Batch fault identification method of smart electric energy meters based on DBSCAN algorithm

Abstract

The present application relates to a method for identifying batch faults of smart electric energy meters based on the DBSCAN algorithm, including the following specific steps: data governance: obtaining a set of error values ​​of dismantled meters and in-service meters; file association: obtaining the manufacturer, production year, model and production batch information of the electric energy meters; batch aggregation: according to the definition of the electric energy meter batch, the data of the same electric energy meter batch are composed; cluster analysis: using the electric energy meter batch as a unit, clustering the electric energy meter error values ​​using the DBSCAN algorithm; fault type: using the DBSCAN algorithm to perform secondary clustering on the current error value and voltage error value of the batch of electric energy meters, respectively, to determine the fault type of the batch of electric energy meters; end stage: outputting the batch with faults and issuing an early warning for this batch. The present application has advanced technology and novel means, and provides a reliable solution for batch quality monitoring of smart electric energy meters.

Description

Technical Field

[0001] The present application relates to the field of batch quality monitoring of electric energy meters, and in particular to a method for identifying batch faults of smart electric energy meters based on a DBSCAN algorithm. Background Art

[0002] In industrial production, aging is usually used to reduce early failures. Due to the design defects of the energy meter and the improper production process, the problem of the energy meter is generally exposed after a long time. This time can be shortened by applying certain stresses during the aging process. In the actual life cycle of the energy meter, it is installed on site for use after the first inspection. Some meters that were missed during the factory calibration can be screened and identified. However, due to factors such as insufficient aging time or incomplete aging conditions, even if the energy meter passes the first inspection, the entire batch may still be in the early failure stage, and it cannot be effectively eliminated before installation. After these batches of energy meters are installed on site, the usage time has not reached the designed service life, and the fault phenomenon is concentrated or the failure rate exceeds the requirements of the national grid enterprise standard for energy meters. We call this type of batch early failure problem batch failure.

[0003] A batch failure warning method for electric meters in operation in the prior art relies on the failure rate. The failure rate is obtained through verification and the cumulative failure rate is calculated. It is triggered only when the failure rate reaches the warning value. Usually the warning value is high and a lot of samples need to be accumulated. The cycle of discovering the batch failure rate is long.

[0004] A method for diagnosing batch faults of in-service meters based on the operating characteristics of disassembled meters is highly dependent on sorting data. It requires that each batch of meters be randomly disassembled for sorting at regular intervals, otherwise the sorting data obtained is a severely biased sample. Using biased samples to train and predict the entire set may result in large deviations in the prediction results. The features extracted from the operating data of the electric energy meter, such as current, voltage, and power, have a good degree of discrimination for some typical faults. However, for faults with out-of-tolerance errors, it may not be possible to effectively distinguish whether an out-of-tolerance error has occurred.

[0005] A method for evaluating the operating status of a batch of smart meters uses the dispersion of errors as a feature, but the vast majority of electricity meters in the same batch are normal, that is, they are very concentrated within the qualified range. Summary of the invention

[0006] The purpose of the embodiment of the present application is to provide a method for identifying batch faults of smart electricity meters based on the DBSCAN algorithm, which can identify batch anomalies of electricity meters approaching error tolerance earlier, play a good auxiliary role in the batch quality monitoring of electricity meters, and can focus on monitoring batches of electricity meters with faults caused by defects, saving manpower and material resources required for sampling inspection.

[0007] To achieve the above objectives, this application provides the following technical solutions:

[0008] The embodiment of the present application provides a method for identifying batch faults of smart electric energy meters based on the DBSCAN algorithm, comprising the following specific steps:

[0009] Data governance: Read the sorting data of the fault meter with fault types of error exceeding tolerance and inaccurate indication from the database, read the calculation results of the big data real-time monitoring system from the ES, obtain all records of the error exceeding tolerance of the electric energy meter, retain the fault type of the electric energy meter, and obtain the error value set of the dismantled meter and the in-service meter;

[0010] File association: Obtain the electric energy meter file information from the marketing system, associate the fault meter record with the electric energy meter file with the electric energy meter barcode as the primary key, and obtain the manufacturer, production year, model and production batch information of the electric energy meter;

[0011] Batch aggregation: According to the definition of the electric energy meter batch, the fault table records with the same manufacturer and production batch, or the same manufacturer, production year and meter model are spliced ​​together to form the data of the same electric energy meter batch;

[0012] Cluster analysis: Taking the batch of electric energy meters as the unit, the DBSCAN algorithm is used to cluster the error values ​​of the electric energy meters to obtain one or more clusters of electric energy meters. If the proportion of the number of electric energy meters in a certain cluster to the total number of faulty meters is higher than the upper threshold, and the standard deviation of the excess error values ​​of the electric energy meters in the cluster is lower than the lower threshold, then the electric energy meter batch is marked as having batch failure; if the proportion of the number of electric energy meters in a certain cluster to the total number of faulty meters is not higher than the upper threshold, and the standard deviation of the excess error values ​​of the electric energy meters in the cluster is not lower than the lower threshold, then the end stage step is performed;

[0013] Fault type: The current error value and voltage error value of the batch of electric energy meters are clustered twice using the DBSCAN algorithm to determine whether the current error value and voltage error value of the batch of electric energy meters also have a relatively concentrated distribution. If the current error value distribution is concentrated, it is marked as batch current measurement abnormality; if the voltage error value distribution is concentrated, it is marked as batch voltage measurement abnormality; current and voltage measurement abnormalities are marked at the same time; if neither the current nor the voltage error value distribution is concentrated, it is marked as electricity measurement abnormality;

[0014] Ending stage: If the analysis of all batches of electricity meters is completed, the batches with faults are output and an early warning is issued for this batch; if the analysis of all batches of electricity meters is not completed, return to the cluster analysis step to analyze the unfinished batches of electricity meters.

[0015] In the data management step, the results where the active power verification is normal and the basic power error exceeds the error value of -50%, -25%, +33%, -33%, and -100% are eliminated.

[0016] The clustering of the error values ​​of the electric energy meters using the DBSCAN algorithm is specifically as follows:

[0017] a) Input the error value of all the meters in this batch of meters;

[0018] b) Mark all error values ​​as unvisited;

[0019] c) Traverse each error value, count the number of other error value data points whose difference with the error value is less than or equal to 2%, and check whether the number of data points is at least 5. If so, mark the error value data point as a core point;

[0020] d) Identify data points with error values ​​less than or equal to 2% from core points and not core points, and mark them as boundary points;

[0021] e) Starting from any unvisited core point, add the points with a difference of less than or equal to 2% to the same cluster to form a cluster; continue to process the boundary points of the cluster, and add the points with a difference of less than or equal to the threshold of 2% to the cluster;

[0022] f) Mark the noise points that are not assigned to any cluster;

[0023] g) Repeat steps d~f until all points are visited and classified;

[0024] h) Finally, a clustering result of a set of data points is obtained, in which each point is assigned to a cluster or marked as a noise point, and the classification result is output.

[0025] The upper threshold is 20%, and the lower threshold is 2%.

[0026] Compared with the prior art, the beneficial effect of the present application is: by clustering the calibration errors of the electric energy meter sorting data or the errors calculated by real-time monitoring using big data technology, it is determined whether the electric energy meters in the batch have the same failure trend. The method is advanced in technology and novel in means, and provides a reliable solution for batch quality monitoring of smart meters. The present application can identify batch anomalies of electric energy meters approaching out-of-tolerance errors earlier, and has a good auxiliary effect on batch quality monitoring of electric energy meters. It can focus on monitoring batches of electric energy meters with failures caused by defects, saving manpower and material resources required for sampling inspection. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0028] Figure 1 is a flow chart of a method provided by an embodiment of the present invention;

[0029] Figure 2 is a file association flow chart provided by an embodiment of the present invention;

[0030] Figure 3 It is a batch polymerization flow chart provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0032] The terms "comprises," "comprising," or any other variation thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0033] The terms "first", "second", etc. are only used to distinguish one entity or operation from another entity or operation, and should not be understood as indicating or implying relative importance, nor should they be understood as requiring or implying any such actual relationship or order between these entities or operations.

[0034] See also Figure 1-Figure 3 The embodiment of the present application provides a method for identifying batch faults of smart electric energy meters based on the DBSCAN algorithm. From the perspective of batch quality monitoring of smart electric energy meters, the distribution characteristics of error values ​​obtained by the disassembly and sorting of the same batch and the real-time error monitoring system of big data in operation are proposed, and a method and application for clustering analysis of error values ​​and identifying batch faults are proposed. And it is used for batch fault warning.

[0035] Based on the sorting errors of dismantled electric energy meters and the inaccurate operation errors of operating electric energy meters, the present invention performs cluster analysis on samples with out-of-tolerance errors. DBSCAN cluster analysis is performed on batches of electric energy meters, and the degree of clustering of error values ​​is used to determine whether the batch may have abnormalities. Then, the current error values ​​and voltage error values ​​are clustered again to determine the type of measurement abnormality that the batch may have (current measurement abnormality, voltage measurement abnormality, or shared module abnormality). According to the explanation of electric energy meter batches in the counting sampling inspection procedures of various network provinces of the State Grid, electric energy meter batches are defined as the following two types:

[0036] Electricity meters from the same manufacturer and the same production batch;

[0037] Electricity meters of the same manufacturer, same production year and same model.

[0038] Delete the error-exceeding-tolerance data caused by constants, pulse port communication and other issues in the sorted data. Because the error-exceeding-tolerance values ​​of these faulty meters are caused by constant setting errors or pulse port errors, noise will be introduced in the error value clustering process, and these abnormal samples need to be deleted in the data governance stage. At the same time, cluster analysis is meaningful only when the number of electric energy meters with error-exceeding-tolerance reaches a certain scale, and the results are more reliable. The batch failure rate also increases to a level worthy of focusing on monitoring its quality. Therefore, the analysis step is temporarily skipped for batches with fewer faulty meters.

[0039] The batch fault identification scheme for clustering error values ​​of smart electric energy meters based on the DBSCAN algorithm is as follows:

[0040] Data governance: Read the sorting data (electricity meter barcode, fault type, basic error and other fields) of the fault meters with fault types of error exceeding tolerance and inaccurate indication from the database, and remove the results with normal active power verification and basic power error exceeding tolerance of some common constant error values ​​such as -50%, -25%, +33%, -33%, -100%. Read the calculation results of the big data real-time monitoring system (electricity meter barcode, fault type, basic error and other fields) from ES, obtain all records of electric energy meter error exceeding tolerance, and retain the fault type of the electric energy meter. Combine the two sets to obtain the error value set of the dismantled meter and the in-service meter.

[0041] File association: Get the electric energy meter file information from the marketing system, associate the fault meter record with the electric energy meter file with the electric energy meter barcode as the primary key, and obtain the manufacturer, production year, model and production batch information of the electric energy meter. The schematic diagram is as follows Figure 2 ;

[0042] Batch aggregation: According to the definition of the meter batch, the fault meter records with the same manufacturer and production batch, or the same manufacturer, production year, and meter model are spliced ​​together to form the data of the same meter batch, and the batches with the number of fault meters below a certain threshold <= 20 are filtered out, such as Figure 3 As shown;

[0043] Cluster analysis: Taking the batch of electric energy meters as the unit, the DBSCAN algorithm is used to cluster the error values ​​of the electric energy meters:

[0044] Enter the error value of all the energy meters in this energy meter batch;

[0045] Mark all error values ​​as unvisited;

[0046] Traverse each error value, count the number of other error value data points whose difference with the error value is less than or equal to a certain threshold <= 2%, check whether the number of data points is at least a certain number (5), if so, mark the error value data point as a core point;

[0047] Identify the data points whose difference with the core points is less than or equal to a certain threshold <= 2% and are not core points, and mark them as boundary points;

[0048] Starting from any unvisited core point, add the points whose difference with it is less than or equal to the threshold value <= 2% to the same cluster to form a cluster; continue to process the boundary points of the cluster, and add the points whose difference with the boundary points is less than or equal to the threshold value <= 2% to the cluster;

[0049] Mark out the noise points that are not assigned to any cluster;

[0050] Repeat steps d~f until all points are visited and classified;

[0051] Finally, a clustering result of a set of data points is obtained, in which each point is assigned to a cluster or marked as a noise point, and the classification result is output.

[0052] Get one or more clusters of electric energy meters. If the proportion of the number of electric energy meters in a cluster to the total number of faulty meters is higher than a certain threshold value>=20%, and the standard deviation of the excess value of the electric energy meters in the cluster is lower than a certain threshold value<=2%, then the batch of electric energy meters is marked as having batch faults; otherwise, skip the fault type step and proceed to the end stage step;

[0053] Fault type: Use the DBSCAN algorithm to perform secondary clustering on the current error value and voltage error value of the batch of electric energy meters to determine whether the current error value and voltage error value of the batch of electric energy meters also have a relatively concentrated distribution. If the current error value distribution is concentrated, it is marked as batch current measurement abnormality; if the voltage error value distribution is concentrated, it is marked as batch voltage measurement abnormality; current and voltage measurement abnormalities can be marked at the same time; if neither the current nor the voltage error value distribution is concentrated, it is marked as electricity measurement abnormality;

[0054] Ending stage: If all the batches of electric energy meters have been analyzed, the batches with faults are output and an early warning is issued for this batch; otherwise, the cluster analysis step is returned to analyze the unfinished batches of electric energy meters.

[0055] From the perspective of batch quality monitoring of smart energy meters, this technical solution creatively proposes a method to extract the distribution characteristics of batch energy meter error values ​​and identify faulty energy meter batches based on the sorting system and big data real-time monitoring system of the State Grid using the DBSCAN clustering algorithm. Using the DBSCAN clustering algorithm, cluster analysis is performed on the error values ​​of smart energy meters in the same batch, and combined with the operating status of the energy meters in operation in the big data real-time monitoring system, it is determined whether there is a batch fault and the type of fault in the energy meter batch. This technical solution can reliably and quickly identify batches of energy meters with the same defects and failure modes. When the batch failure rate is low, it is possible to timely discover whether there is a batch failure problem in the batch, without waiting until the batch failure rate accumulates to a high level before issuing an early warning. Because these energy meters have the same failure mode, after a clear analysis of the failure mode, feedback can be given to manufacturers to improve related problems and improve production quality. In the batch quality monitoring program of smart energy meters, it can effectively assist in locating the batches of energy meters that need to be closely watched, giving full play to the value of sorting data and big data analysis and diagnosis systems, and saving the manpower and material resources required for counting sampling inspection.

[0056] The above description is only an embodiment of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for identifying batch faults of smart electric energy meters based on DBSCAN algorithm, characterized in that: The specific steps include: Data governance: Read the sorting data of the fault meter with fault types of error exceeding tolerance and inaccurate indication from the database, read the calculation results of the big data real-time monitoring system from the ES, obtain all records of the error exceeding tolerance of the electric energy meter, retain the fault type of the electric energy meter, and obtain the error value set of the dismantled meter and the in-service meter; File association: Obtain the electric energy meter file information from the marketing system, associate the fault meter record with the electric energy meter file with the electric energy meter barcode as the primary key, and obtain the manufacturer, production year, model and production batch information of the electric energy meter; Batch aggregation: According to the definition of the electric energy meter batch, the fault table records with the same manufacturer and production batch, or the same manufacturer, production year and meter model are spliced ​​together to form the data of the same electric energy meter batch; Cluster analysis: Taking the batch of electric energy meters as the unit, the DBSCAN algorithm is used to cluster the error values ​​of the electric energy meters to obtain one or more clusters of electric energy meters. If the proportion of the number of electric energy meters in a certain cluster to the total number of faulty meters is higher than the upper threshold, and the standard deviation of the excess error values ​​of the electric energy meters in the cluster is lower than the lower threshold, then the electric energy meter batch is marked as having batch failure; if the proportion of the number of electric energy meters in a certain cluster to the total number of faulty meters is not higher than the upper threshold, and the standard deviation of the excess error values ​​of the electric energy meters in the cluster is not lower than the lower threshold, then the end stage step is performed; Fault type: The current error value and voltage error value of the batch of electric energy meters are clustered twice using the DBSCAN algorithm to determine whether the current error value and voltage error value of the batch of electric energy meters also have a relatively concentrated distribution. If the current error value distribution is concentrated, it is marked as batch current measurement abnormality; If the voltage error value distribution is concentrated, it is marked as batch voltage measurement abnormality; Current and voltage measurement anomalies are marked simultaneously; If the current and voltage error values ​​are not distributed in a concentrated manner, it is marked as an abnormality in power measurement; Ending stage: If all the batches of electric energy meters have been analyzed, the batches with faults will be output and an early warning will be issued for this batch; If all the batches of electric energy meters have not been analyzed, return to the cluster analysis step to analyze the unfinished batches of electric energy meters.

2. The method for identifying batch faults of smart electric energy meters based on the DBSCAN algorithm according to claim 1, characterized in that: In the data management step, the results where the active power verification is normal and the basic power error exceeds the error value of -50%, -25%, +33%, -33%, and -100% are eliminated.

3. The method for identifying batch faults of smart electric energy meters based on the DBSCAN algorithm according to claim 1 is characterized in that: The clustering of the error values ​​of the electric energy meters using the DBSCAN algorithm is specifically as follows: a) Input the error value of all the energy meters in this energy meter batch; b) Mark all error values ​​as unvisited; c) Traverse each error value, count the number of other error value data points whose difference with the error value is less than or equal to 2%, and check whether the number of data points is at least 5. If so, mark the error value data point as a core point; d) Identify the data points whose error values ​​are less than or equal to 2% from the core points and are not core points, and mark them as boundary points; e) Starting from any unvisited core point, add the points with a difference of less than or equal to 2% to the same cluster to form a cluster; continue to process the boundary points of the cluster, and add the points with a difference of less than or equal to the threshold of 2% to the cluster; f) Mark the noise points that are not assigned to any cluster; g) Repeat steps d~f until all points are visited and classified; h) Finally, a clustering result of a set of data points is obtained, in which each point is assigned to a cluster or marked as a noise point, and the classification result is output.

4. The method for identifying batch faults of smart electric energy meters based on the DBSCAN algorithm according to claim 1, characterized in that: The upper threshold is 20%, and the lower threshold is 2%.

Images