Real-time fault intelligent detection method for electric energy meter

Abstract

The invention relates to the technical field of data processing, in particular to an intelligent real-time fault detection method for an electric energy meter, which comprises the following steps of: obtaining a historical neighborhood radius when each historical data sequence is clustered according to data fluctuation of the historical data sequence of any monitoring index of the electric energy meter in each historical time period; according to the data rule similarity of any monitoring index of the transformer in the current time period and each historical time period, obtaining the neighborhood radius adaptation degree of the neighborhood radius when each historical neighborhood radius is used as the clustering of the real-time data sequence of any monitoring index of the electric energy meter in the current time period; according to each historical neighborhood radius and the neighborhood radius adaptation degree thereof, obtaining a self-adaptive neighborhood radius corresponding to the real-time data sequence, clustering the real-time data sequence according to the self-adaptive neighborhood radius to obtain a clustering result, and according to the clustering result under each monitoring index of the electric energy meter, detecting the fault of the electric energy meter. And the accuracy of detecting the fault of the electric energy meter is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a real-time fault intelligent detection method for an electric energy meter. BACKGROUND

[0002] The electric energy meter fault refers to an abnormal phenomenon that the monitoring data of the electric energy meter jumps, is too high or too low, or fluctuates sharply due to aging, damage, software anomaly and other factors, which does not conform to the normal change rule of the data. The health status of the electric energy meter is the key to ensure the measurement accuracy, improve the operation efficiency of the power system and enhance the user experience satisfaction. With the development of technology, the fault detection of the electric energy meter is gradually intelligentized, which can realize remote real-time monitoring. Compared with manual inspection, the detection efficiency can be improved, and the maintenance cost of the electric energy meter can be effectively reduced.

[0003] Because the common abnormality of the electric energy meter fault is the sudden change of the monitoring data, which is obviously different from the original data, the DBSCAN clustering algorithm is usually used for identification. However, the electric energy meter fault may also show data drift, sharp jitter and fixed invariance. There may be more abnormal data in the monitoring period, and the key factor affecting the abnormal identification effect of the DBSCAN clustering algorithm is the selection of the neighborhood radius. When the proportion of abnormal data is high, the traditional DBSCAN clustering algorithm cannot well identify the abnormal problem, resulting in poor identification effect of these persistent abnormalities and affecting the accuracy of the detection result of the electric energy meter fault.

[0004] Therefore, how to improve the accuracy of the detection of the electric energy meter fault becomes a problem to be solved. SUMMARY

[0005] Therefore, the present application provides a real-time fault intelligent detection method for an electric energy meter to solve the problem of how to improve the accuracy of the detection of the electric energy meter fault.

[0006] The present application provides a real-time fault intelligent detection method for an electric energy meter, which comprises the following steps: Obtaining a historical data sequence of any monitoring index of the electric energy meter in a preset number of historical periods, and obtaining a historical neighborhood radius when clustering each historical data sequence according to the data fluctuation in each historical data sequence; Obtaining a transformer real-time sequence of the any monitoring index under the transformer in a current period and a transformer historical sequence in each historical period, and obtaining an electricity consumption similarity index of each historical period and the current period according to the data rule similarity between each transformer historical sequence and the transformer real-time sequence; obtaining a real-time data sequence of the any monitoring index of the electric energy meter in a current period, obtaining a neighborhood radius adaptation degree of each historical neighborhood radius as a neighborhood radius when clustering the real-time data sequence according to a power consumption condition similarity index of each historical period and the current period and an abnormal performance of each historical data sequence; obtaining an adaptive neighborhood radius when clustering the real-time data sequence according to the adaptive neighborhood radius according to each historical neighborhood radius and the corresponding neighborhood radius adaptation degree, clustering the real-time data sequence according to the adaptive neighborhood radius to obtain a clustering result, and detecting a fault of the electric energy meter in real time according to the clustering result of each monitoring index of the electric energy meter.

[0007] Preferably, the historical neighborhood radius when clustering each historical data sequence is obtained according to data fluctuation in each historical data sequence, and the historical neighborhood radius is obtained by the following steps. For any historical data sequence, if a time interval between any data in the any historical data sequence and a first data in the any historical data sequence is greater than or equal to a preset time interval, and a time interval between the any data and a last data in the any historical data sequence is greater than or equal to the preset time interval, the any data is taken as analysis data, a window of a preset length is constructed in the any historical data sequence with the analysis data as a center, and the preset time interval is one half of a difference between the preset length and a constant 1. The absolute value of a difference between each two adjacent data in the window is calculated, and the fluctuation degree of the analysis data is obtained by linear normalization of an accumulated value of all the absolute values. The fluctuation degree of each analysis data in the any historical data sequence is obtained, and the historical neighborhood radius when clustering the any historical data sequence is obtained by using a K-distance graph method according to the fluctuation degree of each analysis data in the any historical data sequence.

[0008] Preferably, the power consumption condition similarity index of each historical period and the current period is obtained according to a data rule similarity between each transformer historical sequence and a transformer real-time sequence, and the power consumption condition similarity index is obtained by the following steps. For any transformer historical sequence, a trend item of the any transformer historical sequence and a transformer real-time sequence is obtained by using an STL time sequence decomposition algorithm, and a change trend similarity between the any transformer historical sequence and the transformer real-time sequence is obtained according to a similarity of the trend items of the any transformer historical sequence and the transformer real-time sequence. obtain the stability degree similarity between the any transformer historical sequence and the transformer real-time sequence according to the difference between the residual terms of the any transformer historical sequence and the transformer real-time sequence. obtain the power consumption situation similarity index of the historical period and the current period where the any transformer historical sequence is located according to the change trend similarity and the stability degree similarity.

[0009] Preferably, the change trend similarity between the any transformer historical sequence and the transformer real-time sequence is obtained according to the similarity of the trend terms of the any transformer historical sequence and the transformer real-time sequence, and the change trend similarity between the any transformer historical sequence and the transformer real-time sequence comprises: the trend term of the any transformer historical sequence is recorded as a historical trend term, and the trend term of the transformer real-time sequence is recorded as a real-time trend term; a preset representative number of historical trend values in the historical trend term are uniformly selected and recorded as historical representative trend values; for any historical representative trend value, the position of the any historical representative trend value in the historical trend term is obtained, the real-time trend value corresponding to the position in the real-time trend term is obtained, the absolute value of the difference between the any historical representative trend value and the real-time trend value is calculated to obtain the change trend difference value between the any historical representative trend value and the real-time trend value; the change trend difference value corresponding to each historical representative trend value in the any transformer historical sequence is calculated, all change trend difference values are accumulated to obtain an accumulated value, and the reciprocal of the sum of a preset constant and the accumulated value is calculated to obtain the change trend similarity between the any transformer historical sequence and the transformer real-time sequence.

[0010] Preferably, the stability degree similarity between the any transformer historical sequence and the transformer real-time sequence is obtained according to the difference between the residual terms of the any transformer historical sequence and the transformer real-time sequence, and the stability degree similarity between the any transformer historical sequence and the transformer real-time sequence comprises: the average value of all residual values in the residual term of the any transformer historical sequence is calculated and recorded as a historical residual average value, and the average value of all residual values in the residual term of the transformer real-time sequence is calculated and recorded as a real-time residual average value; the absolute value of the difference between the historical residual average value and the real-time residual average value is calculated to obtain a residual consistency index, and the reciprocal of the sum of a preset constant and the residual consistency index is calculated to obtain the stability degree similarity between the any transformer historical sequence and the transformer real-time sequence.

[0011] Preferably, the power consumption situation similarity index of the historical period and the current period where the any transformer historical sequence is located is obtained according to the change trend similarity and the stability degree similarity, and the power consumption situation similarity index of the historical period and the current period where the any transformer historical sequence is located comprises: calculating a product between the change trend similarity and the stability degree similarity to obtain a power consumption similarity index between a historical period where the any transformer historical sequence is located and a current period.

[0012] Preferably, the power consumption similarity index between each historical period and the current period, and the abnormal performance of each historical data sequence itself, are used to obtain a neighborhood radius adaptation degree of each historical neighborhood radius as a neighborhood radius when clustering the real-time data sequence, including: For any historical data sequence, a standard deviation of all data in the any historical data sequence is calculated, and a product of the power consumption similarity index corresponding to the historical period where the any historical data sequence is located and an inverse of the standard deviation is linearly normalized to obtain a neighborhood radius adaptation degree of a historical neighborhood radius corresponding to the any historical data sequence as a neighborhood radius when clustering the real-time data sequence.

[0013] Preferably, the adaptive neighborhood radius when clustering the real-time data sequence is obtained according to each historical neighborhood radius and the neighborhood radius adaptation degree corresponding to the historical neighborhood radius, including: For any historical neighborhood radius, a proportion of the neighborhood radius adaptation degree corresponding to the any historical neighborhood radius in all neighborhood radius adaptation degrees is calculated to obtain an adaptation weight of the any historical neighborhood radius; The adaptation weights of all historical neighborhood radii are obtained, and a weighted average value of all historical neighborhood radii is calculated according to the adaptation weights of all historical neighborhood radii to obtain the adaptive neighborhood radius when clustering the real-time data sequence.

[0014] Preferably, the clustering result is obtained by clustering the real-time data sequence according to the adaptive neighborhood radius, including: For any real-time data in the real-time data sequence, if a time interval between the any real-time data and a first data in the real-time data sequence is greater than or equal to a preset time interval, and a time interval between the any real-time data and a last data in the real-time data sequence is greater than or equal to the preset time interval, the any real-time data is taken as real-time to-be-analyzed data, a real-time window of a preset length is constructed in the real-time data sequence with the real-time to-be-analyzed data as a center, and the preset time interval is one half of a difference between the preset length and a constant 1. The absolute values of the differences between every two adjacent data in the real-time window are calculated, and an accumulated value of all the absolute values is linearly normalized to obtain a fluctuation degree of the real-time to-be-analyzed data. obtain fluctuation degrees of each of the real-time data to be analyzed in the real-time data sequence, and construct a scatter plot according to the fluctuation degrees of each of the real-time data to be analyzed, wherein the abscissa of the scatter plot represents the real-time data to be analyzed, and the ordinate represents the fluctuation degrees; cluster the data points in the scatter plot according to the adaptive neighborhood radius corresponding to the real-time data sequence, and obtain a clustering result.

[0015] Preferably, the real-time detection of the faults of the electric energy meter according to the clustering results under each of the monitoring indexes of the electric energy meter comprises: If there is an isolated point in the clustering result of any monitoring index, the electric energy meter is given a fault warning.

[0016] Compared with the prior art, the embodiment of the present application has the following beneficial effects: The present application obtains a historical data sequence of any monitoring index of an electric energy meter in a preset number of historical periods, respectively obtains a historical neighborhood radius when clustering each historical data sequence according to the data fluctuation in each historical data sequence, obtains a transformer real-time sequence of the any monitoring index of the transformer in a current period and a transformer historical sequence in each historical period, obtains an electricity consumption similarity index of each historical period and the current period according to the data regularity similarity between each transformer historical sequence and the transformer real-time sequence, obtains a neighborhood radius adaptation degree of taking each historical neighborhood radius as a neighborhood radius when clustering a real-time data sequence according to the electricity consumption similarity index of each historical period and the current period and the abnormal performance of each historical data sequence, obtains an adaptive neighborhood radius when clustering the real-time data sequence according to each historical neighborhood radius and the corresponding neighborhood radius adaptation degree, clusters the real-time data sequence according to the adaptive neighborhood radius, obtains a clustering result, and detects the faults of the electric energy meter in real time according to the clustering result under each monitoring index of the electric energy meter. According to the neighborhood radius when clustering the monitoring data of the electric energy meter in the historical period and the data change regularity similarity of the monitoring data of the transformer in the historical period and the current period, the adaptive neighborhood radius when clustering the monitoring data of the electric energy meter in the current period is obtained, and then the monitoring data of the electric energy meter in the current period is clustered according to the adaptive neighborhood radius, so as to effectively reduce the problem that the neighborhood radius directly obtained is not good at identifying abnormalities due to the great influence of the data in the current period on the electric energy meter faults, and improve the accuracy of identifying the electric energy meter faults by using the DBSCAN clustering algorithm. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.

[0018] Figure 1 is a method flow chart of a real-time fault intelligent detection method of an electric energy meter provided by the embodiment one of the present application. DETAILED DESCRIPTION

[0019] The embodiments of the present disclosure will be described in detail below, and examples of the embodiments are shown in the drawings. The embodiments described below by referring to the drawings are exemplary and are intended to explain the present disclosure, and cannot be understood as a limitation of the present disclosure.

[0020] It should be noted that the terms "first", "second" and the like in the specification of the present disclosure and the above drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The implementation described in the following exemplary embodiments does not represent all the implementations consistent with the present disclosure. Rather, they are only examples of devices and methods consistent with some aspects of the present disclosure.

[0021] In order to illustrate the technical solutions of the present application, the following will be illustrated by specific embodiments.

[0022] Referring to Figure 1 , is a method flow chart of a real-time fault intelligent detection method of an electric energy meter provided by the embodiment one of the present application, as Figure 1 shown, the method can include: Step S101, obtaining a historical data sequence of any monitoring index of the electric energy meter in a preset number of historical periods, and obtaining a historical neighborhood radius when clustering each historical data sequence according to the data fluctuation in each historical data sequence.

[0023] The common abnormal performance of the power meter fault is the monitoring data mutation. In the traditional way, the clustering algorithm such as DBSCAN is used for identification. The power meter fault may also show the data drift, severe jitter, fixed and other persistent abnormalities. There may be more abnormal data in the monitoring period, and the key factor affecting the abnormal identification effect of the DBSCAN clustering algorithm is the selection of the neighborhood radius. The neighborhood radius should try to cluster similar normal data into the same cluster, and also ensure that the abnormal data can be separated. When clustering the data with a high proportion of abnormal data, the neighborhood radius determined according to the current data segment itself is greatly affected by the abnormal data, resulting in low accuracy of the identification result of the power meter abnormality.

[0024] Considering that the power meter is used to monitor the power consumption of the user, if the power consumption of the user in the historical period is similar to that in the current period, and the power meter has a fault in the current period, the neighborhood radius for DBSCAN clustering of the historical data segment of the power meter can be used as the neighborhood radius for DBSCAN clustering of the current data segment of the power meter, which can effectively reduce the influence of the overall abnormality of the current data segment on the clustering result and improve the accuracy of the power meter fault detection. Therefore, in the embodiment of the present application, the neighborhood radius for DBSCAN clustering of the historical data segment of the power meter is adjusted according to the similarity of the power consumption between the historical period and the current period of the user, and the neighborhood radius applicable to DBSCAN clustering of the current data segment of the power meter is obtained.

[0025] Firstly, the monitoring indicators for power meter fault detection are set as current, voltage and power, and the current, voltage and power monitored by the power meter are recorded every 1s. At the same time, the time period length is set as 5min, that is, the fault condition of the power meter is evaluated every 5min. Here, the monitoring indicators, recording frequency and time period length are not limited, and the implementer can set them according to the specific scene.

[0026] Then, the data of each monitoring indicator of the power meter in the historical period is obtained. Since the value range and fluctuation of the monitoring data of the power meter should be basically consistent under the condition that there is no circuit adjustment or change in power demand, only the data in the last 2 hours before the current period is analyzed. Here, the last 2 hours before the current period is not limited, and the implementer can set it according to the specific scene, and each time period (the time period length is 5min) in the last 2 hours before the current period is recorded as a historical period, that is, the neighborhood radius for DBSCAN clustering of each historical period is adjusted to obtain the neighborhood radius for DBSCAN clustering of the data in the current period. Taking current as an example, the current value monitored by the power meter at each time in the last 2 hours before the current period is linearly normalized to obtain historical current data. Here, linear normalization is a prior art, which will not be described again.

[0027] Finally, the historical current data in each historical period is grouped into a historical data sequence, and the neighborhood radius when the historical data sequence of the current of the electric energy meter in each historical period is DBSCAN clustered is obtained. Considering that the abnormal data generated by the electric energy meter failure may also appear in the case of severe data fluctuation in addition to the case of too high or too low value, in the embodiment of the present application, the monitoring data of the electric energy meter is DBSCAN clustered according to the fluctuation of the monitoring data of the electric energy meter, so as to more accurately detect the failure of the electric energy meter. Therefore, the neighborhood radius when the data of the current of the electric energy meter in each historical period is DBSCAN clustered also needs to be obtained according to the fluctuation of the historical data sequence of the current of the electric energy meter in each historical period, so as to adjust the neighborhood radius when the current data of the electric energy meter in the current period is DBSCAN clustered. Taking the mth historical period as an example, that is, taking the mth historical data sequence (the historical current data of the electric energy meter in the mth historical period) as an example, the steps of obtaining the neighborhood radius when the mth historical data sequence is DBSCAN clustered are as follows: (1) Obtain the data to be analyzed in the mth historical data sequence, and obtain the fluctuation degree of each data to be analyzed according to the fluctuation in the neighborhood range of each data to be analyzed.

[0028] The time interval is set to 5s, which is not limited here, and the implementer can set it according to the specific scene. Taking each data in the mth historical data sequence as the center, the data fluctuation within 5s at the time of each data is analyzed. Since the monitoring data of the electric energy meter is recorded once every 1s in the embodiment of the present application, a window with a length of 5 is constructed with each data in the mth historical data sequence as the center, and the data in the window of each data is the data within 5s. Because the number of data before the first two data in the mth historical data sequence and the number of data of the last two data are less than 2, that is, the window cannot be constructed, if the time interval between any data in the mth historical data sequence and the first data in the mth historical data sequence is greater than or equal to 2, and the time interval between the any data and the last data in the mth historical data sequence is greater than or equal to 2, the any data is taken as the data to be analyzed, and a window with a length of 5 is constructed with the data to be analyzed as the center, so as to obtain the fluctuation degree of the data to be analyzed according to the data fluctuation in the window of the data to be analyzed.

[0029] Taking the ith data to be analyzed in the mth historical data sequence as an example, the specific way of obtaining the fluctuation degree of the ith data to be analyzed is as follows: The absolute value of the difference between each adjacent two data in the window of the ith data to be analyzed is calculated, and the cumulative value of all difference absolute values is linearly normalized to obtain the fluctuation degree of the ith data to be analyzed.

[0030] In an embodiment, the formula for calculating the fluctuation degree of the i-th data to be analyzed is: wherein, represents the fluctuation degree of the i-th data to be analyzed, represents the number of all data within the window of the i-th data to be analyzed (in the embodiment of the present application, ), represents the j+1-th data within the window of the i-th data to be analyzed, represents the j-th data within the window of the i-th data to be analyzed, represents a linear normalization function, represents an absolute value symbol.

[0031] It should be noted that, the greater the current fluctuation is, and thus the greater the fluctuation degree of the i-th data to be analyzed is.

[0032] Similarly, the fluctuation degree of each data to be analyzed in the m-th historical data sequence is obtained.

[0033] (2) According to the fluctuation degree of each data to be analyzed in the m-th historical data sequence, the neighborhood radius when performing DBSCAN clustering on the data in the m-th historical data sequence is obtained, denoted as historical neighborhood radius.

[0034] The K-distance graph method is a commonly used method for determining the neighborhood radius in the DBSCAN clustering algorithm. According to the fluctuation degree of each data to be analyzed in the m-th historical data sequence, the neighborhood radius when clustering the m-th historical data sequence is obtained using the K-distance graph method. The general process is as follows: (a) construct a coordinate graph according to each data to be analyzed in the m-th historical data sequence and its fluctuation degree, with the horizontal coordinate being the fluctuation degree and the vertical coordinate being the data to be analyzed; (b) calculate the distance from each data point in the coordinate graph to its k-th nearest neighbor, with k set to 4, which is not limited here and can be set according to the specific scene by the implementer; (c) sort all distances in descending order to construct a line graph, and take the distance corresponding to the first valley point in the line graph as the neighborhood radius when clustering the m-th historical data sequence, denoted as historical neighborhood radius, represented by . The K-distance graph method is a prior art and will not be described here.

[0035] Similarly, the neighborhood radius when performing DBSCAN clustering on each historical data sequence is obtained, that is, the historical neighborhood radius when performing DBSCAN clustering on the historical current data of the electric energy meter in each historical period is obtained.

[0036] At this point, the historical neighborhood radius when the current data of the electric energy meter in each historical period is DBSCAN clustered is obtained, so as to obtain the neighborhood radius when the current data of the electric energy meter in the previous period is DBSCAN clustered by adjusting each historical neighborhood radius.

[0037] In step S102, the transformer real-time sequence of the any monitoring index under the transformer in the current period and the transformer historical sequence in each historical period are obtained, and the power consumption similarity index of each historical period and the current period is obtained according to the data rule similarity between each transformer historical sequence and the transformer real-time sequence.

[0038] In the embodiment of the application, the neighborhood radius suitable for the current period is calculated according to the historical neighborhood radius corresponding to the historical period, and the neighborhood radius corresponding to the historical period is obtained based on the fluctuation degree of the historical monitoring data. Since the fluctuation performance of different periods may be different, for example, the current, voltage and power value will have a relatively obvious difference compared with other periods during the power consumption peak period, the more similar the power consumption of the user in the historical period and the current period is, the better the adaptability of the historical neighborhood radius corresponding to the historical period as the neighborhood radius corresponding to the current period is.

[0039] Since the fault condition of the electric energy meter in the current period is unknown, the similarity of the monitoring data of the electric energy meter between the historical period and the current period cannot be used to judge the similarity of the power consumption of the user in the historical period and the current period. Considering that the electric energy meter failure only affects the data accuracy of the user monitored by the electric energy meter, but does not affect the data of other nodes, and the transformer node connected with the user node is the previous node of the electric energy meter, the monitoring data at the transformer will not be affected by the failure of the electric energy meter, so in the embodiment of the application, a resistance voltage divider, a current transformer and a metering chip are respectively installed at the transformer, which are used to monitor the voltage, current and power at the transformer position, and the voltage value, current value and power value at the transformer are recorded once every 1s according to the recording frequency of the monitoring data of the electric energy meter.

[0040] Since step S101 takes current as an example, the historical neighborhood radius corresponding to the historical data sequence of the current of the electric energy meter in each historical period is obtained, so still taking current as an example, the current value monitored at the transformer in the current period and 2 hours before the current period is linearly normalized to obtain the transformer historical sequence of the current monitored at the transformer in each historical period and the transformer real-time sequence of the current monitored at the transformer in the current period.

[0041] Further, according to the similarity between each transformer historical sequence and the transformer real-time sequence, a user's electricity consumption similarity index about current in each historical period and the current period is obtained, which is used to judge the electricity consumption similarity of the user in the historical period and the current period. Taking the mth historical period as an example, that is, taking the mth transformer historical sequence as an example, the electricity consumption similarity index of the current in the mth historical period and the current period at the transformer is obtained as follows: (1) By comparing the data change trend between the mth transformer historical sequence (the current data at the transformer in the mth historical period) and the transformer real-time sequence (the current data at the transformer in the current period), the change trend similarity between the mth transformer historical sequence and the transformer real-time sequence is obtained.

[0042] Specifically, since the current data fluctuates to a certain extent under normal circumstances, the current data in different periods cannot be analyzed for similarity by one-to-one comparison, but the overall change trend of two segments of current data under the same change law is the same, so the STL time series decomposition algorithm is used to obtain the trend item of the mth transformer historical sequence and the transformer real-time sequence respectively. If the data change trends of the historical period and the current period are similar, the corresponding trend items should also be consistent. The STL time series decomposition algorithm is a prior art and will not be described here. The trend item of the mth transformer historical sequence is denoted as a historical trend item, and the trend item of the transformer real-time sequence is denoted as a real-time trend item. In the historical trend item, a preset representative number of historical trend values are uniformly selected and denoted as historical representative trend values. The preset representative number is set to 30, which is not limited here, and the implementer can set it according to the specific scene. For any historical representative trend value, the position of the any historical representative trend value in the historical trend item is obtained, the same real-time trend value as the position in the real-time trend item is obtained, and the absolute value of the difference between the any historical representative trend value and the real-time trend value is calculated to obtain the change trend difference value between the any historical representative trend value and the real-time trend value. The change trend difference value corresponding to each historical representative trend value in the mth transformer historical sequence is calculated, all change trend difference values are accumulated to obtain an accumulated value, and the reciprocal of the sum of the preset constant and the accumulated value is calculated to obtain the change trend similarity between the mth transformer historical sequence and the transformer real-time sequence.

[0043] In an embodiment, the calculation formula of the change trend similarity between the mth transformer historical sequence and the transformer real-time sequence is as follows: Wherein, represents the change trend similarity between the mth transformer historical sequence and the transformer real-time sequence, represents a preset representative quantity, that is, the number of all historical representative trend values, represents the kth historical representative trend value in the trend item (historical trend item) of the mth transformer historical sequence, represents the real-time trend value in the trend item (real-time trend item) of the transformer real-time sequence, which is located at the same position as the kth historical representative trend value, represents a preset constant, which is used to ensure that the fraction is meaningful, and is set to Here, no limitation is made, and the implementer can set it according to the specific scene, represents the absolute value symbol.

[0044] It should be noted that, represents the difference between the two trend values at the same position in the trend item corresponding to the mth historical period and the current period, The smaller the difference is, the more consistent the current change trend of the transformer is with the current change trend of the transformer in the mth historical period, and thus The greater the difference is, the more similar the change trend between the mth transformer historical sequence and the transformer real-time sequence is.

[0045] (2) By comparing the fluctuation between the mth transformer historical sequence and the transformer real-time sequence, the stability similarity between the mth transformer historical sequence and the transformer real-time sequence is obtained.

[0046] Because the current data has a certain volatility, and the volatility degree also reflects the stability degree of the current power consumption, if there is a large difference between the power consumption stability degrees of the mth historical period and the current period, it is possible that the user frequently adjusts the power consumption equipment in one of the periods, and the power consumption equipment is in continuous operation in the other period, at this time, the consistency of the user's power consumption in the mth historical period and the current period is low.

[0047] Therefore, in the embodiment of the present application, the STL time sequence decomposition algorithm is used to obtain the residual items of the mth transformer historical sequence and the transformer real-time sequence respectively, the fluctuation between the mth transformer historical sequence and the transformer real-time sequence is reflected through the residual items, and then the stability similarity between the mth transformer historical sequence and the transformer real-time sequence is obtained, which is used to reflect the consistency of the user's power consumption in the mth historical period and the current period.

[0048] Specifically, the average value of all residual values in the residual item of the mth transformer historical sequence is calculated, which is denoted as historical residual average value, and the average value of all residual values in the residual item of the transformer real-time sequence is calculated, which is denoted as real-time residual average value; calculating an absolute value of a difference between the historical residual mean and the real-time residual mean to obtain a residual consistency index, and calculating an inverse of a sum of a preset constant and the residual consistency index to obtain the stability degree similarity between the mth transformer historical sequence and the transformer real-time sequence.

[0049] In an embodiment, a calculation formula of the stability degree similarity between the mth transformer historical sequence and the transformer real-time sequence is as follows: wherein, the stability degree similarity between the mth transformer historical sequence and the transformer real-time sequence is represented by, the historical residual mean is represented by an average value of all residual values in a residual term of the mth transformer historical sequence, the real-time residual mean is represented by an average value of all residual values in a residual term of the transformer real-time sequence, the preset constant is represented by, and is set to Here, no limitation is made, and an implementer can set it according to a specific scene, the absolute value symbol is represented by.

[0050] It should be noted that, The smaller the value is, the more consistent the data stability degree between the mth historical period and the current period is, and further The larger the value is, the more similar the stability degree between the mth transformer historical sequence and the transformer real-time sequence is.

[0051] (3) According to the change trend similarity and the stability degree similarity between the mth transformer historical sequence and the transformer real-time sequence, an electricity consumption situation similarity index of the user about current in the mth historical period and the current period is obtained.

[0052] Specifically, a product of the change trend similarity and the stability degree similarity is calculated to obtain the electricity consumption situation similarity index of the historical period (i.e., the mth historical period) in which the mth transformer historical sequence is located and the current period about current.

[0053] In an embodiment, a calculation formula of the electricity consumption situation similarity index of the mth historical period and the current period about current is as follows: wherein, the electricity consumption situation similarity index of the mth historical period and the current period about current is represented by, the change trend similarity between the mth transformer historical sequence (current data of the transformer in the mth historical period) and the transformer real-time sequence is represented by, The stability similarity between the mth transformer historical sequence and the transformer real-time sequence is represented.

[0054] It should be noted that, The greater the value is, the more consistent the current change trend at the transformer in the mth historical period and the current period is, and thus The greater the value is, the more similar the user's power consumption about the current in the mth historical period and the current period is, and the better the adaptability of the historical neighborhood radius corresponding to the current data of the electric energy meter in the mth historical period (the mth historical data sequence) to the neighborhood radius corresponding to the current data of the electric energy meter in the current period is. The greater the value is, the more consistent the data stability of the current at the transformer in the mth historical period and the current period is, and thus The greater the value is, the more similar the user's power consumption about the current in the mth historical period and the current period is, and the better the adaptability of the historical neighborhood radius corresponding to the current data of the electric energy meter in the mth historical period (the mth historical data sequence) to the neighborhood radius corresponding to the current data of the electric energy meter in the current period is.

[0055] At this point, the similarity index of the user's power consumption about the current in the mth historical period and the current period is obtained. Similarly, the similarity index of the user's power consumption about the current in each historical period and the current period is obtained, so as to adjust the historical neighborhood radius corresponding to each historical data sequence (the historical current data of the electric energy meter in each historical period) according to the similarity index of the user's power consumption about the current in each historical period and the current period, to obtain the neighborhood radius when the current data of the electric energy meter in the current period is clustered by DBSCAN.

[0056] In step S103, the real-time data sequence of the any monitoring index of the electric energy meter in the current period is obtained, and the neighborhood radius adaptability of taking each historical neighborhood radius as the neighborhood radius when clustering the real-time data sequence is obtained according to the similarity index of the power consumption in each historical period and the current period and the abnormal performance of each historical data sequence.

[0057] According to the acquisition method of the historical current data of the current of the electric energy meter in each historical period, the real-time current data of the current of the electric energy meter in the current period is obtained to obtain the real-time data sequence. After obtaining the similarity index of the power consumption about the current in each historical period and the current period by step S102, the neighborhood radius adaptability of taking the historical neighborhood radius corresponding to each historical data sequence (the historical current data of the electric energy meter in each historical period) as the neighborhood radius when clustering the real-time data sequence (the real-time current data of the electric energy meter in the current period) by DBSCAN is obtained according to the similarity index of the power consumption about the current in each historical period and the current period.

[0058] Considering that if the data in the historical period has a large degree of abnormality, the accuracy of the historical neighborhood radius corresponding to the historical period obtained by step S101 will also be reduced, in order to reduce the interference of abnormal historical periods, the influence of historical periods with large abnormalities on the calculation results should be reduced. Therefore, in an embodiment of the present invention, the standard deviation of all data in each historical data sequence is calculated to characterize the degree of abnormality of the historical current data in each historical period, and then based on the similarity index of the current consumption of each historical period and the current period, and the standard deviation of each historical data sequence, the neighborhood radius adaptability of using the neighborhood radius corresponding to each historical data sequence as the neighborhood radius when performing DBSCAN clustering on the real-time data sequence is obtained.

[0059] Taking the mth historical period as an example, that is, taking the mth historical data sequence as an example, the specific method of obtaining the neighborhood radius fitness of using the historical neighborhood radius corresponding to the mth historical data sequence as the neighborhood radius for performing DBSCAN clustering on the real-time data sequence is: The product of the electricity consumption similarity index corresponding to the historical period (mth historical period) of the mth historical data sequence (i.e., the electricity consumption similarity index of the mth historical period and the current period regarding current) and the inverse of the standard deviation of the mth historical data sequence is linearly normalized to obtain the neighborhood radius adaptability of using the historical neighborhood radius corresponding to the mth historical data sequence as the neighborhood radius when performing DBSCAN clustering on the real-time data sequence.

[0060] In one embodiment, the calculation formula for the neighborhood radius fitness of the real-time data sequence using the historical neighborhood radius corresponding to the mth historical data sequence as the neighborhood radius when performing DBSCAN clustering on the real-time data sequence is: in, Indicates the neighborhood radius adaptability of using the historical neighborhood radius corresponding to the mth historical data sequence as the neighborhood radius for clustering the real-time data sequence. It represents the similarity index of the electricity consumption between the mth historical period and the current period. Represents the standard deviation of all data in the mth historical data sequence, represents the linear normalization function.

[0061] It should be noted that The larger the value is, the more similar the user's electricity consumption in the mth historical period is to the current period. The larger it is, the better the adaptability of using the historical neighborhood radius corresponding to the mth historical data sequence as the neighborhood radius for DBSCAN clustering of the real-time data sequence; The smaller, the smaller the abnormality of the mth historical data sequence, and further The larger, the better adaptability of the historical neighborhood radius corresponding to the mth historical data sequence as the neighborhood radius when DBSCAN clustering is performed on the real-time data sequence.

[0062] Similarly, the neighborhood radius adaptability of taking the historical neighborhood radius corresponding to each historical data sequence as the neighborhood radius when DBSCAN clustering is performed on the real-time data sequence is obtained, In step S104, the adaptive neighborhood radius when clustering the real-time data sequence is obtained according to each historical neighborhood radius and the neighborhood radius adaptability corresponding thereto, the real-time data sequence is clustered according to the adaptive neighborhood radius, and the clustering result is obtained. According to the clustering result of each monitoring index of the electric energy meter, the fault of the electric energy meter is detected in real time.

[0063] After obtaining the neighborhood radius adaptability of taking the historical neighborhood radius corresponding to each historical data sequence as the neighborhood radius when DBSCAN clustering is performed on the real-time data sequence through step S103, the adaptive neighborhood radius when DBSCAN clustering is performed on the real-time data sequence is obtained according to the historical neighborhood radius corresponding to each historical data sequence and the neighborhood radius adaptability thereof. Specifically: For any historical neighborhood radius, the proportion of the neighborhood radius adaptability corresponding to the any historical neighborhood radius in all neighborhood radius adaptabilities is calculated to obtain the adaptive weight of the any historical neighborhood radius; The adaptive weights of all historical neighborhood radii are obtained, and the weighted average value of all historical neighborhood radii is calculated according to the adaptive weights of all historical neighborhood radii to obtain the adaptive neighborhood radius when DBSCAN clustering is performed on the real-time data sequence.

[0064] In an embodiment, the calculation formula of the adaptive neighborhood radius when DBSCAN clustering is performed on the real-time data sequence is: Wherein, E represents the adaptive neighborhood radius when DBSCAN clustering is performed on the real-time data sequence, N represents the number of all historical time periods, that is, the number of all historical data sequences, E m represents the historical neighborhood radius corresponding to the mth historical data sequence, E m represents the neighborhood radius adaptability of taking the historical neighborhood radius corresponding to the mth historical data sequence as the neighborhood radius when DBSCAN clustering is performed on the real-time data sequence, E q represents the neighborhood radius adaptability of taking the historical neighborhood radius corresponding to the qth historical data sequence as the neighborhood radius when DBSCAN clustering is performed on the real-time data sequence.

[0065] It should be noted that, The larger, the better the adaptability of the historical neighborhood radius corresponding to the mth historical data sequence as the neighborhood radius when DBSCAN clustering is performed on the real-time data sequence, and the adaptive neighborhood radius when DBSCAN clustering is performed on the real-time data sequence The neighborhood radius when DBSCAN clustering is performed on the historical data sequence The closer.

[0066] After obtaining the adaptive neighborhood radius when DBSCAN clustering is performed on the real-time data sequence, the fluctuation degree of each real-time data to be analyzed in the real-time data sequence is obtained in the manner of obtaining the fluctuation degree of the data to be analyzed in each historical data sequence in step S101, so as to cluster the real-time data to be analyzed in the real-time data sequence according to the fluctuation degree of each real-time data to be analyzed in the real-time data sequence and the adaptive neighborhood radius of the real-time data sequence, and obtain the clustering result of the current period of the current of the electric energy meter by using the DBSCAN clustering algorithm. The specific manner of obtaining the fluctuation degree of each real-time data to be analyzed in the real-time data sequence is: For any real-time data in the real-time data sequence, if the time interval between the any real-time data and the first data in the real-time data sequence is greater than or equal to 2, and the time interval between the any real-time data and the last data in the real-time data sequence is greater than or equal to 2, the any real-time data is taken as a real-time data to be analyzed, and a real-time window with a length of 5 is constructed in the real-time data sequence with the real-time data to be analyzed as the center; The absolute value of the difference between each adjacent two data in the real-time window is calculated, and the cumulative value of all difference absolute values is linearly normalized to obtain the fluctuation degree of the real-time data to be analyzed.

[0067] Similarly, the fluctuation degree of each real-time data to be analyzed in the real-time data sequence is obtained, and further, a scatter plot is constructed according to the fluctuation degree of each real-time data to be analyzed, the horizontal coordinate of the scatter plot represents the real-time data to be analyzed, and the vertical coordinate represents the fluctuation degree; the data points in the scatter plot are DBSCAN clustered according to the adaptive neighborhood radius corresponding to the real-time data sequence, and the clustering result of the current period of the current of the electric energy meter is obtained.

[0068] Similarly, the adaptive neighborhood radius when DBSCAN clustering is performed on the voltage of the electric energy meter in the current period is obtained, denoted as , and the adaptive neighborhood radius when DBSCAN clustering is performed on the power of the electric energy meter in the current period is obtained, denoted as Further, by using the DBSCAN clustering algorithm, the monitoring data of each monitoring index in the current period is DBSCAN clustered according to the adaptive neighborhood radius corresponding to each monitoring index in the current period, to obtain the clustering result of each monitoring index of the electric energy meter in the current period. The DBSCAN clustering is prior art, which will not be described here. If there is an isolated point in the clustering result of any monitoring index in the current period, the electric energy meter is fault warned to inform the relevant technical personnel to troubleshoot. Specifically, since the power is obtained according to the current and voltage, if the clustering result of the power in the current period is normal (i.e. there is no isolated point), and the clustering result of the current or voltage in the current period is abnormal (i.e. there is an isolated point), it may be that there is interference or other abnormalities in the data transmission process of the electric energy meter. At this time, the electric energy meter should also be fault warned to inform the relevant technical personnel to troubleshoot the interference or other abnormalities in the data transmission process of the electric energy meter.

[0069] It is worth noting that the focus of the present application is to use the historical monitoring data of the electric energy meter to adaptively generate the neighborhood radius when DBSCAN clustering is performed on the monitoring data of the current period, so as to improve the accuracy of identifying the electric energy meter fault by using the DBSCAN algorithm. It is prior art to identify the fault type of the electric energy meter according to the clustering result, which will not be described here.

[0070] In summary, the application obtains a historical data sequence of any monitoring index of an electric energy meter in a preset number of historical periods, obtains a historical neighborhood radius when clustering each historical data sequence according to data fluctuation in each historical data sequence, obtains a transformer real-time sequence of the any monitoring index under the transformer in a current period and a transformer historical sequence in each historical period, obtains an electricity consumption similarity index of each historical period and the current period according to data rule similarity between each transformer historical sequence and the transformer real-time sequence, obtains a real-time data sequence of the any monitoring index of the electric energy meter in the current period, obtains a neighborhood radius adaptation degree of taking each historical neighborhood radius as a neighborhood radius when clustering the real-time data sequence according to the electricity consumption similarity index of each historical period and the current period and the abnormal performance of each historical data sequence, obtains an adaptive neighborhood radius when clustering the real-time data sequence according to each historical neighborhood radius and the corresponding neighborhood radius adaptation degree, clusters the real-time data sequence according to the adaptive neighborhood radius to obtain a clustering result, and detects a fault of the electric energy meter in real time according to the clustering result of each monitoring index of the electric energy meter. According to the neighborhood radius when clustering the monitoring data of the electric energy meter in the historical period and the data change rule similarity of the monitoring data of the transformer in the historical period and the current period, the adaptive neighborhood radius when clustering the monitoring data of the electric energy meter in the current period is obtained, and then the monitoring data of the electric energy meter in the current period is clustered according to the adaptive neighborhood radius, so as to effectively reduce the problem that the neighborhood radius directly obtained is not good at identifying the abnormality due to the fact that the data in the current period is greatly affected by the fault of the electric energy meter, and improve the accuracy of identifying the fault of the electric energy meter by using the DBSCAN clustering algorithm.

[0071] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit it; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for intelligent detection of real-time faults of electric energy meters, characterized in that: The method for intelligently detecting real-time faults of electric energy meters comprises: Obtain a historical data sequence of any monitoring indicator of the electric energy meter within a preset number of historical time periods, and obtain a historical neighborhood radius for clustering each historical data sequence based on data fluctuations in each historical data sequence; Obtain the transformer real-time sequence in the current period and the transformer historical sequence in each historical period for any monitoring indicator under the transformer, and obtain the power consumption similarity index of each historical period and the current period based on the data regularity similarity between each transformer historical sequence and the transformer real-time sequence; Obtaining a real-time data sequence of any monitoring indicator of the electric energy meter in the current period, and obtaining a neighborhood radius adaptability of using each historical neighborhood radius as a neighborhood radius for clustering the real-time data sequence based on a similarity index of electricity usage between each historical period and the current period, and an abnormal performance of each historical data sequence; According to each historical neighborhood radius and its corresponding neighborhood radius fitness, the adaptive neighborhood radius for clustering the real-time data sequence is obtained. The real-time data sequence is clustered according to the adaptive neighborhood radius to obtain the clustering result. According to the clustering result under each monitoring indicator of the electric energy meter, the fault of the electric energy meter is detected in real time.

2. The method for real-time intelligent fault detection of electric energy meters according to claim 1, characterized in that: The historical neighborhood radius for clustering each historical data sequence is obtained based on the data fluctuations in each historical data sequence, including: For any historical data sequence, if the time interval between any data in the historical data sequence and the first data in the historical data sequence is greater than or equal to a preset time interval, and the time interval between any data and the last data in the historical data sequence is greater than or equal to a preset time interval, then the data is used as the data to be analyzed, and a window of preset length is constructed in the historical data sequence with the data to be analyzed as the center, where the preset time interval is half of the difference between the preset length and a constant of 1; Calculating the absolute value of the difference between every two adjacent data in the window, and linearly normalizing the accumulated value of all the absolute values ​​of the differences to obtain the degree of fluctuation of the data to be analyzed; Obtain the degree of fluctuation of each data to be analyzed in any historical data sequence, and use the K-distance graph method to obtain the historical neighborhood radius when clustering any historical data sequence based on the degree of fluctuation of each data to be analyzed in any historical data sequence.

3. The method for real-time intelligent fault detection of electric energy meters according to claim 1, characterized in that: The similarity index of power consumption between each historical period and the current period is obtained based on the data regularity similarity between each transformer historical sequence and the transformer real-time sequence, including: For any transformer history sequence, the STL time series decomposition algorithm is used to obtain trend items of the transformer history sequence and the transformer real-time sequence respectively, and according to the similarity of the trend items of the transformer history sequence and the transformer real-time sequence, the similarity of the change trend between the transformer history sequence and the transformer real-time sequence is obtained; Using the STL time series decomposition algorithm, respectively obtaining the residual terms of the any transformer historical sequence and the transformer real-time sequence, and obtaining the stability similarity between the any transformer historical sequence and the transformer real-time sequence according to the difference between the residual terms of the any transformer historical sequence and the transformer real-time sequence; According to the similarity of the change trend and the similarity of the stability, a similarity index of the power consumption between the historical period in which the historical sequence of any transformer is located and the current period is obtained.

4. The method for intelligent real-time fault detection of electric energy meters according to claim 3, characterized in that: The obtaining of the similarity of the change trend between any transformer historical sequence and the transformer real-time sequence according to the similarity of the trend items between any transformer historical sequence and the transformer real-time sequence includes: Recording the trend item of any transformer historical sequence as a historical trend item, recording the trend item of the transformer real-time sequence as a real-time trend item, and evenly selecting a preset number of representative historical trend values ​​from the historical trend items and recording them as historical representative trend values; For any historical representative trend value, obtaining a position of the historical representative trend value in the historical trend item, obtaining a real-time trend value having the same position in the real-time trend item, calculating an absolute value of a difference between the historical representative trend value and the real-time trend value, and obtaining a change trend difference value between the historical representative trend value and the real-time trend value; Calculate the change trend difference value corresponding to each historical representative trend value in any transformer historical sequence, accumulate all change trend difference values ​​to obtain an accumulated value, calculate the inverse of the sum of a preset constant and the accumulated value, and obtain the change trend similarity between any transformer historical sequence and the transformer real-time sequence.

5. The method for intelligent real-time fault detection of electric energy meters according to claim 3, characterized in that: The obtaining of the stability similarity between any transformer historical sequence and the transformer real-time sequence according to the difference in residual terms between any transformer historical sequence and the transformer real-time sequence includes: Calculate the average value of all residual values ​​in the residual items of any transformer historical sequence, record it as the historical residual mean, calculate the average value of all residual values ​​in the residual items of the transformer real-time sequence, record it as the real-time residual mean; The absolute value of the difference between the historical residual mean and the real-time residual mean is calculated to obtain a residual consistency index, and the reciprocal of the sum of a preset constant and the residual consistency index is calculated to obtain a stability similarity between any transformer historical sequence and the transformer real-time sequence.

6. The method for intelligent real-time fault detection of electric energy meters according to claim 3, characterized in that: The obtaining, based on the similarity of the change trend and the similarity of the stability, of a similarity index of power consumption between the historical period and the current period of the historical sequence of any transformer includes: The product of the change trend similarity and the stability similarity is calculated to obtain a similarity index of power consumption between the historical period and the current period of the historical sequence of any transformer.

7. The method for intelligent real-time fault detection of electric energy meters according to claim 1, characterized in that: The neighborhood radius adaptability of each historical neighborhood radius as the neighborhood radius for clustering the real-time data sequence is obtained based on the similarity index of electricity consumption between each historical period and the current period, and the abnormal performance of each historical data sequence itself, including: For any historical data sequence, the standard deviation of all data in the said historical data sequence is calculated, and the product of the electricity consumption similarity index corresponding to the historical period of the said historical data sequence and the inverse of the said standard deviation is linearly normalized to obtain the neighborhood radius adaptability of using the historical neighborhood radius corresponding to the said any historical data sequence as the neighborhood radius when clustering the real-time data sequence.

8. The method for intelligent real-time fault detection of electric energy meters according to claim 1, characterized in that: The adaptive neighborhood radius for clustering the real-time data sequence is obtained based on each historical neighborhood radius and its corresponding neighborhood radius adaptability, including: For any historical neighborhood radius, calculate the proportion of the neighborhood radius adaptation degree corresponding to the any historical neighborhood radius in all neighborhood radius adaptation degrees, and obtain the adaptation weight of the any historical neighborhood radius; The adaptation weights of all historical neighborhood radii are obtained, and the weighted average of all historical neighborhood radii is calculated according to the adaptation weights of all historical neighborhood radii to obtain the adaptive neighborhood radius when clustering the real-time data series.

9. The method for intelligent real-time fault detection of electric energy meters according to claim 1, characterized in that: The clustering of the real-time data sequence according to the adaptive neighborhood radius to obtain the clustering result includes: For any real-time data in a real-time data sequence, if the time interval between the any real-time data and the first data in the real-time data sequence is greater than or equal to a preset time interval, and the time interval between the any real-time data and the last data in the real-time data sequence is greater than or equal to a preset time interval, then the any real-time data is used as the real-time data to be analyzed, and in the real-time data sequence, a real-time window of a preset length is constructed with the real-time data to be analyzed as the center, where the preset time interval is half of the difference between the preset length and a constant of 1; Calculating the absolute value of the difference between every two adjacent data in the real-time window, and linearly normalizing the accumulated value of all the absolute values ​​of the differences to obtain the degree of fluctuation of the real-time data to be analyzed; Obtaining a fluctuation degree of each of the real-time data to be analyzed in the real-time data sequence, and constructing a scatter plot according to the fluctuation degree of each of the real-time data to be analyzed, wherein the abscissa of the scatter plot represents the real-time data to be analyzed, and the ordinate represents the fluctuation degree; The data points in the scatter plot are clustered according to the adaptive neighborhood radius corresponding to the real-time data sequence to obtain a clustering result.

10. The method for intelligent real-time fault detection of electric energy meters according to claim 1, characterized in that: The method of performing real-time fault detection on the electric energy meter based on the clustering result under each monitoring indicator of the electric energy meter includes: If there are isolated points in the clustering results of any monitoring indicator, a fault warning will be issued for the electricity meter.

Images