Distribution line single-phase earth fault diagnosis method, automatic early warning system and device

Abstract

The invention belongs to the field of electric power safety, and particularly relates to a distribution line single-phase earth fault diagnosis method, an automatic early warning system and an automatic early warning device. The method comprises the following steps: firstly, acquiring zero-sequence voltage and zero-sequence current of each measuring point, and then determining a single-phase grounding type corresponding to each risk node by adopting a mode identification method. When there are risk nodes of undetermined types still in the pattern recognition result, three-phase voltages and three-phase currents of all the detection points are obtained, two-stage detection is carried out through a single-phase earth fault detection model based on a deep neural network, the single-phase earth fault detection model is constructed based on Transform, and the single-phase earth fault detection model is constructed based on the deep neural network. And training by using field data and simulation data of the current line. The problems that an existing single-phase earth fault detection scheme is low in recognition efficiency and long in response time are solved, and active elimination and remote regulation and control of faults can be achieved by combining existing power grid equipment.

Description

Technical Field

[0001] The present invention belongs to the field of power safety, and particularly relates to a diagnosis method, an automatic warning system and a device for single-phase grounding faults of distribution lines. Background Art

[0002] Electric power provides convenience for modern human life, and it is crucial to ensure the safe and stable operation of the power system. With the rapid development of intelligent sensing equipment, analysis equipment and automatic control equipment, the traditional method of manually identifying power faults by sampling urgently needs to be replaced by a more intelligent method. Especially in remote areas, the trinity cooperation of intelligent sensing, discrimination and execution plays a crucial role in power operation safety and ensuring power consumption safety. Among them, intelligent sensing equipment can collect real-time signals of key information such as current, voltage and frequency in power operation at the millisecond level; intelligent discrimination devices can analyze the information collected by the sensing equipment in real time, transmit it to the control hall immediately when a fault is found, and send the fault information to each execution device; the execution device can solve the problem immediately to avoid greater harm and loss to power operation and passers-by around. Therefore, it is particularly important to achieve automatic diagnosis and disposal when a single-phase grounding fault occurs in the line.

[0003] Single-phase grounding fault is the most common fault in the 10kV (35kV) small-current grounding system of distribution, and mostly occurs in humid and rainy weather. It is caused by many factors such as tree obstacles, single-phase breakdown of insulators on distribution lines, single-phase disconnection and small animal hazards. Single-phase grounding not only affects the normal power supply of users, but also may generate overvoltage, burn out equipment, and even cause phase-to-phase short circuit and expand the accident. It is very important for duty and dispatching personnel to be familiar with the identification and treatment methods of grounding faults.

[0004] After a single-phase grounding fault occurs, the traditional solution is that the duty personnel should record the data and quickly report to the on-duty dispatcher and relevant responsible personnel, and then the duty personnel search for the grounding fault according to the order of the on-duty dispatcher. First, check whether there are obvious fault traces in the substation equipment. If not found, then conduct line grounding search. The line grounding search needs to be carried out step by step according to the bus and sub-lines to gradually reduce the fault range, finally locate the fault point, and arrange personnel to handle and restore the fault immediately. The traditional method relies on the personal experience of technical personnel, the workload of investigation is large, and the response time for finding and handling faults is long, which will pose unsafe factors to the equipment and residents running on the faulty line, and may seriously cause personal injury and property loss. Summary of the Invention

[0005] To solve the problems of low recognition efficiency and long response time existing in the manual detection method for single-phase grounding faults in existing distribution lines, the present invention provides a diagnosis method, an automatic warning system and a device for single-phase grounding faults in distribution lines.

[0006] The present invention is implemented by the following technical solutions:

[0007] A diagnosis method for single-phase grounding faults in distribution lines, which includes the following steps:

[0008] S1: Real-time obtain the values of zero-sequence voltage U0 and zero-sequence current I0 at the measuring points corresponding to each sectionalizing switch.

[0009] S2: Judge whether the zero-sequence current of any measuring point exceeds a preset threshold. If so, take it as a risk measuring point; otherwise, judge that no single-phase grounding fault has occurred in the current line.

[0010] S3: Calculate the phase difference between the zero-sequence current and the zero-sequence voltage of each risk node

[0011]

[0012] In the above formula, and respectively represent the phases of I0 and U0.

[0013] S4: Obtain the neutral point grounding type of the current distribution line, and make the following judgments according to the value of :

[0014] (1) If the current line is an ungrounded neutral system and satisfies then identify this measuring point as a fault node.

[0015] (2) If the current line is not an ungrounded neutral system and satisfies or then identify this measuring point as a fault node.

[0016] S5: After the above steps, if there are still risk nodes where faults have not been identified, obtain the monitoring data of all measuring points and input it into a single-phase grounding fault detection model, and the model outputs the single-phase grounding fault type of the risk nodes.

[0017] Among them, the single-phase grounding fault detection model is constructed based on Transformer and trained using the current line data; the input of the single-phase grounding fault detection model is the monitoring data of each measuring point in the current line, and the output is the classification result of the single-phase grounding fault of each measuring point.

[0018] As a further improvement of the present invention, in step S4, the monitoring data of each measuring point includes three-phase voltages U A 、U B 、U C and three-phase currents I A 、I B 、I C waveform data within a plurality of consecutive sampling periods.

[0019] As a further improvement of the present invention, in step S5, the monitoring data of each measuring point is the data within 4 cycles before and 8 cycles after the fault moment of the known fault node, and 64 points are collected in each cycle.

[0020] As a further improvement of the present invention, the single-phase grounding fault detection model constructed is composed of a linear embedding layer, a backbone network, and a prediction module. Among them, the linear embedding layer is used to encode the monitoring data of each input measuring point into corresponding feature vectors. The backbone network includes 6 feature encoding layers, and each feature encoding layer sequentially includes a multi-head attention module, a feed-forward network, an LN layer, and a Dropout layer; each feature encoding layer is used to extract multi-level information contained in the monitoring data of the measuring point. The prediction module is composed of two fully connected layers and is used to generate classification results of the fault types of each measuring point according to the extracted feature information.

[0021] As a further improvement of the present invention, the multi-head attention module in the feature encoding layer extracts the time features of a single measuring point and the relationship features between multiple measuring points through a hybrid attention mechanism.

[0022] As a further improvement of the present invention, the backbone network further includes an absolute position encoding module; the absolute position encoding module is used to extract position encodings representing the spatial relationships between measuring points according to the spatial distribution or topological structure relationship of the measuring points in the substation area line.

[0023] And / or

[0024] The feature encoding layer further includes a global normalization module; it is used to perform global normalization processing on the time features and relationship features extracted in the feature encoding layer and the spatial features extracted by the absolute position encoding module.

[0025] As a further improvement of the present invention, the single-phase grounding fault detection model further includes a preprocessing module; the preprocessing module is used to extract multi-scale time-frequency features in the original monitoring signals of each measuring point through wavelet transform, and input the multi-scale time-frequency features after being fused with the original detection signals into the linear embedding layer.

[0026] As a further improvement of the present invention, the sample data set required for the training stage is jointly composed of the real monitoring data collected from the real line and the simulated data in the virtual scenario simulated by the Pscad simulation model.

[0027] The single-phase grounding fault detection model uses the following Focal loss as the loss function during the training phase:

[0028] FL(p t ) = -(1 - p t ) γ log(p t ),

[0029] In the above formula, p t represents the degree of closeness between the classification result and the true label; γ is an adjustable factor, γ > 0.

[0030] The present invention also includes an automatic monitoring method for single-phase grounding faults, which includes the following steps:

[0031] 1. Adopt the diagnosis method of single-phase grounding faults in the distribution line as described above to monitor in real time whether a single-phase grounding fault occurs in the substation area line.

[0032] 2. Judge whether there is a single-phase grounding fault in the current period according to the monitoring result of the above step:

[0033] (1) If yes, send an alarm prompt to the command and control center and report relevant information;

[0034] (2) Otherwise, continuously detect the substation area line.

[0035] 3. The command and control center conducts a secondary judgment based on the reported alarm prompt and its relevant information:

[0036] (1) When the judgment result is true, select active intervention, and then start remote control of the substation area power grid according to the specific situation of the single-phase grounding fault measurement point, leaving enough time for personnel to go to the scene to eliminate the fault;

[0037] (2) When the judgment result is false, this alarm ends and continuous monitoring continues.

[0038] The present invention also includes an automatic early warning system for single-phase grounding faults, which is applied to a power substation area including DTU and FTU, and adopts the diagnosis method of single-phase grounding faults in the distribution line as described above to realize the identification and early warning of whether a single-phase grounding fault occurs at each measurement point according to the monitoring data on the substation area line. This automatic early warning system includes: a data acquisition module, a single-point analysis module, a multi-point analysis module, an early warning module, and a decision-making unit.

[0039] Among them, the data acquisition module is used to collect the phase voltage, phase current, zero-sequence voltage, and zero-sequence current at the measurement points corresponding to each sectionalizing switch on the line in real time through the DTU unit deployed in the substation area power grid.

[0040] The single - point analysis module is used to first obtain the values of zero - sequence voltage and zero - sequence current at each measuring point, determine whether the zero - sequence current at any measuring point exceeds a preset threshold. If so, it is regarded as a risk measuring point; otherwise, it is judged that no single - phase grounding fault has occurred on the current line. Then, it calculates the phase difference between the zero - sequence current and zero - sequence voltage at each risk node; finally, according to the preset judgment logic, it identifies the fault types of each measuring point based on the calculated phase difference and the neutral - point grounding type of the current distribution line.

[0041] The multi - point analysis module includes a single - phase grounding fault detection model trained based on a Transformer and using the current line data. The multi - point analysis module is used to, when the single - point analysis module fails to identify the fault types of all measuring points, input the waveform data of the phase - separated voltage and phase - separated current at each measuring point into the single - phase grounding fault detection model, and generate the fault types of the remaining nodes according to the output of the single - phase grounding fault detection model.

[0042] The early - warning module is used to generate corresponding early - warning information according to the identification results of single - phase grounding faults at each measuring point on the line output by the single - point analysis module and the multi - point analysis module, and send it to the command center in the background.

[0043] The decision - making unit is used to receive the control instructions issued by the command center according to the received early - warning information, and send the control instructions to the FTU units deployed in the substation power grid. The FTU units execute the control instructions and adjust the operating states of relevant power equipment.

[0044] The present invention also includes a diagnostic device for single - phase grounding faults in a distribution line, which includes a memory, a processor, and a computer program stored on the memory and running on the processor. When the processor executes the computer program, it implements the diagnostic method for single - phase grounding faults in a distribution line as described above, and further realizes automatically identifying whether a single - phase grounding fault occurs at each measuring point according to the monitoring information of each measuring point in the substation area.

[0045] The technical solution provided by the present invention has the following beneficial effects:

[0046] The present invention provides a two - stage fault discrimination method combining pattern recognition and artificial intelligence detection based on a deep neural network. This solution fully considers the abnormal characteristics of a single measuring point and the impact of line grounding faults on the entire line system. For the changes in the values of each observation point, it has more advantages in improving the diagnostic accuracy of single - phase grounding fault measuring points. The present invention combines the two recognition methods to avoid missed detection of single - phase grounding faults. And under the new two - step strategy, the real - time performance of the solution is significantly improved, and the recognition efficiency of faults is correspondingly increased. Therefore, it helps to realize real - time analysis and remote control of the power system, and is more conducive to timely discovery and handling of related faults.

[0047] The solution provided by the present invention can make full use of the data obtained by the signal acquisition devices in the current distribution network lines. There is no need to add new devices to the distribution lines, the deployment cost of the solution is lower, and the practical value is higher. After applying the present invention, the whole process of data acquisition, fault identification, and fault handling can be automated, greatly reducing the workload of on-site operation and maintenance personnel, reducing errors or mistakes introduced by humans, making signal processing more transparent, fault discrimination more controllable, and the whole system more intelligent. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a flowchart of the steps of the method for diagnosing single-phase grounding faults in a distribution line provided in Embodiment 1 of the present invention.

[0049] Figure 2 It is a distribution diagram of measurement points (marked by horizontal lines in the figure) in a typical power grid line topology diagram provided in Embodiment 1 of the present invention.

[0050] Figure 3 It is a waveform diagram of three-phase voltage and three-phase current input to the network model in Embodiment 1 of the present invention.

[0051] Figure 4 It is a model architecture diagram of the single-phase grounding fault detection model based on Ttansfomer constructed in Embodiment 1 of the present invention.

[0052] Figure 5 It is a flowchart of the steps of the automatic monitoring method for single-phase grounding faults provided in Embodiment 2 of the present invention.

[0053] Figure 6 It is a schematic diagram of the modules of the automatic early warning system for single-phase grounding faults provided in Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0055] Embodiment 1

[0056] This embodiment provides a method for diagnosing single-phase grounding faults in distribution lines. This method belongs to an online monitoring solution, which can combine the detection data obtained in real time from the substation lines to identify whether single-phase grounding faults occur at the measurement points corresponding to each sectionalizing switch, thereby significantly improving the real-time performance of fault handling and reducing the possible losses. In addition, the solution of this embodiment is combined with the actual use scenario, and the accuracy of single-phase grounding fault discrimination is ensured by adopting a combination of pattern discrimination and artificial intelligence discrimination. Among them, pattern discrimination realizes the diagnosis of single-phase grounding faults at individual measurement points in the line, and the artificial intelligence method utilizes all the measurement point data in the line to discriminate the measurement point faults from a more macroscopic perspective, obtaining a higher diagnostic accuracy rate.

[0057] Specifically, as Figure 1 shown, a method for diagnosing single-phase grounding faults in distribution lines provided in this embodiment includes the following steps:

[0058] S1: Obtain the values of zero-sequence voltage U0 and zero-sequence current I0 at the measurement points corresponding to each sectionalizing switch in real time.

[0059] In the power substation targeted by the solution of this embodiment, during the normal operation of the transmission line, the real-time signals such as current and voltage at each node fluctuate stably and orderly. However, when a single-phase grounding fault suddenly occurs at a certain point in the line, it will cause the ground connection of the grounded line, breaking the original stable and balanced state. It is manifested as a greater fluctuation of the zero-sequence voltage beyond the stable stage and disorderly. At the same time, the measurement points closer to the fault point show more obvious fluctuations, and other measurement points are less affected. Therefore, this embodiment first needs to monitor the changes of zero-sequence voltage U0 and zero-sequence current I0 at each node in real time. In the actual application process, this embodiment selects each sectionalizing switch in the power grid line topology diagram as shown in Figure 2 as the measurement point, and actively collects the zero-sequence voltage and zero-sequence current at each sectionalizing switch by using the DTU (data transmission unit) installed in the substation.

[0060] S2: Judge whether the zero-sequence current of any measurement point exceeds the preset threshold. If so, regard it as a risk measurement point; otherwise, judge that no single-phase grounding fault has occurred in the current line.

[0061] Zero-sequence current is the average value of the sum of three-phase currents in a power system, also known as zero-sequence signal. When a grounding fault occurs in the power system, the zero-sequence current will increase. In this embodiment, a safety threshold is set for the change of zero-sequence current. When the fluctuation of zero-sequence current at any node in the entire substation area line exceeds the threshold (such as the standard deviation is greater than 50), it is determined that there may be a fault at this measurement point. Therefore, in this embodiment, this node is determined as a risk measurement point. It should be noted that: a risk measurement point refers to a measurement point where a single-phase grounding fault may occur. As for whether this node actually belongs to a single-phase grounding fault, it still needs to be comprehensively judged in combination with subsequent criteria. On the contrary, when the zero-sequence currents of all nodes in the substation area line are below the threshold, it indicates that the substation area line operates smoothly and no single-phase grounding fault has occurred.

[0062] S3: For the measurement points where the zero-sequence current determined in the previous step exceeds the threshold, after identifying the risk, it is necessary to find out the transient and steady-state intervals. The transient state refers to a period of time after the fault occurs when the current fluctuates greatly and disorderly; the steady state refers to the interval that gradually stabilizes after being disorderly. Then, calculate the phase difference between the zero-sequence current and the zero-sequence voltage in the steady-state interval, and then identify the single-phase grounding fault in combination with the phase difference to determine the final identification result.

[0063] Specifically, the phase difference between the zero-sequence current and the zero-sequence voltage of each risk node The calculation formula is as follows:

[0064]

[0065] In the above formula, and respectively represent the phases of I0 and U0.

[0066] S4: When different neutral grounding methods are adopted in the power system, there are differences in the phase difference between the zero-sequence current and the zero-sequence voltage. Therefore, in this embodiment, it is necessary to first obtain the neutral grounding type of the current distribution line, and then discuss the situation of whether a single-phase grounding fault occurs at the measurement point according to the value. Specifically, the judgment strategy in this embodiment is as follows:

[0067] (1) If the current line is a non-grounded neutral system and satisfies then this measurement point is identified as a fault node.

[0068] (2) If the current line is not a non-grounded neutral system and satisfies or then this measurement point is identified as a fault node.

[0069] The above content is all about using the pattern recognition method to find single-phase grounding faults. In addition to this, if there are other situations where the zero-sequence of the measuring point exceeds the safety threshold, but the phase difference does not conform to the above two criteria, it is still uncertain whether the corresponding measuring point has a grounding fault. In the subsequent steps of this embodiment, the monitoring data of all measuring points will be combined to conduct a comprehensive evaluation on it.

[0070] S5: After the above steps are completed, if there are still risk nodes where the faults have not been identified, then obtain the monitoring data of all measuring points. Specifically, the monitoring data of each measuring point includes the three-phase voltages U A , U B , U C and the three-phase currents I A , I B , I C waveform data within multiple consecutive sampling periods. Next, this embodiment will input the monitoring data of each measuring point into a single-phase grounding fault detection model together, and the model will output the single-phase grounding fault type of the risk node.

[0071] When a fault occurs, the change direction of the current at the observation point closest to the fault will be opposite to that of other measuring points on the line. Based on such a phenomenon, this embodiment selects to use the data of similar faults occurring on the line to train a deep learning model, and then the trained network model constructs a model by integrating the current data collected by all measuring points to predict the fault measuring point and the normal measuring point, and the output is the normal or abnormal state of each measuring point.

[0072] Among them, the single-phase grounding fault detection model actually applied in this embodiment is constructed based on Transformer and trained using the current line data; the input of the single-phase grounding fault detection model is the monitoring data of each measuring point in the current line, and the output is the classification result of the single-phase grounding fault of each measuring point. As introduced above, as Figure 3 shown, the network model of this embodiment uses the waveform diagrams of the three-phase voltages and three-phase currents of each measuring point as input data. In a more optimized solution, the network model of this embodiment uses the 4 cycles before and 8 cycles after the fault moment of the known fault node at each measuring point as monitoring data, and 64 points are collected in each cycle. This special format of input data enables the network model to more accurately distinguish the changes in the monitoring data of the measuring points before and after the fault moment, and thus is more helpful to improve the recognition accuracy.

[0073] Specifically, as Figure 4As shown in the figure, the single-phase grounding fault detection model constructed in this embodiment consists of a linear embedding layer, a backbone network, and a prediction module. Among them, the linear embedding layer is used to encode the monitoring data of each measurement point into corresponding feature vectors. The single-phase grounding fault detection model constructed in this embodiment inputs the data of multiple measurement points into the network simultaneously, realizing the joint modeling of the time series current fluctuation characteristics of a single measurement point and the spatial difference characteristics between different measurement points, and adding new dimensional information.

[0074] In this embodiment, as Figure 4 shown, the backbone network includes 6 feature encoding layers, and each feature encoding layer sequentially includes a multi-head attention module, a feed-forward network, an LN layer, and a Dropout layer; each feature encoding layer is used to extract multi-level information contained in the monitoring data of the measurement points. In the feature encoding stage, the network model of this embodiment introduces a hybrid attention mechanism, which can effectively focus on the time characteristics of a single measurement point and the relationship characteristics between multiple measurement points, and is more flexible and comprehensive than traditional single-measurement point prediction methods. As Figure 4 shown, the prediction module consists of two fully connected layers and is used to generate the classification results of the fault types of each measurement point according to the extracted feature information. The classification result of the network in this embodiment reduces the dependence on a single measurement point signal through the joint analysis of multi-measurement point data, and improves the recognition accuracy and robustness of the model for single-phase grounding faults in complex scenarios.

[0075] In a more optimized solution of this embodiment, in addition to the traditional time series position encoding (processing the time series characteristics of 1024 time steps) in the Transformer model of the backbone network, an absolute position encoding module can also be additionally introduced; the absolute position encoding module is used to extract the position encoding representing the spatial relationship between measurement points according to the spatial distribution or topological structure relationship of the measurement points in the substation area line. The introduction of the absolute position encoding module enables the network model to simultaneously learn the spatio-temporal correlation of current fluctuations (such as the influence of the physical distance between adjacent measurement points on the current direction), improving the physical interpretability of fault location. Solve the problem that traditional methods ignore spatial topology information and are more in line with the actual scenario of the distribution network. On this basis, the feature encoding layer also includes a global normalization module; it is used to perform global normalization processing on the time features and relationship features extracted in the feature encoding layer and the spatial features extracted by the absolute position encoding module.

[0076] In a further optimized solution of this embodiment, for the transient characteristics of single-phase grounding fault signals, a preprocessing module is further included in the single-phase grounding fault detection model constructed in this embodiment; the preprocessing module is used to extract multi-scale time-frequency features (such as high-frequency transient components) in the original monitoring signals of each measurement point through wavelet transform, and fuse the multi-scale time-frequency features with the original detection signals and then input them into the linear embedding layer. This can enhance the sensitivity of the network model to fault features (such as high-frequency transient current) and reduce noise interference in the improvement. Thus, it can be seen that the traditional method relies on single-time domain or frequency domain analysis, while the improved method in this embodiment automatically fuses multi-scale features through deep learning to improve the detection ability of minor faults.

[0077] The single-phase grounding fault detection model constructed in this embodiment needs to be trained using real sample data in the local line to have the performance of accurately identifying single-phase grounding faults at each measurement point in the substation area. In practical applications, the probability of various single-phase grounding faults actually occurring in the substation area line is relatively low, and usually only limited to a small number of high-risk nodes each time a fault occurs. Therefore, the sample size of real data is limited and the sample types are single. Such sample data will affect the classification accuracy of the finally trained network model and reduce the generalization of the model.

[0078] To address the above problems, in the training stage of the single-phase grounding fault detection model in this embodiment, the real monitoring data collected from the real line and the simulated data in the virtual scenario simulated by the Pscad simulation model are jointly used to form the sample data set required for the training stage. The simulated data in the virtual scenario greatly increases the sample size of the original real data and can specifically simulate the sample data that is difficult to collect in the real scenario, thus greatly improving the quality of the sample data set.

[0079] In addition, the single-phase grounding fault detection model in this embodiment uses the following Focal loss as the loss function in the training stage:

[0080] FL(p t )=-(1-p t ) γ log(p t ),

[0081] In the above formula, p t represents the degree of proximity between the classification result and the true label; γ is an adjustable factor, γ>0.

[0082] Embodiment 2

[0083] As is well known, single-phase grounding faults, also known as small-current grounding faults, account for more than 80% of the distribution system faults and mostly occur in humid and rainy weather. They are caused by many factors such as tree obstacles, single-phase breakdown of insulators on distribution lines, single-phase disconnection, and damage caused by small animals. Such faults can lead to current passing through the ground, resulting in differences in relative ground potential, thus causing serious consequences such as current imbalance, voltage fluctuation, equipment damage, and even fires. Therefore, how to quickly identify the fault point, locate and dispose of the fault point is the key to ensuring the safe and stable operation of the power grid.

[0084] Based on the technology of the solution in Embodiment 1, this embodiment further provides an automatic monitoring method for single-phase grounding faults. The design idea of this monitoring method is to use the existing monitoring and control equipment in the substation area to collect the power data of each measurement point on the line, and then use the solution of Embodiment 1 combined with the collected data to monitor the single-phase grounding faults in the substation area in real time. Finally, the operation status of the power equipment in the substation area is actively regulated in combination with the monitoring results, in order to minimize the harm caused by the single-phase grounding faults occurring on the line.

[0085] Specifically, as Figure 5 shown, the automatic monitoring method for single-phase grounding faults provided in this embodiment includes the following steps:

[0086] First, use the diagnostic method for single-phase grounding faults of the distribution line as in Embodiment 1 to monitor in real time whether there is a single-phase grounding fault in the substation area line.

[0087] In the actual application process, the collection of measurement point information can be realized by using the existing data collection automation equipment on the line. In the existing circuit system, the real-time data necessary for power operation such as the current, voltage, and frequency of the bus and branch lines can be collected and transmitted in real time through DTU (Data Transmission Unit) and FTU (Feeder Terminal Unit). Among them, FTU (Feeder Terminal Unit) pays more attention to the monitoring and control of power equipment, and obtains the working state information of these equipment in real time, including key parameters such as current, voltage, and temperature. These information is of great significance for ensuring the normal operation of power equipment and preventing potential faults. In addition to data collection, FTU also has the ability to remotely control power equipment, such as remotely switching switches, adjusting current, etc. DTU, that is, the data transmission unit, is an important terminal device in the distribution network automation system, and its main responsibility is to establish a communication connection with the upper computer. Specifically, DTU can collect various data on site, such as current, voltage, temperature, etc., and then send these data to the upper computer through the communication network for processing and analysis.

[0088] Second, judge whether there is a single-phase grounding fault in the current period according to the monitoring results of the above step:

[0089] (1) If yes, send an alarm to the command and control center and report relevant information;

[0090] (2) Otherwise, the substation lines are continuously tested.

[0091] 3. The command and control center conducts secondary analysis based on the reported alarm prompts and related information:

[0092] (1) When the judgment result is true, active intervention is selected, and then remote control of the substation power grid is initiated according to the specific situation of the single-phase grounding fault measurement point, leaving enough time for personnel to arrive at the site to troubleshoot the fault;

[0093] (2) If the result of the analysis is false, the alarm ends and monitoring continues.

[0094] Based on the above, it can be seen that the automatic monitoring method for single-phase grounding faults provided in this embodiment can simultaneously realize data collection automation, fault monitoring and alarm automation, and decision-making execution mechanism automation, which greatly improves the efficiency of handling single-phase grounding faults in the substation area.

[0095] Among them, data collection automation is crucial for monitoring, fault warning, and operating status evaluation of power systems. In actual application, it is also necessary to align the timestamps of the current, voltage and other information collected on the line so that the clocks of each sampling point remain consistent to avoid the impact of inconsistent information on diagnostic results. Fault monitoring and alarm automation means that the solution can automatically realize the early warning and identification of single-phase grounding faults, determine the section where the fault occurs, and provide decision-making information for the executive agency to solve the problem. The automation of the decision-making executive agency is that after the fault alarm automation device feeds back the fault information to the command center, the system automatically completes the short-term troubleshooting work, buying time for the staff to arrive at the scene to solve the problem.

[0096] Example 3

[0097] On the basis of Examples 1 and 2, this embodiment further provides an automatic early warning system for single-phase grounding faults, which is applied to power substations including DTUs and FTUs, and adopts the diagnosis method for single-phase grounding faults of distribution lines as in Example 1, so as to identify and warn whether a single-phase grounding fault occurs at each measuring point based on the monitoring data on the substation line. Figure 6 As shown, the automatic early warning system includes: a data acquisition module, a single-point analysis module, a multi-point analysis module, an early warning module and a decision-making unit.

[0098] Among them, the data acquisition module is used to collect the phase voltage, phase current, zero-sequence voltage and zero-sequence current at the measuring points corresponding to each boundary switch on the line in real time through the DTU unit deployed in the substation power grid.

[0099] The single - point analysis module is used to first obtain the values of zero - sequence voltage and zero - sequence current at each measuring point, and determine whether the zero - sequence current at any measuring point exceeds a preset threshold. If so, it is regarded as a risk measuring point; otherwise, it is determined that no single - phase grounding fault has occurred on the current line. Then, it calculates the phase difference between the zero - sequence current and the zero - sequence voltage at each risk node; finally, according to the preset judgment logic, it identifies the fault types at each measuring point based on the calculated phase difference and the neutral - point grounding type of the current distribution line.

[0100] The multi - point analysis module includes a single - phase grounding fault detection model based on a Transformer and trained using the current line data. The multi - point analysis module is used to, when the single - point analysis module fails to identify the fault types at all measuring points, input the waveform data of the phase - split voltage and phase - split current at each measuring point into the single - phase grounding fault detection model, and generate the fault types of the remaining nodes according to the output of the single - phase grounding fault detection model.

[0101] The early - warning module is used to generate corresponding early - warning information according to the identification results of single - phase grounding faults at each measuring point on the line output by the single - point analysis module and the multi - point analysis module, and send it to the command center in the background.

[0102] The decision - making unit is used to receive the control instructions issued by the command center according to the received early - warning information, and send the control instructions to the FTU units deployed in the distribution network of the substation area. The FTU units execute the control instructions and adjust the operating states of relevant power equipment.

[0103] In the actual deployment and application process of the automatic early - warning system for single - phase grounding faults provided in this embodiment, the data acquisition module and the decision - making unit can be implemented based on existing DTU and FTU devices, while the single - point analysis module, the multi - point analysis module, and the early - warning module can be implemented through the background server deployed in the command control center. The background server obtains the real - time grid monitoring data uploaded by the DTU devices and FTU devices on the substation area line, and combines the collected data to identify the single - phase grounding faults in the substation area. After the background server identifies a fault, it immediately sends an early - warning message to the operation and maintenance personnel and issues an instruction to the FTU device to actively regulate the operating states of the power equipment on the grid according to the preset disposal strategy.

[0104] Embodiment 4

[0105] This embodiment provides a diagnostic device for single - phase grounding faults in a distribution line, which includes a memory, a processor, and a computer program stored on the memory and running on the processor. When the processor executes the computer program, it implements the diagnostic method for single - phase grounding faults in a distribution line as in Embodiment 1, and further realizes automatically identifying whether a single - phase grounding fault occurs at each measuring point according to the monitoring information of each measuring point in the substation area.

[0106] The diagnostic device for single-phase grounding faults in the distribution line provided in this embodiment is essentially a computer device. In actual application, this computer device can adopt an embedded device and be deployed in the control device on the power grid. It can also sample an independent computer device and be deployed in the control center of the power grid. The latter computer device can adopt medium and large-sized computer devices such as notebook computers, tablet computers, desktop computers, or rack-mounted servers, blade servers, tower servers, or cabinet servers (including independent servers or server clusters composed of multiple servers) that can execute computer programs.

[0107] The computer device in this embodiment at least includes but is not limited to: a memory and a processor that can communicate with each other through a system bus. In this embodiment, the memory (i.e., the readable storage medium) includes flash memory, hard disks, multimedia cards, card-type memories (such as SD or DX memories, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memories, magnetic disks, optical disks, etc. In some embodiments, the memory can be an internal storage unit of the computer device, such as the hard disk or memory of the computer device. In other embodiments, the memory can also be an external storage device of the computer device, such as the plug-in hard disk equipped on the computer device, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc. Of course, the memory can also include both the internal storage unit and the external storage device of the computer device. In this embodiment, the memory is usually used to store the operating system and various application software installed on the computer device. In addition, the memory can also be used to temporarily store various data that have been output or will be output.

[0108] In some embodiments, the processor can be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips. The processor is usually used to control the overall operation of the computer device.

[0109] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A diagnostic method for single-phase grounding faults in a distribution line, characterized in that, It includes the following steps: S1: Obtain the values of zero-sequence voltage U0 and zero-sequence current I0 at the measurement points corresponding to each sectionalizing switch in real time; S2: Determine whether the zero-sequence current at any measurement point exceeds a preset threshold. If so, take it as a risk measurement point; otherwise, determine that no single-phase grounding fault has occurred in the current line; S3: Calculate the phase difference between the zero-sequence current and the zero-sequence voltage of each risk node In the above formula, and respectively represent the phases of I0 and U0; S4: Obtain the neutral grounding type of the current distribution line, and make the following judgments based on the value: (1) If the current line is an ungrounded neutral system and meets then identify this measuring point as a fault node; (2) If the current line is not an isolated neutral system and meets or then identify this measuring point as a fault node; S5: After the above steps are completed, if there are still risk nodes for which the fault has not been identified, obtain the monitoring data of all measurement points and input it into a single-phase grounding fault detection model, and the model outputs the single-phase grounding fault type of the risk nodes; The single-phase grounding fault detection model is constructed based on Transformer and trained using the on-site data and simulation data of the current line; Its input is the monitoring data of each measurement point in the current line, and the output is the classification result of the single-phase grounding fault of each measurement point.

2. The diagnostic method for single-phase grounding faults of a distribution line according to claim 1, wherein: In step S4, the monitoring data of each measuring point includes the three-phase voltages U A , U B , U C and the waveform data of the three-phase currents I A , I B , I C .

3. The diagnostic method for single-phase grounding faults of a distribution line according to claim 2, wherein: In step S5, the 4 cycles before the fault moment of the known fault node and the 8 cycles after it for each measurement point are used as the monitoring data, and 64 points are collected in each cycle.

4. The diagnostic method for single-phase grounding fault of a distribution line according to claim 2, characterized in that: The single-phase grounding fault detection model consists of a linear embedding layer, a backbone network based on Transformer, and a prediction module; the linear embedding layer is used to encode the monitoring data of each input measurement point into corresponding feature vectors; the backbone network includes 6 feature encoding layers, and each feature encoding layer sequentially includes a multi-head attention module, a feed-forward network, an LN layer, and a Dropout layer; the feature encoding layer is used to extract multi-level information contained in the monitoring data of the measurement points; the prediction module consists of two fully connected layers and is used to generate the classification result of the fault type of each measurement point according to the extracted feature information.

5. The diagnostic method for single-phase grounding faults of a distribution line according to claim 4, characterized in that: The multi-head attention module in the feature encoding layer extracts the time features of a single measurement point and the relationship features between multiple measurement points through a hybrid attention mechanism; And / or The backbone network also includes an absolute position encoding module; the absolute position encoding module is used to extract the position encoding representing the spatial relationship between measurement points according to the spatial distribution or topological structure relationship of the measurement points in the substation area line; And / or The feature encoding layer also includes a global normalization module; it is used to perform global normalization processing on the time features and relationship features extracted in the feature encoding layer and the spatial features extracted by the absolute position encoding module.

6. The diagnostic method for single-phase grounding faults in distribution lines according to claim 5, characterized in that: The single-phase grounding fault detection model also includes a preprocessing module; the preprocessing module is used to extract the multi-scale time-frequency features in the original monitoring signals of each measurement point through wavelet transform, and fuse the multi-scale time-frequency features with the original detection signals and then input them into the linear embedding layer.

7. The diagnostic method for single-phase grounding faults of a distribution line according to claim 1, characterized in that: Use the real monitoring data collected from the real line and the simulation data in the virtual scenario simulated by the Pscad simulation model to jointly form the sample data set required for the training stage; The single-phase grounding fault detection model uses the following Focal loss as the loss function in the training stage: FL(p t ) = -(1 - p t ) γ log(p t ) In the above formula, p t represents the degree of proximity between the classification result and the true label; γ is an adjustable factor, where γ > 0.

8. An automatic monitoring method for single-phase grounding faults, characterized in that, It includes: One, use the diagnostic method for single-phase grounding faults in distribution lines as described in any one of claims 1-7 to monitor in real time whether a single-phase grounding fault has occurred in the substation area line; Two, judge whether there is a single-phase grounding fault in the current period according to the monitoring result of the above step: If so, send an alarm prompt to the command and control center and report relevant information; Otherwise, continuously detect the substation line; 3. The command and control center conducts secondary judgment based on the reported alarm prompt and its relevant information: When the judgment result is true, select active intervention, and then start remote control of the substation power grid according to the specific situation of the single-phase grounding fault measurement point, leaving enough time for personnel to eliminate the fault on site; When the judgment result is false, this alarm ends and continuous monitoring continues.

9. An automatic early warning system for single-phase grounding faults, characterized in that: It is applied to a power substation including DTU and FTU, and adopts the diagnostic method for single-phase grounding fault of a distribution line described in any one of claims 1-7, so as to realize the identification and early warning of whether a single-phase grounding fault occurs at each measurement point according to the monitoring data on the substation line; The automatic early warning system includes: A data acquisition module, which is used to collect the phase voltage, phase current, zero-sequence voltage and zero-sequence current at the measurement points corresponding to each sectionalizing switch on the line in real time through the DTU unit deployed in the substation power grid; A single-point analysis module, which is used to first obtain the values of the zero-sequence voltage and zero-sequence current of each measurement point, judge whether the zero-sequence current of any measurement point exceeds a preset threshold, if so, regard it as a risk measurement point, otherwise judge that no single-phase grounding fault has occurred on the current line; then calculate the phase difference between the zero-sequence current and the zero-sequence voltage of each risk node, and finally identify the fault type of each measurement point according to the predicted judgment logic based on the calculated phase difference and the neutral grounding type of the current distribution line; A multi-point analysis module, which includes a single-phase grounding fault detection model based on Transformer and trained using the current line data; the multi-point analysis module is used to input the waveform data of the phase voltage and phase current of each measurement point into the single-phase grounding fault detection model when the single-point analysis module fails to identify the fault types of all measurement points, and generate the fault types of the remaining nodes according to the output of the single-phase grounding fault detection model; An early warning module, which is used to generate corresponding early warning information according to the identification results of single-phase grounding faults at each measurement point on the line output by the single-point analysis module and the multi-point analysis module, and send it to the command center in the background; A decision-making unit, which is used to receive the control instruction issued by the command center according to the received early warning information, and send the control instruction to the FTU unit deployed in the substation power grid. The FTU unit executes the control instruction and adjusts the operating state of relevant power equipment.

10. A diagnostic device for single-phase grounding faults in a distribution line, comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that: When the processor executes the computer program, it realizes the diagnostic method for single-phase grounding fault of a distribution line described in any one of claims 1-8, and further realizes automatically identifying whether a single-phase grounding fault occurs at each measurement point according to the monitoring information of each measurement point in the substation.

Images