Single-phase earth fault recording file screening system and method

Abstract

The invention provides a single-phase earth fault recording file screening system and method, and relates to the technical field of recording file screening. The system comprises a data acquisition module used for acquiring all wave recording files and all alarm signals when a single-phase earth fault occurs in the power distribution network; the first data cleaning module is used for cleaning out-of-limit data of the target wave recording file according to the triggering moment of each alarm signal and / or the starting moment of the target wave recording file to obtain a first wave recording file; the second data cleaning module is used for cleaning out-of-limit data of the target wave recording file according to the alarm signal corresponding to the target wave recording file and the adjacent wave recording file to obtain a second wave recording file; and the third data cleaning module is used for cleaning out-of-limit data of the target wave recording file according to the transient characteristics and the steady-state characteristics of the target wave recording file to obtain a third wave recording file, and obtaining a screened wave recording file according to the first wave recording file to the third wave recording file. According to the method, the noise data in the wave recording file can be effectively eliminated.

Description

Technical Field

[0001] This invention relates to the field of waveform recording screening technology, and in particular to a system and method for screening waveform recordings of single-phase grounding faults. Background Technology

[0002] As the types of equipment in the distribution network increase and the structure of the distribution network becomes more complex, once a single-phase ground fault occurs in the distribution network, it will generate a large number of waveform files, which will bring a large amount of data flow impact to the single-phase ground fault judgment algorithm, affecting the timeliness and accuracy of the judgment.

[0003] Currently, waveform recordings are typically cleaned to remove outlier data. For example, a reasonable threshold range is set for voltage or current data in the recording to eliminate data points outside the range and reduce noise interference. However, this data cleaning method is relatively simple and generally cannot effectively remove noisy data. Sometimes, removing noisy data may also remove actual fault data, thus affecting the accuracy of subsequent single-phase grounding fault location. Summary of the Invention

[0004] This invention provides a system and method for screening single-phase grounding fault waveform recordings, in order to solve the problem of difficulty in effectively removing noisy data when screening waveform recordings corresponding to single-phase grounding faults.

[0005] In a first aspect, embodiments of the present invention provide a single-phase grounding fault waveform file screening system, comprising: a data acquisition module, a first data cleaning module, a second data cleaning module, and a third data cleaning module; The data acquisition module is used to acquire all waveform files and all alarm signals when a single-phase ground fault occurs in the distribution network; The first data cleaning module is used to clean the out-of-limit data of the target waveform file according to the trigger time of each alarm signal and / or the start time of the target waveform file, so as to obtain the first waveform file; The second data cleaning module is used to clean the out-of-limit data of the target waveform file based on the topology of the distribution network, according to the alarm signal corresponding to the target waveform file and adjacent waveform files, to obtain the second waveform file; The third data cleaning module is used to clean the out-of-limit data of the target waveform file according to the transient and steady-state characteristics of the target waveform file to obtain a third waveform file, and to obtain a filtered waveform file based on the first waveform file, the second waveform file and the third waveform file.

[0006] Secondly, embodiments of the present invention provide a method for screening single-phase grounding fault recording files, including: Acquire all waveform recordings and all alarm signals when a single-phase ground fault occurs in the power distribution network; The out-of-limit data of the target waveform file is cleaned according to the trigger time of each alarm signal and / or the start time of the target waveform file to obtain the first waveform file; Based on the topology of the distribution network, the over-limit data of the target waveform file is cleaned according to the alarm signal corresponding to the target waveform file and the adjacent waveform files to obtain the second waveform file; The out-of-limit data of the target waveform file is cleaned according to the transient and steady-state characteristics of the target waveform file to obtain a third waveform file, and a filtered waveform file is obtained based on the first waveform file, the second waveform file and the third waveform file.

[0007] In this embodiment of the invention, all waveform files and all alarm signals when a single-phase ground fault occurs in the distribution network are acquired. Then, the out-of-limit data of the target waveform file is cleaned according to the trigger time of each alarm signal and / or the start time of the target waveform file to obtain a first waveform file. Based on the topology of the distribution network, the out-of-limit data of the target waveform file is cleaned according to the alarm signal corresponding to the target waveform file and adjacent waveform files to obtain a second waveform file. The out-of-limit data of the target waveform file is cleaned according to the transient and steady-state characteristics of the target waveform file to obtain a third waveform file. The first, second, and third waveform files are used to obtain a filtered waveform file. The out-of-limit data in the waveform file can be comprehensively filtered by integrating alarm signals, the waveform file itself, the topology of the distribution network, and the transient and steady-state characteristics of the waveform file. This allows for cross-validation of the waveform file using multi-source data, effectively eliminating noisy data while ensuring the integrity of the fault information, thereby improving the accuracy of subsequent single-phase ground fault location. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of the structure of the single-phase grounding fault recording file screening system provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the implementation of the single-phase grounding fault recording file screening method provided in this embodiment of the invention. Detailed Implementation

[0009] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0010] Figure 1 A schematic diagram of the single-phase grounding fault waveform file screening system provided in an embodiment of the present invention is shown. For ease of explanation, only the parts related to the embodiment of the present invention are shown, and are described in detail below: like Figure 1As shown, the single-phase ground fault waveform file screening system includes: a data acquisition module 11, a first data cleaning module 12, a second data cleaning module 13, and a third data cleaning module 14.

[0011] The data acquisition module 11 is used to acquire all waveform files and all alarm signals when a single-phase ground fault occurs in the distribution network. The first data cleaning module 12 is used to clean the out-of-limit data of the target waveform file according to the trigger time of each alarm signal and / or the start time of the target waveform file, so as to obtain the first waveform file.

[0012] The second data cleaning module 13 is used to clean the out-of-limit data of the target waveform file based on the topology of the distribution network and the alarm signals and adjacent waveform files corresponding to the target waveform file, so as to obtain the second waveform file.

[0013] The third data cleaning module 14 is used to clean the out-of-limit data of the target waveform file according to the transient and steady-state characteristics of the target waveform file, to obtain the third waveform file, and to obtain the filtered waveform file based on the first waveform file, the second waveform file and the third waveform file.

[0014] In this embodiment, the data acquisition module 11 can collect waveform files transmitted by fault indicators, FTUs (feeder terminal units), and DTUs (distribution terminal units) in real time, as well as zero-sequence overcurrent stage I alarm signals, zero-sequence overcurrent stage II alarm signals, substation switch grounding alarm signals, and bus grounding alarm signals from FTUs and DTUs. Relevant data from FTUs and DTUs can be uploaded to the distribution network master station in real time via wireless communication, while data from fault indicators can be obtained from the provincial company's central control station via an interface. Main network-related information such as substation outgoing switch and bus grounding alarm signals can be obtained through the D5000 interface.

[0015] After the data acquisition module 11 collects the waveform recording files and alarm signals, in order to improve the accuracy of single-phase ground fault location using the waveform recording files, data cleaning of the waveform recording files is also required. Each waveform recording file to be cleaned is considered a target waveform recording file. In this embodiment, the out-of-limit data in the waveform recording files is cleaned by fusing alarm signals, the waveform recording files themselves, the topology of the distribution network, and the transient and steady-state characteristics of the waveform recording files. This allows for cross-validation of the waveform recording files using multi-source data, effectively eliminating noisy data while ensuring the integrity of the fault information.

[0016] Optionally, the first data cleaning module 12 is specifically used for: When it is determined that the acquisition time corresponding to the target waveform file contains the trigger time of a certain alarm signal, the data in the target waveform file within a preset range before and after the trigger time are marked according to the trigger time of the alarm signal to obtain the first waveform file.

[0017] And / or, if the out-of-limit data of the target waveform file appears at the start time of the target waveform file, then the out-of-limit data of the target waveform file is retained, and the first waveform file is obtained.

[0018] In this embodiment, the first data cleaning module 12 corresponds to the event-triggered protection mechanism in the data cleaning process. The event-triggered protection mechanism includes alarm signal linkage verification protection and waveform file timestamp association protection.

[0019] The alarm signal linkage verification protection means that when an FTU / DTU zero-sequence overcurrent alarm or a grounding alarm in the station is detected, the voltage / current data within 500ms before and after that moment is automatically marked as "fault characteristic data" and the 3σ principle is not applied to remove it temporarily to avoid accidentally deleting fault transient over-limit data.

[0020] The waveform recording file timestamp association protection means that if the out-of-limit data appears at the start time of the waveform recording file (such as from 0.02s before the fault to 0.1s after the fault), it is determined to be a normal out-of-limit during the fault transient process and is not removed.

[0021] Optionally, the second data cleaning module 13 is specifically used for: Based on the topology of the distribution network, it is determined whether the fault section pointed to by the over-limit data of the target waveform file is consistent with the fault section pointed to by the alarm signal corresponding to the target waveform file, and whether there is over-limit data in the adjacent waveform files corresponding to the target waveform file.

[0022] Based on the judgment results, the out-of-limit data in the target waveform file is cleaned to obtain the second waveform file.

[0023] For example, the second data cleaning module 13 is specifically used for: If the fault segment pointed to by the out-of-limit data in the target waveform file is consistent with the fault segment pointed to by the alarm signal corresponding to the target waveform file, and / or, there is out-of-limit data in the adjacent waveform file corresponding to the target waveform file, then the out-of-limit data of the target waveform file is retained.

[0024] If there is no out-of-limit data in the adjacent waveform files corresponding to the target waveform file, then the out-of-limit data in the target waveform file is removed. If the fault segment pointed to by the out-of-limit data in the target waveform file is inconsistent with the fault segment pointed to by the alarm signal corresponding to the target waveform file, the out-of-limit data in the target waveform file shall be confirmed a second time.

[0025] In this embodiment, the second data cleaning module 13 corresponds to the multi-source data cross-validation in the data cleaning process, including data comparison of equipment on the same line and topology logic verification.

[0026] The data comparison with equipment on the same line is as follows: if the out-of-limit data of a certain equipment does not show any abnormality in the synchronous waveform recording of adjacent FTU / DTU or fault indicator, it is determined that the sensor of that equipment is abnormal or there is communication interference, and it is removed; if multiple equipment show out-of-limit data synchronously, it is confirmed as real fault data and retained.

[0027] Topology logic verification means that, based on the topology of the distribution network, if the over-limit data is consistent with the fault section pointed to by the alarm signal (such as a branch line FTU exceeding the limit and triggering an alarm), the data is retained; if the over-limit data appears in a non-fault section, it is further checked whether it is a false alarm.

[0028] For example, to further verify whether it is a false alarm, one could re-determine the fault segment pointed to by the out-of-limit data, and then determine again whether the fault segment pointed to by the out-of-limit data is consistent with the fault segment pointed to by the alarm signal.

[0029] Optionally, the third data cleaning module 14 is specifically used for: Determine whether the transient characteristics of the target waveform file are transient current polarity changes or voltage drops, and determine whether the steady-state characteristics of the target waveform file have corresponding alarm signals.

[0030] If the transient characteristics of the target waveform file are a sudden change in the polarity of the transient current or a sudden drop in voltage, then the out-of-limit data of the target waveform file is retained as the third waveform file.

[0031] If the steady-state characteristics of the target waveform file do not have a corresponding alarm signal, then the out-of-limit data of the target waveform file will be reconfirmed.

[0032] This embodiment corresponds to the fault feature priority marking in the data cleaning process, including transient feature priority and steady-state limit over-limit secondary confirmation.

[0033] Transient feature priority means that out-of-limit data containing fault characteristics such as transient current polarity change and voltage drop are automatically marked as "high priority" and retained regardless of whether they exceed the 3σ range.

[0034] The steady-state limit violation secondary confirmation means that for steady-state limit violation data without alarm association (such as continuously exceeding the threshold for more than 10 seconds), a secondary confirmation is made to determine whether it is a real fault. If it is not a fault, it is removed.

[0035] For example, during secondary confirmation, the rationality of steady-state over-limit data can be analyzed in conjunction with the distribution network topology (such as line connection relationships and switch status). If the section containing the steady-state over-limit data is in a "hot standby" or "out of service" state (which can be obtained by acquiring the switch position signal through the SCADA system), then the steady-state over-limit data is determined to be a false alarm (such as a false alarm caused by misjudging that the equipment is energized), and the steady-state over-limit data is removed.

[0036] If the section containing the steady-state over-limit data is in "operation" mode and its electrical quantity change trend is consistent with that of the adjacent section (e.g., upstream voltage drop accompanied by downstream current increase), then the steady-state over-limit data is confirmed to correspond to a real fault, and the steady-state over-limit data is retained.

[0037] Optionally, the third data cleaning module 14 is specifically used for: Candidate waveform files are obtained based on the first, second, and third waveform files.

[0038] Obtain the historical data corresponding to the target waveform file, and calculate the mean and standard deviation of the historical data.

[0039] If the data in the candidate waveform file is within the normal operating period, then the data in the candidate waveform file within the normal operating period are further filtered based on the standard deviation of the mean and the first preset multiple.

[0040] If the data in the candidate waveform file falls within the suspected fault period, the data in the candidate waveform file within the suspected fault period will be further filtered based on the standard deviation of the mean and the second preset multiple.

[0041] The filtered waveform files are obtained based on the results of the second filtering.

[0042] This embodiment corresponds to the dynamic threshold adjustment of the time window during the data cleaning process. For normal operating periods (i.e., times without alarms), the 3σ principle is strictly adopted: the mean μ and standard deviation σ of historical data (e.g., the past 24 hours) are calculated, and data exceeding the range of [μ-3σ, μ+3σ] are removed. For periods suspected of faults (i.e., times with alarms or when waveform recording starts), the threshold is temporarily extended to [μ-5σ, μ+5σ], removing only extreme outliers (e.g., noise data exceeding 5σ) while retaining reasonable limits exceeded due to faults (e.g., short-circuit current peaks are typically 3-5 times the rated current).

[0043] This embodiment, through the above criteria, can effectively eliminate noisy data while ensuring the integrity of fault information, thereby helping to improve the accuracy of grounding fault assessment.

[0044] Optionally, the single-phase ground fault recording file screening system also includes a fault location module 15, which is used for: Based on the topology of the distribution network and the alarm level and alarm status corresponding to each alarm signal, the first fault location result is obtained.

[0045] Extract transient and steady-state features from the screened waveform files, and obtain the second fault location result based on the transient and steady-state features.

[0046] Based on the first fault location result and the second fault location result, the final fault location result is determined.

[0047] Optionally, the fault location module 15 is specifically used for: Based on the distribution network topology, a topology tree is constructed with the substation switch as the root node, following the hierarchy of "substation switch - main line distribution terminal - branch line distribution terminal".

[0048] For each alarm signal corresponding to a node in the topology tree, compare them according to alarm time and alarm level, and obtain preliminary fault location results based on the comparison results.

[0049] For each fault segment in the preliminary fault location results, the alarm score of the fault segment is calculated based on the alarm level and alarm status corresponding to each alarm signal in the fault segment.

[0050] The alarm scores of each faulty section are compared, and the first fault location result is obtained based on the comparison results.

[0051] The following example illustrates the process of obtaining the first fault location result: First, with the station switch as the root node, a topology tree is constructed according to the hierarchy of "station switch → main line FTU → branch line FTU", and each node records the branch relationship of its line.

[0052] Then, for each alarm level, the alarm signals are sorted in ascending order of timestamp (that is, all zero-sequence overcurrent stage I alarm signals are sorted in ascending order of timestamp, and all zero-sequence overcurrent stage II alarm signals are sorted in ascending order of timestamp), and the topology tree is traversed starting from the root node: If the time difference between the grounding alarm signal of the station switch and the zero-sequence overcurrent stage I alarm of the FTU is ≤5ms, it is determined that the same fault event has been triggered.

[0053] If a branch FTU has the earliest alarm time (e.g., t=10.0ms) and its upstream node (e.g., station switch) has already triggered an alarm, then the branch is determined to be a fault area.

[0054] If a branch FTU triggers an alarm earliest (e.g., t=10.0ms), and subsequent branch FTUs do not trigger alarms, then the faulty section is the line between that FTU and the next node.

[0055] If multiple branch FTUs alarm simultaneously (e.g., the time difference between multiple branch FTU alarms is ≤0.1ms), then the branch containing the deepest FTU in the topology tree is taken as the priority fault area, or the branch containing the earliest alarming FTU is taken as the priority fault area.

[0056] If multiple FTU zero-sequence overcurrent stage II alarms are triggered consecutively within 10ms, an "alarm chain" is constructed, and its common path (such as the intersection of branch lines) is the suspected fault segment.

[0057] In this embodiment, by comparing the alarm signals corresponding to each node in the topology tree according to alarm time and alarm level, a preliminary fault location result is obtained based on the comparison result, enabling rapid coarse location. Utilizing the propagation timing characteristics of fault signals (fault current propagates from the fault point to the power supply side), and through the logic of "earliest alarm is closest to the fault point," the branch where the fault is located can be quickly locked within 10ms, resulting in high location efficiency. This is suitable for single-branch fault scenarios. An "alarm chain" priority mechanism is introduced, and when multiple FTUs alarm, they are sorted according to "Segment I alarm > Segment II alarm" and "earliest time > latest time," which can quickly prune non-faulty branches, improving traversal efficiency by more than 30%. Topology constraint filtering can also be performed, combined with the topology tree structure, to avoid misjudging non-faulty branches. For example, if the alarm time of FTU3 is earlier than that of FTU2, but the branch where FTU3 is located is downstream of FTU2, then the fault area is the line between FTU2 and FTU3.

[0058] Optionally, the fault location module 15 is specifically used for: For each fault segment in the preliminary fault location results, the alarm level weight corresponding to each alarm signal in the fault segment is determined according to the alarm level.

[0059] Calculate the product of the alarm level weight and the alarm status for each alarm signal within the faulty section to obtain the alarm score for each alarm signal within the faulty section.

[0060] Calculate the sum of the alarm scores for each alarm signal within the faulty section to obtain the alarm score for that faulty section.

[0061] Specifically, the alarm level corresponding to each alarm signal in the fault section and the current meteorological data of the distribution network can be input into the preset alarm level weight optimization model, and the output of the preset alarm level weight optimization model can be used as the alarm level weight corresponding to the alarm level. The preset alarm level weight optimization model is trained by machine learning based on the historical fault data and historical meteorological data of the distribution network.

[0062] The alarm scores of each fault segment can be compared. If there is only one highest alarm score, the fault segment corresponding to the highest alarm score is determined as the first fault location result.

[0063] If there are at least two highest alarm scores, compare the alarm levels of the alarm signals contained in the fault segment corresponding to each highest alarm score, and determine the fault segment corresponding to the highest alarm score with the highest alarm level as the first fault location result.

[0064] In this embodiment, the alarm signals sent by each power distribution terminal within each fault section are quantified, and the alarm status of each device is converted into a numerical value. For example, quantification can be achieved by setting the alarm level weight corresponding to the alarm level to a preset value. For instance, if the alarm level of the alarm signal is a zero-sequence overcurrent stage I alarm, the alarm level weight can be recorded as 2 points; if the alarm level of the alarm signal is a zero-sequence overcurrent stage II alarm, the alarm level weight can be recorded as 1 point; and if the alarm level of the alarm signal is a normal state, the alarm level weight can be recorded as 0 points. Alternatively, historical fault data and historical meteorological data can be combined, and the alarm level weight can be adaptively adjusted through machine learning (e.g., the stage I alarm weight is increased to 3 during thunderstorms) to adapt to changes in fault characteristics under different operating scenarios.

[0065] Then, for each possible segment in the distribution network (such as the FTU1-FTU2 line), the alarm scores of all equipment within that segment are accumulated: .

[0066] Among them, is the first Alarm scores for all devices within each section. For the first Within each section Alarm level weights for individual devices (for example, Segment I = 2, Segment II = 1). For the first Within each section The alarm status of the device indicates the alarm status of the first device. Within each section Does the device issue an alarm (for example, 1 = yes, 0 = no)?

[0067] Then compare the scores of all segments, and the segment with the highest score is the faulty segment; if the scores are the same, the segment containing segment I alarm is selected first.

[0068] In this embodiment, alarm level weights reflect the severity of alarms (Level I > Level II), which are positively correlated with the distance to the fault point (the closer to the fault point, the greater the zero-sequence current, and the easier it is to trigger a higher-level alarm). This quantifies the weight of different alarms and distinguishes between high and low fault probabilities. Alarm status reflects whether the device has actually triggered an alarm, which is used to filter valid signals and devices that have not triggered alarms, avoiding interference from invalid data.

[0069] In this embodiment, preliminary location is performed based on a topology tree. The location logic is based on signal propagation timing and topology hierarchy, which is suitable for single-branch faults and scenarios with accurate timestamps. The location accuracy can reach the branch level (error ±500m), but it depends on the accuracy of timestamps and is easily affected by communication delays. Alternatively, location is performed by calculating the alarm score of the fault section. The location logic is a quantitative decision based on the alarm level and the number of alarm states. This is suitable for scenarios with multi-branch faults and complex alarm signals, and the location accuracy can reach the section level (error ±200m). The location process is not sensitive to timing errors, but it depends on the integrity of alarm signals. Combining the two methods, the branch is first locked using a timestamp algorithm (i.e., preliminary location based on the topology tree), and then the section is refined using weighted voting. This is suitable when there are multiple grounding points or communication delays in the distribution network. After the timestamp algorithm narrows the range, weighted voting further refines the location. The timestamp algorithm resists interference through topology constraints, and weighted voting reduces misjudgments through multi-signal averaging. Combining the two methods improves accuracy.

[0070] The following are device embodiments of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0071] See Figure 2 The flowchart illustrating the implementation of the single-phase grounding fault waveform file filtering method provided in this embodiment of the invention is described in detail below: Step 201: Obtain all waveform files and all alarm signals when a single-phase ground fault occurs in the distribution network.

[0072] Step 202: Clean the out-of-limit data of the target waveform file according to the trigger time of each alarm signal and / or the start time of the target waveform file to obtain the first waveform file.

[0073] Step 203: Based on the topology of the distribution network, clean the over-limit data of the target waveform file according to the alarm signal corresponding to the target waveform file and the adjacent waveform files to obtain the second waveform file.

[0074] Step 204: Clean the out-of-limit data of the target waveform file according to the transient and steady-state characteristics of the target waveform file to obtain the third waveform file, and obtain the filtered waveform file based on the first, second and third waveform files.

[0075] In one possible implementation, step 202 may include: When it is determined that the acquisition time corresponding to the target waveform file contains the trigger time of a certain alarm signal, the data within a preset range before and after the trigger time in the target waveform file are marked according to the trigger time of the alarm signal to obtain the first waveform file.

[0076] And / or, if the out-of-limit data of the target waveform file appears at the start time of the target waveform file, then the out-of-limit data of the target waveform file is retained to obtain the first waveform file.

[0077] In one possible implementation, step 203 may include: Based on the topology of the distribution network, it is determined whether the fault segment pointed to by the over-limit data of the target waveform file is consistent with the fault segment pointed to by the alarm signal corresponding to the target waveform file, and it is also determined whether there is over-limit data in the adjacent waveform files corresponding to the target waveform file.

[0078] Based on the judgment result, the out-of-limit data of the target waveform file is cleaned to obtain a second waveform file.

[0079] In one possible implementation, step 203 may include: If the fault segment pointed to by the out-of-limit data of the target waveform file is consistent with the fault segment pointed to by the alarm signal corresponding to the target waveform file, and / or, there is out-of-limit data in the adjacent waveform file corresponding to the target waveform file, then the out-of-limit data of the target waveform file is retained.

[0080] If there is no out-of-limit data in the adjacent waveform files corresponding to the target waveform file, then the out-of-limit data in the target waveform file is removed.

[0081] If the fault segment pointed to by the out-of-limit data of the target waveform file is inconsistent with the fault segment pointed to by the alarm signal corresponding to the target waveform file, then the out-of-limit data of the target waveform file shall be confirmed a second time.

[0082] In one possible implementation, step 204 may include: Determine whether the transient characteristics of the target waveform file are transient current polarity changes or voltage drops, and determine whether the steady-state characteristics of the target waveform file have corresponding alarm signals.

[0083] If the transient characteristics of the target waveform file are transient current polarity change or voltage drop, then the out-of-limit data of the target waveform file is retained as the third waveform file.

[0084] If the steady-state characteristics of the target waveform file do not have a corresponding alarm signal, then the out-of-limit data of the target waveform file shall be reconfirmed.

[0085] In one possible implementation, step 204 may include: Candidate waveform files are obtained based on the first waveform file, the second waveform file, and the third waveform file.

[0086] Obtain the historical data corresponding to the target waveform file, and calculate the mean and standard deviation of the historical data.

[0087] If the data in the candidate waveform file is within the normal operating period, then the data in the candidate waveform file within the normal operating period are further filtered according to the mean and the standard deviation of the first preset multiple.

[0088] If the data in the candidate waveform file is within a suspected fault period, then the data in the candidate waveform file within the suspected fault period are further filtered based on the mean and the standard deviation of the second preset multiple.

[0089] The filtered waveform files are obtained based on the results of the second filtering.

[0090] One possible implementation also includes: Based on the topology of the distribution network and the alarm level and alarm status corresponding to each alarm signal, the first fault location result is obtained.

[0091] Extract the transient and steady-state features of the filtered waveform file, and obtain the second fault location result based on the transient and steady-state features.

[0092] Based on the first fault location result and the second fault location result, the final fault location result is determined.

[0093] In one possible implementation, based on the distribution network topology, a topology tree is constructed with the substation switch as the root node, following the hierarchy of "substation switch - main line distribution terminal - branch line distribution terminal".

[0094] For each alarm signal corresponding to a node in the topology tree, the alarm time and alarm level are compared, and a preliminary fault location result is obtained based on the comparison result.

[0095] For each fault segment in the preliminary fault location results, the alarm score of the fault segment is calculated based on the alarm level and alarm status corresponding to each alarm signal in the fault segment.

[0096] The alarm scores of each faulty section are compared, and the first fault location result is obtained based on the comparison results.

[0097] In one possible implementation, for each fault segment in the preliminary fault location result, an alarm level weight corresponding to the alarm level is determined based on the alarm level corresponding to each alarm signal within the fault segment.

[0098] Calculate the product of the alarm level weight and the alarm status corresponding to each alarm signal in the fault section to obtain the alarm score for each alarm signal in the fault section.

[0099] The alarm score of the faulty section is obtained by summing the alarm scores of each alarm signal within the faulty section.

[0100] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0101] For the sake of simplicity and clarity, only the above-described functional modules / units are used as examples. In practical applications, the functions described above can be assigned to different functional modules / units as needed. These modules / units can be implemented in hardware, software, or a combination of both.

[0102] In the above embodiments, the descriptions of each embodiment have their own emphasis. Parts not detailed or described in a particular embodiment can be referred to in the relevant descriptions of other embodiments. Unless otherwise specified or in conflict with logic, the terminology and / or descriptions between different embodiments are consistent and can be referenced interchangeably. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0103] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A system for screening single-phase ground fault recording files, characterized in that, include: The system comprises a data acquisition module, a first data cleaning module, a second data cleaning module, and a third data cleaning module. The data acquisition module is used to acquire all waveform files and all alarm signals when a single-phase ground fault occurs in the distribution network; The first data cleaning module is used to clean the out-of-limit data of the target waveform file according to the trigger time of each alarm signal and / or the start time of the target waveform file, so as to obtain the first waveform file; The second data cleaning module is used to clean the out-of-limit data of the target waveform file based on the topology of the distribution network, according to the alarm signal corresponding to the target waveform file and adjacent waveform files, to obtain the second waveform file; The third data cleaning module is used to clean the out-of-limit data of the target waveform file according to the transient and steady-state characteristics of the target waveform file to obtain a third waveform file, and to obtain a filtered waveform file based on the first waveform file, the second waveform file and the third waveform file.

2. The single-phase grounding fault recording file screening system according to claim 1, characterized in that, The first data cleaning module is specifically used for: When it is determined that the acquisition time corresponding to the target waveform file contains the trigger time of a certain alarm signal, the data within a preset range before and after the trigger time in the target waveform file are marked according to the trigger time of the alarm signal to obtain the first waveform file; And / or, if the out-of-limit data of the target waveform file appears at the start time of the target waveform file, then the out-of-limit data of the target waveform file is retained to obtain the first waveform file.

3. The single-phase grounding fault recording file screening system according to claim 1, characterized in that, The second data cleaning module is specifically used for: Based on the topology of the distribution network, it is determined whether the fault segment pointed to by the over-limit data of the target waveform file is consistent with the fault segment pointed to by the alarm signal corresponding to the target waveform file, and it is also determined whether there is over-limit data in the adjacent waveform files corresponding to the target waveform file. Based on the judgment result, the out-of-limit data of the target waveform file is cleaned to obtain a second waveform file.

4. The single-phase grounding fault recording file screening system according to claim 3, characterized in that, The second data cleaning module is specifically used for: If the fault segment pointed to by the out-of-limit data of the target waveform file is consistent with the fault segment pointed to by the alarm signal corresponding to the target waveform file, and / or, there is out-of-limit data in the adjacent waveform file corresponding to the target waveform file, then the out-of-limit data of the target waveform file is retained. If there is no out-of-limit data in the adjacent waveform files corresponding to the target waveform file, then the out-of-limit data in the target waveform file is removed. If the fault segment pointed to by the out-of-limit data of the target waveform file is inconsistent with the fault segment pointed to by the alarm signal corresponding to the target waveform file, then the out-of-limit data of the target waveform file shall be confirmed a second time.

5. The single-phase grounding fault recording file screening system according to claim 4, characterized in that, The third data cleaning module is specifically used for: Determine whether the transient characteristics of the target waveform file are transient current polarity change or voltage drop, and determine whether the steady-state characteristics of the target waveform file have corresponding alarm signals; If the transient characteristics of the target waveform file are transient current polarity change or voltage drop, then the out-of-limit data of the target waveform file is retained as the third waveform file. If the steady-state characteristics of the target waveform file do not have a corresponding alarm signal, then the out-of-limit data of the target waveform file shall be reconfirmed.

6. The single-phase grounding fault recording file screening system according to claim 1, characterized in that, The third data cleaning module is specifically used for: Candidate waveform files are obtained based on the first waveform file, the second waveform file, and the third waveform file; Obtain the historical data corresponding to the target waveform file, and calculate the mean and standard deviation of the historical data; If the data in the candidate waveform file is in the normal operating period, the data in the candidate waveform file that is in the normal operating period are further filtered according to the mean and the standard deviation of the first preset multiple; If the data in the candidate waveform file is in the suspected fault period, the data in the candidate waveform file in the suspected fault period are further filtered according to the standard deviation of the mean and the second preset multiple. The filtered waveform files are obtained based on the results of the second filtering.

7. The single-phase grounding fault recording file screening system according to claim 1, characterized in that, It also includes a fault location module, which is used for: Based on the topology of the distribution network and the alarm level and alarm status corresponding to each alarm signal, the first fault location result is obtained; Extract the transient and steady-state features of the screened waveform files, and obtain the second fault location result based on the transient and steady-state features; Based on the first fault location result and the second fault location result, the final fault location result is determined.

8. The single-phase grounding fault recording file screening system according to claim 7, characterized in that, The fault location module is specifically used for: Based on the topology of the distribution network, a topology tree is constructed with the station switch as the root node and in the hierarchy of "station switch - main line distribution terminal - branch line distribution terminal". For each node in the topology tree, the alarm signals are compared according to alarm time and alarm level, and preliminary fault location results are obtained based on the comparison results. For each fault segment in the preliminary fault location results, the alarm score of the fault segment is calculated based on the alarm level and alarm status corresponding to each alarm signal in the fault segment. The alarm scores of each faulty section are compared, and the first fault location result is obtained based on the comparison results.

9. The single-phase grounding fault recording file screening system according to claim 8, characterized in that, The fault location module is specifically used for: For each fault segment in the preliminary fault location results, the alarm level weight corresponding to each alarm signal in the fault segment is determined according to the alarm level. Calculate the product of the alarm level weight and the alarm status corresponding to each alarm signal in the fault section to obtain the alarm score of each alarm signal in the fault section; The alarm score of the faulty section is obtained by summing the alarm scores of each alarm signal within the faulty section.

10. A method for screening single-phase ground fault recording files, characterized in that, include: Acquire all waveform recordings and all alarm signals when a single-phase ground fault occurs in the power distribution network; The out-of-limit data of the target waveform file is cleaned according to the trigger time of each alarm signal and / or the start time of the target waveform file to obtain the first waveform file; Based on the topology of the distribution network, the over-limit data of the target waveform file is cleaned according to the alarm signal corresponding to the target waveform file and the adjacent waveform files to obtain the second waveform file; The out-of-limit data of the target waveform file is cleaned according to the transient and steady-state characteristics of the target waveform file to obtain a third waveform file, and a filtered waveform file is obtained based on the first waveform file, the second waveform file and the third waveform file.

Images