A method and system for processing electric energy metering data

By dividing the power metering data into multiple datasets and adaptively adjusting the nu parameter, the accuracy problem of power metering data detection in different time-series stages of the One-Class SVM algorithm is solved, achieving higher accuracy in anomaly data detection.

CN118569509BActive Publication Date: 2024-11-15JIANGYIN CHANGYI GRP CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411038635.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-11-15
Estimated Expiration
2044-07-31

AI Technical Summary

Technical Problem

In existing technologies, the fixed nu parameter of the One-Class SVM algorithm cannot adapt to power metering data in different time series stages, resulting in the problem of missed or false alarms of abnormal data in power metering data in some time series stages.

Method used

The electricity metering data is divided into several datasets. The nu parameter is adaptively obtained according to the degree of anomaly in each dataset. Anomaly detection is performed on the electricity metering data in each dataset. Correction is made by obtaining the degree of offset and stationarity. The nu parameter is adjusted to adapt to the data characteristics of different time series stages.

Benefits of technology

It improves the accuracy of abnormal data detection, avoids missed and false alarms, and ensures the accurate identification of abnormal data in power metering data at different time stages.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118569509B_ABST
    Figure CN118569509B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of data processing technology, and in particular to a method and system for processing electric energy metering data. The method comprises the steps of collecting electric energy metering data and obtaining each electric energy metering data set, obtaining the degree of deviation of each electric energy metering data set according to the distribution of the electric energy metering data in each electric energy metering data set, obtaining the stability of each electric energy metering data set according to the degree of fluctuation of each electric energy metering data set, correcting the degree of deviation of each electric energy metering data set according to the stability of each electric energy metering data set, obtaining the degree of abnormality of each electric energy metering data set, obtaining the nu parameter of each electric energy metering data set based on the degree of abnormality, performing abnormality detection on the electric energy metering data in each electric energy metering data set, and obtaining the abnormal data in each electric energy metering data set, and the present invention avoids missed detection or false detection of abnormal data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and system for processing electric energy metering data. Background Art

[0002] With the acceleration of the construction of smart grids in my country, the intelligence of electricity metering equipment and systems has been greatly improved. Therefore, electricity metering data has grown exponentially and presents the characteristics of big data. When processing electricity metering data, it includes multiple links, such as data acquisition, data transmission, data preprocessing, data feature extraction, data anomaly detection, etc. Among them, due to various equipment failures, communication failures, power grid fluctuations and management, abnormal data will appear in the electricity metering data. These abnormal data are important for the accuracy, completeness, self-consistency and dynamic requirements of electricity metering data. They contain important time event information of the power grid. Therefore, it is of great significance to analyze, identify and process abnormal data of electricity metering data. It is necessary to detect abnormal data in electricity metering data first.

[0003] The patent document with the current authorization publication number CN116720753A is a method, system and readable storage medium for processing hydrological data. By collecting hydrological data and performing data verification, it is determined whether the data verification has passed; if not, the abnormal data sequence that has not passed the verification is subsequently processed; the data points are detected for abnormality using the One-ClassSVM model to obtain a first index value; the data points are detected for abnormality using the EWMA algorithm, the COF algorithm, and the IsolationForest algorithm respectively, and the second index value is obtained by weighted summation of the weights of each algorithm; the historical data corresponding to the data point is input into the LSTM model to obtain the fitting value, and the abnormal data point is determined based on the variance of the fitting value and the abnormal data sequence to obtain the third index value; the first index value, the second index value and the third index value are input into the hidden Markov model as observation values ​​to obtain the target abnormal data point.

[0004] The nu parameter of the One-Class SVM algorithm in the above method limits the proportion of abnormal data allowed to exist in the data to be detected. A smaller nu parameter means that fewer abnormal data are allowed to exist, while a larger nu parameter means that more abnormal data are allowed to exist. It is known that in power metering data, there are different numbers of abnormal data in the power metering data at different time series stages. Therefore, when detecting abnormal data in power metering data according to the One-Class SVM algorithm in the above method, using a fixed nu parameter may not be able to adapt to the power metering data at all time series stages, which may lead to omissions or false alarms in the abnormal data obtained in the power metering data at some time series stages, thereby reducing the abnormal data recognition result of the model. Summary of the invention

[0005] In order to solve the problem that a fixed nu parameter may not be able to adapt to the power metering data in all time stages, thereby causing omissions or false alarms in abnormal data obtained in the power metering data in some time stages, the present invention provides a method and system for processing electric energy metering data.

[0006] In a first aspect, the present invention provides a method for processing electric energy metering data, which adopts the following technical solution:

[0007] A method for processing electric energy metering data comprises the following steps:

[0008] Collecting power metering data; dividing the power metering data into a plurality of power metering data sets; obtaining the degree of deviation of each power metering data set, wherein the degree of deviation represents the deviation of the power metering data in the power metering data set; obtaining the stability of each power metering data set, wherein the stability represents the fluctuation of the power metering data in the power metering data set;

[0009] Obtain the abnormality level of each power metering data set , where Represents the abnormality degree of the i-th power metering data set; Represents the degree of deviation of the i-th power metering data set; represents the stationarity of the ith power metering data set; exp() represents an exponential function with a natural constant as the base; the abnormality degree of each power metering data set is taken as the nu parameter of each power metering data set; based on the nu parameter of each power metering data set, anomaly detection is performed on the power metering data in each power metering data set to obtain the abnormal data in each power metering data set.

[0010] The innovation of the present invention lies in dividing the power metering data into several power metering data sets, and adaptively acquiring the nu parameter of each power metering data set according to the abnormality degree of each power metering data set to detect the abnormal data in each power metering data set, thereby solving the problem that the use of a fixed nu parameter may not be able to adapt to the power metering data in all time series stages, so that the abnormal data in the power metering data in different time series stages can be accurately detected, avoiding the phenomenon of underreporting or false reporting of abnormal data. Furthermore, the abnormality degree of each power metering data set is obtained by correcting the offset degree of each power metering data set according to the stability of the power metering data in each power metering data set, thereby improving the accuracy of the offset degree result and avoiding the offset degree of each power metering data set being obtained due to the volatility of the power metering data set itself.

[0011] Preferably, dividing the power metering data into a plurality of power metering data sets comprises the steps of:

[0012] Each power metering data is sorted in ascending order of sampling time to obtain a power metering data sequence, and the power metering data of each hour in the power metering data sequence is sequentially obtained to form each power metering data set.

[0013] This facilitates the subsequent adaptive acquisition of the abnormality level of each power metering data set.

[0014] Preferably, the calculation formula for obtaining the offset degree of each power metering data set is:

[0015] Each cluster of each power metering data set is obtained, and the ratio of the number of power metering data in each cluster of each power metering data set to the number of power metering data in each power metering data set is recorded as the proportion of each cluster of each power metering data set, and the cluster with the largest proportion in each power metering data set is recorded as the largest cluster of each power metering data set, and the cluster with the smallest proportion in each power metering data set is recorded as the smallest cluster of each power metering data set; the Euclidean distance between the cluster center of the largest cluster of each power metering data set and the cluster center of the smallest cluster of each power metering data set is used as the offset degree of each power metering data set.

[0016] The greater the degree of deviation, the more abnormal data there are in the power metering data set, which facilitates the adaptation of the nu parameter of the power metering data set according to the degree of deviation of the power metering data set.

[0017] Preferably, the obtaining of each cluster of each power metering data set includes:

[0018] The k-means clustering algorithm is used to cluster the power metering data in each power metering data set to obtain each cluster cluster of each power metering data set.

[0019] Preferably, the obtaining of the stability of each electric power metering data set includes:

[0020] Obtain the degree of fluctuation of each power metering data set; preset the number of reference data sets M, record the first M power metering data sets of each power metering data set in time series as the reference data set of each power metering data set, obtain the absolute value of the difference between the degrees of fluctuation of any two time-series adjacent reference data sets of each power metering data set, and take the sum of the absolute values ​​of the differences between the degrees of fluctuation of all two time-series adjacent reference data sets of each power metering data set as the stationarity of each power metering data set.

[0021] The smaller the stability of each power metering data set, the more accurate the offset of each power metering data set. Subsequently, the offset of each power metering data set is corrected according to the stability of each power metering data set, avoiding the offset of each power metering data set being caused by the volatility of the power metering data set itself.

[0022] Preferably, the fluctuation degree of each power metering data set is obtained by calculating the formula:

[0023] ;

[0024] In the formula, represents the fluctuation degree of the i-th power metering data set; represents the number of power metering data in the i-th power metering data set; represents the value of the ath power metering data in the i-th power metering data set; represents the mean of all power metering data in the i-th power metering data set; || represents the absolute value symbol.

[0025] Preferably, the obtaining of abnormal data in each power metering data set includes:

[0026] According to the nu parameter of the ith power metering data set, the One-ClassSVM algorithm is applied to the ith power metering data set to obtain a segmentation hyperplane, which is recorded as the segmentation hyperplane of the ith power metering data set. The vertical distance between each power metering data in the ith power metering data set and the segmentation hyperplane of the ith power metering data set is obtained, and the average of the vertical distances between all power metering data in the ith power metering data set and the segmentation hyperplane of the ith power metering data set is recorded as the abnormal threshold of the ith power metering data set. The power metering data in the ith power metering data set whose vertical distance exceeds the abnormal threshold of the ith power metering data set is taken as the abnormal data in the ith power metering data set, and the abnormal data in each power metering data set is obtained.

[0027] Anomaly detection is performed on each power metering data set according to the nu parameter of each power metering data set, so that the abnormal data in each power metering data set can be obtained more accurately.

[0028] In a second aspect, the present invention provides a system for processing electric energy metering data, which adopts the following technical solution:

[0029] A system for processing electric energy metering data comprises: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned method for processing electric energy metering data is implemented.

[0030] By adopting the above technical solution, the above-mentioned method for processing electric energy metering data is generated into a computer program and stored in a memory so as to be loaded and executed by a processor, thereby making a terminal device based on the memory and the processor for easy use.

[0031] The present invention has the following technical effects: the present invention first divides the power metering data into a plurality of power metering data sets, and according to the abnormality degree of each power metering data set, the nu parameter of each power metering data set is adaptively used to detect the abnormal data in each power metering data set, so as to solve the problem that the use of a fixed nu parameter may not be able to adapt to the power metering data in all time series stages, and can better adapt to different power metering data sets, so that the abnormal data in different power metering data sets can be accurately detected, and the phenomenon of missing or false alarm of abnormal data is avoided. Furthermore, the abnormality degree of each power metering data set is obtained by correcting the offset degree of each power metering data set according to the stability of the power metering data in each power metering data set, so as to improve the accuracy of the offset degree result, and also avoid the offset degree of each power metering data set being obtained due to the volatility of the power metering data set itself. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts.

[0033] Figure 1 The present invention is a flowchart of a method for processing electric energy metering data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0035] It should be understood that when the terms "first", "second", etc. are used in the claims, descriptions, and drawings of the present invention, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the description and claims of the present invention indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their collections.

[0036] The embodiment of the present invention discloses a method for processing electric energy metering data, referring to Figure 1 , comprising steps S1 to S3:

[0037] S1: Collect electricity metering data.

[0038] In the embodiment of the present invention, every 1 minute is a sampling moment, and the power metering data of the smart meter is collected in turn each time, and the collection is carried out for a total of five days to obtain a plurality of power metering data; each power metering data is sorted in ascending order according to the sampling time to obtain a power metering data sequence.

[0039] S2: According to the power metering data, each power metering data set is obtained, according to the distribution of the power metering data in each power metering data set, the offset degree of each power metering data set is obtained, the fluctuation degree of each power metering data set is obtained, based on the fluctuation degree, the stationarity of each power metering data set is obtained, according to the stationarity of each power metering data set, the offset degree of each power metering data set is corrected, the abnormality degree of each power metering data set is obtained, and according to the abnormality degree, the nu parameter of each power metering data set is obtained.

[0040] It should be noted that the One-Class SVM algorithm is a two-classification algorithm. This algorithm is used in the detection of abnormal data. The algorithm first obtains a segmentation hyperplane. The segmentation hyperplane is a straight line in two-dimensional space and a plane in three-dimensional space. The segmentation hyperplane is actually a boundary, or a range. If the distance from the data to the hyperplane is within a certain range, the data point can be considered to be "inside the segmentation hyperplane". If it exceeds this range, it can be considered to be "outside the segmentation hyperplane", and the data is abnormal data. Therefore, the algorithm can divide the data into two categories according to the segmentation hyperplane: "inside the segmentation hyperplane" and "outside the segmentation hyperplane" to obtain abnormal data. Since the algorithm has an important Parameter, namely the nu parameter, the nu parameter limits the proportion of abnormal data allowed to exist in the detection model. A smaller nu parameter means that less abnormal data is allowed, while a larger nu parameter means that more abnormal data is allowed. However, in the power metering data, the power metering data in different time series stages have different numbers of abnormal data, that is, the power metering data in different time series stages have different degrees of abnormality. The use of a fixed nu parameter may not be able to adapt to the power metering data in all time series stages, which may lead to omissions or false alarms in the abnormal data obtained in the power metering data in some time series stages, thereby reducing the abnormal data recognition results of the model.

[0041] Step S2 includes steps S20-S21, which are specifically as follows:

[0042] S20: acquiring each power metering data set according to the power metering data, and acquiring the degree of deviation of each power metering data set according to the distribution of the power metering data in each power metering data set.

[0043] It should be noted that it is known that there are different numbers of abnormal data in the power metering data at different time series stages, that is, there are different degrees of abnormality in the power metering data at different time series stages. The use of a fixed nu parameter may not be able to adapt to the power metering data at all time series stages. Therefore, the present invention first divides the collected power addition data into each power metering data set; according to the distribution characteristics of the data in each power metering data set, the degree of abnormality of each power metering data set is obtained; based on the degree of abnormality of each power metering data set, the nu parameter of each power metering data set is adjusted to perform abnormality detection on the power metering data in each power metering data set.

[0044] It should be further explained that the degree of deviation of the power metering data in each power metering data set can be used to measure the degree of abnormality of each power metering data set. When the degree of deviation of the power metering data in any power metering data set is greater, it means that the degree of abnormality of the power metering data set is also greater. At this time, a larger nu parameter is required to allow more abnormal points to exist and promote their accurate identification. In addition, since similar power metering data will be clustered together when the power metering data in the power metering data set are clustered, the deviated power metering data and the non-deviated power metering data have different characteristics in some aspects, and the deviated power metering data are a minority in the entire power metering data set, and there is a certain distance between them and other non-deviated power metering data. Therefore, the power metering data in each power metering data set are first clustered to obtain each cluster cluster of each power metering data set. According to the number of power metering data in the cluster cluster, the maximum cluster cluster and the minimum cluster cluster are obtained. Finally, the degree of deviation of each power metering data set is quantified according to the distance between the maximum cluster cluster and the minimum cluster cluster.

[0045] In the embodiment of the present invention, the power metering data of each hour in the power metering data sequence is sequentially acquired to form each power metering data set.

[0046] The k-means clustering algorithm is used to cluster the power metering data in each power metering data set to obtain each cluster cluster of each power metering data set. It should be noted that the k-means clustering algorithm is a well-known technology and will not be described in detail in the embodiments of the present invention.

[0047] The ratio of the number of power metering data in the jth cluster of the i-th power metering data set to the number of power metering data in the i-th power metering data set is recorded as the proportion of the jth cluster of the i-th power metering data set. Similarly, the proportion of each cluster of the i-th power metering data set is obtained; the cluster with the largest proportion in the i-th power metering data set is recorded as the largest cluster of the i-th power metering data set, and the cluster with the smallest proportion in the i-th power metering data set is recorded as the smallest cluster of the i-th power metering data set; the Euclidean distance between the cluster center of the largest cluster of the i-th power metering data set and the cluster center of the smallest cluster of the i-th power metering data set is taken as the degree of deviation of the i-th power metering data set.

[0048] Similarly, the deviation degree of each power metering data set is obtained.

[0049] S21: Obtain the degree of fluctuation of each power metering data set, obtain the stability of each power metering data set based on the degree of fluctuation, correct the degree of deviation of each power metering data set according to the stability of each power metering data set, obtain the degree of abnormality of each power metering data set, and obtain the nu parameter of each power metering data set according to the degree of abnormality.

[0050] It should be noted that it is known that the stability of each power metering data set can also affect the degree of abnormality of each power metering data set. When the power metering data set is stable, the degree of deviation of the power metering data set is more credible. If the power metering data set itself is unstable, the credibility of the degree of deviation of the obtained power metering data set is lower. Therefore, it is necessary to obtain the stability of each power metering data set and correct the degree of deviation of each power metering data set, so as to obtain the degree of abnormality of each power metering data set.

[0051] It should be further explained that, according to the fluctuation of the power metering data in each power metering data set, the fluctuation degree of each power metering data set is obtained. When the sum of the differences in the fluctuation degrees between all two power metering data sets before the time series of any power metering data set is small, it means that the possibility of the fluctuation degree of the power metering data set changing over time is smaller, the stability of the power metering data set is higher, and the offset degree of the power metering data set is more reliable.

[0052] In an embodiment of the present invention, the fluctuation degree of each power metering data set is obtained:

[0053]

[0054] In the formula, represents the fluctuation degree of the i-th power metering data set; represents the number of power metering data in the i-th power metering data set; represents the value of the ath power metering data in the i-th power metering data set; represents the mean value of all power metering data in the i-th power metering data set; || represents the absolute value symbol; The larger the value of , the greater the fluctuation of the value of the power metering data in the i-th power metering data set.

[0055] Obtaining the stationarity of each power metering data set: recording the first M power metering data sets of each power metering data set in time series as the reference data set of each power metering data set, obtaining the absolute value of the difference between the fluctuation degree of any two time-series adjacent reference data sets of each power metering data set, and taking the sum of the absolute values ​​of the difference between the fluctuation degrees of all two time-series adjacent reference data sets of each power metering data set as the stationarity of each power metering data set; in an embodiment of the present invention, the preset number of reference data sets M=5, and in other embodiments, the implementer may preset the number of reference data sets according to specific implementation conditions.

[0056] Get the degree of anomaly for each electricity metering dataset:

[0057]

[0058] In the formula, Represents the abnormality degree of the i-th power metering data set; Represents the degree of deviation of the i-th power metering data set; represents the stationarity of the i-th power metering data set; when the stationarity of the i-th power metering data set is smaller, the deviation degree of the i-th power metering data set is more credible; The larger the value of is, the greater the abnormality of the ith power metering data set is. In this case, a larger nu parameter is needed to allow more abnormal points to exist and enable them to be accurately identified.

[0059] Obtaining the nu parameter of each power metering data set: taking the abnormality degree of each power metering data set as the nu parameter of each power metering data set.

[0060] S3: According to the nu parameter of each power metering data set, anomaly detection is performed on the power metering data in each power metering data set to obtain the abnormal data in each power metering data set.

[0061] In an embodiment of the present invention, according to the nu parameter of the ith power metering data set, a One-Class SVM algorithm is applied to the ith power metering data set to obtain a segmentation hyperplane, which is recorded as the segmentation hyperplane of the ith power metering data set, and the vertical distance between each power metering data in the ith power metering data set and the segmentation hyperplane of the ith power metering data set is obtained, and the average of the vertical distances between all the power metering data in the ith power metering data set and the segmentation hyperplane of the ith power metering data set is recorded as the abnormal threshold of the ith power metering data set, and the power metering data in the ith power metering data set whose vertical distance exceeds the abnormal threshold of the ith power metering data set is taken as the abnormal data in the ith power metering data set. Similarly, the abnormal data in each power metering data set is obtained.

[0062] It should be noted that the power metering data processed by the One-Class SVM algorithm in the present invention is two-dimensional time series data (sampling time and data value), so the segmentation hyperplane of each power metering data set obtained is a straight line in the two-dimensional space. Therefore, the vertical distance of each power metering data in the power metering data set from the segmentation hyperplane of the power metering data set is actually the vertical distance of each power metering data in the power metering data set from a straight line.

[0063] The beneficial effect of the present invention is that the power metering data is divided into several power metering data sets, and according to the abnormality degree of each power metering data set, the nu parameter of each power metering data set is adaptively obtained to detect the abnormal data in each power metering data set, which solves the problem that the use of a fixed nu parameter may not be able to adapt to the power metering data in all time series stages, so that the abnormal data in the power metering data in different time series stages can be accurately detected, avoiding the phenomenon of underreporting or false reporting of abnormal data.

[0064] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.

[0065] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory, a dynamic random access memory, a static random access memory, an enhanced dynamic random access memory, a high bandwidth memory, a hybrid storage cube, etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of the device or accessible or connectable to the device.

[0066] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

[0067] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for processing electric energy metering data, characterized in that: Includes steps: Collecting power metering data; dividing the power metering data into a plurality of power metering data sets, including the steps of: sorting each power metering data in ascending order of sampling time to obtain a power metering data sequence, and sequentially obtaining the power metering data of each hour in the power metering data sequence to form each power metering data set; The degree of deviation of each power metering data set is obtained, and the calculation formula is: obtain each cluster of each power metering data set, record the ratio of the number of power metering data in each cluster of each power metering data set to the number of power metering data in each power metering data set as the proportion of each cluster of each power metering data set, record the cluster with the largest proportion in each power metering data set as the largest cluster of each power metering data set, and record the cluster with the smallest proportion in each power metering data set as the smallest cluster of each power metering data set; The Euclidean distance between the cluster center of the largest cluster of each power metering data set and the cluster center of the smallest cluster of each power metering data set is used as the offset degree of each power metering data set; The offset degree represents the offset of the power metering data in the power metering data set; Obtaining the stability of each power metering data set includes: obtaining the fluctuation degree of each power metering data set, and the calculation formula is: ; In the formula, represents the fluctuation degree of the i-th power metering data set; represents the number of power metering data in the i-th power metering data set; represents the value of the ath power metering data in the i-th power metering data set; represents the mean of all the power metering data in the i-th power metering data set; || represents the absolute value symbol; the number of reference data sets M is preset, the first M power metering data sets of each power metering data set in time series are recorded as the reference data set of each power metering data set, the absolute value of the difference between the fluctuation degree of any two reference data sets adjacent in time series of each power metering data set is obtained, and the sum of the absolute values ​​of the difference between the fluctuation degree of all two reference data sets adjacent in time series of each power metering data set is taken as the stationarity of each power metering data set; the stationarity represents the fluctuation amount of the power metering data in the power metering data set; Obtain the abnormality level of each power metering data set , where Represents the abnormality degree of the i-th power metering data set; Represents the degree of deviation of the i-th power metering data set; represents the stationarity of the ith power metering data set; exp() represents an exponential function with a natural constant as the base; the abnormality degree of each power metering data set is taken as the nu parameter of each power metering data set; based on the nu parameter of each power metering data set, anomaly detection is performed on the power metering data in each power metering data set to obtain the abnormal data in each power metering data set.

2. The method for processing electric energy metering data according to claim 1, characterized in that: The step of obtaining each cluster of each power metering data set includes: The k-means clustering algorithm is used to cluster the power metering data in each power metering data set to obtain each cluster cluster of each power metering data set.

3. The method for processing electric energy metering data according to claim 1, characterized in that: The obtaining of abnormal data in each power metering data set includes: According to the nu parameter of the ith power metering data set, the One-Class SVM algorithm is applied to the ith power metering data set to obtain a segmentation hyperplane, which is recorded as the segmentation hyperplane of the ith power metering data set. The vertical distance between each power metering data in the ith power metering data set and the segmentation hyperplane of the ith power metering data set is obtained, and the average of the vertical distances between all power metering data in the ith power metering data set and the segmentation hyperplane of the ith power metering data set is recorded as the abnormal threshold of the ith power metering data set. The power metering data in the ith power metering data set whose vertical distance exceeds the abnormal threshold of the ith power metering data set is taken as the abnormal data in the ith power metering data set, and the abnormal data in each power metering data set is obtained.

4. A system for processing electric energy metering data, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a method for processing electric energy metering data according to any one of claims 1 to 3 is implemented.