is greater than the set amplitude threshold, it is determined that the current filtering data is leakage current data.

[0077] Specifically, the rated line voltage level and real-time line load current of the line to be detected are obtained, and the corresponding optimal sampling frequency and set amplitude threshold are determined according to the preset mapping relationship table. Among them, the mapping relationship table is pre-established based on a large amount of historical data and expert experience, and the optimal sampling frequency and set amplitude threshold under different line voltage levels and line load current conditions are recorded in the mapping relationship table. The line current data of the line to be detected is collected in real time by a current sensor, and the collected analog signal is converted into a digital signal through an analog-to-digital converter, and is transmitted to a designated data buffer area for storage through a data bus. According to the determined optimal sampling frequency and set amplitude threshold, the data in the data buffer area is adaptively filtered. Specifically, an adaptive notch algorithm is used to automatically adjust the center frequency of the adaptive notch filter according to the optimal sampling frequency, and the bandwidth and stopband suppression of the adaptive notch filter are automatically adjusted according to the set amplitude threshold, so as to effectively filter out the power frequency interference and harmonic interference components in the line current data. The current filtering data obtained by filtering is subjected to amplitude judgment. If the amplitude exceeds the set amplitude threshold, it is determined to be leakage current data. If the amplitude does not exceed the set amplitude threshold, it is not leakage current data.

[0078] In one embodiment, based on steps S1 to S4, after extracting the leakage characteristic parameters of the leakage current data, the method further includes:

[0079] When the leakage current amplitude of the circuit to be detected exceeds the set amplitude threshold and the leakage duration is greater than the set leakage time limit, it is determined that the circuit to be detected has a leakage risk; wherein the leakage characteristic parameters include the leakage current amplitude and the leakage duration;

[0080] or,

[0081] Converting the leakage characteristic parameters into leakage current characteristic vectors; wherein the leakage characteristic parameters include leakage current amplitude, leakage current frequency, and leakage duration of the leakage current amplitude exceeding the set amplitude threshold of the leakage current data, and the leakage current amplitude includes mean value, peak value, effective value and peak factor;

[0082] Inputting the leakage current feature vector into a pre-trained SVM classification model to obtain a classification result of the circuit to be detected; wherein the SVM classification model is pre-trained based on a number of historical leakage current feature vectors and corresponding historical classification results;

[0083] When the classification result of the circuit to be detected is dangerous leakage, it is determined that the circuit to be detected has leakage risk;

[0084] When the classification result of the circuit to be detected is general leakage, the data recording module is called to record the occurrence time of the general leakage.

[0085] Specifically, there are two ways to determine whether there is a leakage risk in the line to be detected: 1. Use a simple amplitude comparison method. Specifically, if the leakage current amplitude of the line to be detected exceeds the set amplitude threshold and the leakage duration is longer than the set leakage time limit (such as the leakage current amplitude exceeds 30mA and the duration exceeds 5 seconds), it is considered that the line to be detected has a leakage risk. 2. Use a model analysis method. Use the historical leakage feature vector and the corresponding historical classification results to train the model to obtain an SVM classification model, where the historical leakage features include the historical leakage current amplitude of the line to be detected, the historical leakage current frequency, and the leakage duration of the historical leakage current amplitude exceeding the set amplitude threshold. The leakage current amplitude includes the mean, peak value, effective value and peak factor. Use the trained SVM classification model to analyze the current leakage current feature vector to determine whether there is a dangerous leakage. If so, it is determined that there is a leakage risk in the line to be detected. If it is a general leakage, the data recording module is called to record the occurrence time of the general leakage.

[0086] Compared with the prior art, the leakage detection method of the power system disclosed in the embodiment of the present invention first collects the leakage current data of the line to be detected; then, by extracting the leakage characteristic parameters of the leakage current data, it is determined whether the line to be detected has a leakage risk; then, when it is determined that the line to be detected has a leakage risk according to the leakage characteristic parameters, the line information of the line to be detected is obtained; wherein the line information includes line parameters and state parameters, and the state parameters include line voltage level and line load current; finally, based on the pre-trained fault type diagnosis model, the leakage fault type of the line to be detected is determined according to the leakage characteristic parameters and line information of the line to be detected; wherein the fault type diagnosis model is trained according to several historical leakage fault cases, and the historical leakage fault cases include historical leakage characteristics, historical line information and fault type labels. It can be seen that the embodiment of the present invention determines the leakage fault type by combining the line information of the line to be detected when determining that the line to be detected has a leakage risk according to the leakage current data of the line to be detected, thereby realizing the analysis of the leakage cause, providing an effective maintenance direction for the maintenance of the line to be detected, and being conducive to improving the maintenance efficiency.

[0087] See also Figure 2 The embodiment of the present invention further provides a leakage detection system, comprising:

[0088] The data acquisition module 21 is used to collect leakage current data of the circuit to be detected;

[0089] A feature extraction module 22, used to extract leakage feature parameters of the leakage current data;

[0090] A line information acquisition module 23 is used to acquire line information of the line to be detected when it is determined that the line to be detected has a leakage risk according to the leakage characteristic parameter; wherein the line information includes line parameters and state parameters, and the state parameters include line voltage level and line load current;

[0091] The fault classification module 24 is used to determine the leakage fault type of the line to be detected based on a pre-trained fault type diagnosis model, according to the leakage characteristic parameters and line information of the line to be detected; wherein the fault type diagnosis model is trained based on a number of historical leakage fault cases, and the historical leakage fault cases include historical leakage characteristics, historical line information and fault type labels.

[0092] It is worth noting that the specific working process of the leakage detection system can refer to the working process of the leakage detection method of the power system described in the above embodiment, which will not be repeated here.

[0093] Compared with the prior art, the leakage detection system disclosed in the embodiment of the present invention determines the leakage fault type by combining the line information of the line to be detected when determining that there is a leakage risk in the line to be detected based on the leakage current data of the line to be detected, thereby realizing the analysis of the cause of the leakage, and can provide an effective maintenance direction for the maintenance of the line to be detected, which is conducive to improving maintenance efficiency.

[0094] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for detecting leakage in a power system, characterized in that: include: Collect leakage current data of the circuit to be tested; Extracting leakage characteristic parameters of the leakage current data; When it is determined that the line to be detected has a leakage risk according to the leakage characteristic parameters, line information of the line to be detected is obtained; wherein the line information includes line parameters and state parameters, and the state parameters include line voltage level and line load current; the line parameters of the line to be detected are a group of circuit equivalent parameters that characterize the electromagnetic characteristics of the line to be detected; Based on a pre-trained fault type diagnosis model, the leakage fault type of the line to be detected is determined according to the leakage characteristic parameters and line information of the line to be detected; wherein the fault type diagnosis model is trained based on a number of historical leakage fault cases, and the historical leakage fault cases include historical leakage characteristics, historical line information and fault type labels; The fault type diagnosis model is pre-trained in the following way: Acquire a number of historical leakage fault cases; wherein the historical leakage fault cases include the historical leakage characteristics, the historical line information and the fault type label, and the fault type label is insulation aging, equipment damage or poor grounding; According to the historical leakage fault cases, the support vector machine was trained using the LIBSVM tool, and the optimal model hyperparameters were selected through the cross-validation method to obtain the initial fault classification model; Selecting a plurality of first leakage fault cases from a historical maintenance record database, wherein the first leakage fault cases are historical leakage fault cases that are most similar to line information of the line to be detected; Optimizing the initial fault classification model according to the first leakage fault case to obtain the fault type diagnosis model; The method of determining the leakage fault type of the line to be detected based on the pre-trained fault type diagnosis model and according to the leakage characteristic parameters and line information of the line to be detected includes: Inputting the leakage characteristic parameters and line information of the line to be detected into the fault type diagnosis model to obtain the initial fault type of the line to be detected; Extracting a first set number of second leakage fault cases from the historical maintenance record database; wherein the second leakage fault case is a historical leakage fault case that is most similar to the current leakage fault case, and the current leakage fault case includes leakage characteristic parameters, line information, and initial fault type of the line to be detected; When the number of target leakage fault cases is greater than a second set number, the initial fault type of the line to be detected is used as the leakage fault type of the line to be detected, and the current leakage fault case is recorded in the historical maintenance record database; wherein the second set number is greater than zero and less than the first set number; the target leakage fault case is the second leakage fault case whose fault type label is the same as the initial fault type; The method further includes: acquiring topological structure information of the line to be detected; constructing a line model according to the topological structure information and line information of the line to be detected; adjusting the fault point position in the line model for multiple times; calculating the estimated impedance information of the line to be detected in the line model after each adjustment of the fault point position, until the estimated impedance information is the same as the actual impedance information of the line to be detected obtained by actual measurement; outputting the estimated fault position; wherein the estimated fault position is the fault point position where the estimated impedance information is the same as the actual impedance information; After outputting the predicted fault position, the method further includes: taking the predicted fault position as a positioning starting point, using a handheld leakage locator to perform a fault location search to find the actual fault position; and correcting the line parameters in the line model according to the deviation between the actual fault position and the predicted fault position, until the deviation between the predicted fault position recalculated using the corrected line model and the actual fault position is less than a set deviation threshold.

2. The method for detecting leakage in an electric power system according to claim 1, wherein: The leakage detection method is applied to a leakage finder, which includes a power module, a current sensor and a communication module; Periodically triggering the self-test program of the leakage finder; After the self-test program is triggered, the operating parameters of the various components of the leakage finder are collected; When the working parameter of the first component exceeds the set normal working parameter range, the self-test result of the first component is abnormal, and the first component is an abnormal component; wherein the first component is any one of all components of the leakage finder; Controlling the standby component corresponding to the abnormal component to take over the work of the abnormal component; Sending an abnormal alarm message carrying a unique identifier of the abnormal component to a management system; Recording the working parameters and the self-test results in a historical self-test database; the historical self-test database is used to record the working parameters and self-test results of each self-test; Inputting the working parameters collected in real time into a pre-trained abnormal probability prediction model, and outputting the abnormal probability of the component; wherein the abnormal probability prediction model is pre-trained using the historical self-test database; When the probability of the component being abnormal is less than a first set probability threshold, setting the trigger period of the self-checking program to a first trigger period; When the probability of the component abnormality is greater than a second set probability threshold, setting the trigger period of the self-check program to a second trigger period; wherein the first set probability threshold is less than the second set probability threshold, and the first trigger period is greater than the second trigger period; When the probability of component abnormality is greater than or equal to the first set probability threshold and less than or equal to the second set probability threshold, the trigger period of the self-test program is set to a third trigger period; wherein the third trigger period is less than the first trigger period and greater than the second trigger period.

3. The method for detecting leakage in an electric power system according to claim 1, wherein: Also includes: Obtaining a leakage time node at which the circuit to be detected has the leakage risk; Generate a leakage fault diagnosis report according to the leakage fault type of the to-be-detected line, the estimated fault location and the leakage time node, and perform packet processing on the leakage fault diagnosis report and compress it using a Huffman coding algorithm; The compressed leakage fault diagnosis report is uploaded to the cloud platform through the wireless communication module, so that the cloud platform performs the following steps: Generate a fault work order according to the leakage fault diagnosis report and send the fault work order to the terminal of the operation and maintenance personnel; generate a fault isolation instruction according to the leakage fault diagnosis report and send the fault isolation instruction to the intelligent power distribution terminal, so that the intelligent power distribution terminal performs a trip isolation operation on the section where the fault estimated position is located according to the fault isolation instruction; Receive the on-site maintenance situation and emergency repair measures record uploaded by the operation and maintenance personnel for the fault work order, and based on the correspondence between the preset maintenance status and the maintenance operation guidance, determine the next maintenance operation suggestion according to the on-site maintenance situation and send the next maintenance operation suggestion to the terminal of the operation and maintenance personnel; When receiving the fault repair information uploaded by the operation and maintenance personnel, the status of the fault work order is changed to completed, and the completion time of the fault work order is counted; When the status of the fault work order is completed, a power supply restoration instruction is sent to the intelligent power distribution terminal, so that the intelligent power distribution terminal restores the power supply of the section where the estimated fault location is located; the leakage fault diagnosis report, the emergency repair measure record and the completion time of the fault work order are stored in the emergency repair statistics database; The data in the emergency repair statistical database are used for analysis to optimize the correspondence between the maintenance status and the maintenance operation instructions; statistical analysis is performed on all the leakage fault diagnosis reports in the emergency repair statistical database to obtain leakage time distribution information and leakage area distribution information, and the leakage time distribution information and the leakage area distribution information are presented in the form of visual charts to provide data support and decision-making reference for distribution network operation and maintenance management.

4. The method for detecting leakage in an electric power system according to claim 1, wherein: The collecting of leakage current data of the circuit to be detected includes: Obtaining a line voltage level and a line load current of a line to be detected, and determining an optimal sampling frequency and setting an amplitude threshold according to the line voltage level and the line load current of the line to be detected based on a preset mapping relationship table; wherein the mapping relationship table is pre-established based on expert experience; Using a current sensor to collect line current data of the line to be detected in real time and store it in a designated data buffer area; Obtain the rated voltage level and real-time load current data of the line, and determine the corresponding set amplitude threshold according to the preset mapping relationship table; Adjusting the center frequency of the adaptive notch filter according to the optimal sampling frequency, and adjusting the bandwidth and stopband suppression of the adaptive notch filter according to the set amplitude threshold; Using the adaptive notch filter to filter the line current data in the data buffer area to obtain current filtering data; When the amplitude of the current filtering data is greater than the set amplitude threshold, it is determined that the current filtering data is leakage current data.

5. The method for detecting leakage in an electric power system according to claim 1, wherein: After extracting the leakage characteristic parameters of the leakage current data, the method further includes: When the leakage current amplitude of the circuit to be detected exceeds the set amplitude threshold and the leakage duration is greater than the set leakage time limit, it is determined that the circuit to be detected has a leakage risk; wherein the leakage characteristic parameters include the leakage current amplitude and the leakage duration; or, Converting the leakage characteristic parameters into leakage current characteristic vectors; wherein the leakage characteristic parameters include leakage current amplitude, leakage current frequency, and leakage duration of the leakage current amplitude exceeding the set amplitude threshold of the leakage current data, and the leakage current amplitude includes mean value, peak value, effective value and peak factor; Inputting the leakage current feature vector into a pre-trained SVM classification model to obtain a classification result of the circuit to be detected; wherein the SVM classification model is pre-trained based on a number of historical leakage current feature vectors and corresponding historical classification results; When the classification result of the circuit to be detected is dangerous leakage, it is determined that the circuit to be detected has leakage risk; When the classification result of the circuit to be detected is general leakage, the data recording module is called to record the occurrence time of the general leakage.

6. A leakage detection system, characterized in that: include: A data acquisition module, used to collect leakage current data of the circuit to be detected; A feature extraction module, used to extract leakage characteristic parameters of the leakage current data; A line information acquisition module, used for acquiring line information of the line to be detected when it is determined that the line to be detected has a leakage risk according to the leakage characteristic parameters; wherein the line information includes line parameters and state parameters, and the state parameters include line voltage level and line load current; the line parameters of the line to be detected are a group of circuit equivalent parameters characterizing the electromagnetic characteristics of the line to be detected; A fault classification module is used to determine the leakage fault type of the line to be detected based on a pre-trained fault type diagnosis model and according to the leakage characteristic parameters and line information of the line to be detected; wherein the fault type diagnosis model is trained based on a number of historical leakage fault cases, and the historical leakage fault cases include historical leakage characteristics, historical line information and fault type labels; The fault type diagnosis model is pre-trained in the following manner: obtaining a number of historical leakage fault cases; wherein the historical leakage fault cases include the historical leakage characteristics, the historical line information and the fault type label, and the fault type label is insulation aging, equipment damage or poor grounding; according to the historical leakage fault cases, the support vector machine is trained using the LIBSVM tool, and the optimal model hyperparameter is selected by the cross-validation method to obtain an initial fault classification model; a number of first leakage fault cases are selected from the historical maintenance record database, wherein the first leakage fault case is a historical leakage fault case that is most similar to the line information of the line to be detected; the initial fault classification model is optimized according to the first leakage fault case to obtain the fault type diagnosis model; The fault classification module is specifically used to: input the leakage characteristic parameters and line information of the line to be detected into the fault type diagnosis model to obtain the initial fault type of the line to be detected; extract a first set number of second leakage fault cases from the historical maintenance record database; wherein the second leakage fault case is a historical leakage fault case that is most similar to the current leakage fault case, and the current leakage fault case includes the leakage characteristic parameters, line information and initial fault type of the line to be detected; when the number of target leakage fault cases is greater than the second set number, the initial fault type of the line to be detected is used as the leakage fault type of the line to be detected, and the current leakage fault case is recorded in the historical maintenance record database; wherein the second set number is greater than zero and less than the first set number; the target leakage fault case is a second leakage fault case whose fault type label is the same as the initial fault type; The fault classification module is also used to: obtain the topological structure information of the line to be detected; construct a line model according to the topological structure information and line information of the line to be detected; in the line model, adjust the fault point position multiple times; after each adjustment of the fault point position, calculate the estimated impedance information of the line to be detected in the line model until the estimated impedance information is the same as the actual impedance information of the line to be detected obtained by actual measurement; output the fault estimated position; wherein the fault estimated position is the fault point position where the estimated impedance information is the same as the actual impedance information; use the fault estimated position as the positioning starting point, use a handheld leakage locator to perform fault positioning search, and find the real fault position; according to the deviation between the real fault position and the fault estimated position, correct the line parameters in the line model until the deviation between the fault estimated position recalculated using the corrected line model and the real fault position is less than a set deviation threshold.

Images