Method and system for correcting and additionally recording incomplete data of intelligent electric meter

By screening out similar daily data sets of the days when the smart meter data is missing, and data filling is performed based on its data characteristics, the problem of large errors in the period of adjacent data filling in the existing technology is solved, and more accurate data completion is achieved .

CN118916352APending Publication Date: 2024-11-08ZHEJIANG SUPCON INFORMATION TECH CO LTD
View PDF 0 Cites 4 Cited by

Patent Information

Application Number
CN202411104905.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-13
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

When the prior art supplements the incomplete data of smart electricity meters, especially electricity consumption data with periodic characteristics, there are large errors in filling through adjacent data, resulting in inaccurate completion data.

Method used

By obtaining the time period corresponding to the missing data segment of the missing data on the missing data, filter out the complete data of the same time period from the historical data, forming several similar daily data sets, filter out the similar daily data set with the highest data similarity, and according to them Data characteristics fill the missing data segments to ensure that the filling results more accurately reflect the actual power consumption characteristics.

Benefits of technology

Compared with trend analysis and filling of adjacent data, using a similar daily data set method can more accurately simulate the power consumption trends and characteristics of missing data segments, thereby obtaining more accurate data completion results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118916352A_ABST
    Figure CN118916352A_ABST
Patent Text Reader

Abstract

The invention discloses a method and a system for correcting and additionally recording incomplete data of an intelligent electric meter. The method comprises the following steps: acquiring a time period corresponding to a data missing segment of a data missing day; screening out complete data in the same time period from the historical data, and forming a plurality of similar day data sets according to the relevance with the data missing day; a similar day data set with the highest data similarity is screened out, and data features in the similar day data set are calculated; and filling the data missing segment based on the data range corresponding to the data missing segment and the data feature. According to the method, the similar day data set most similar to the power consumption trend of the data missing segment of the data missing day is searched for to simulate the data feature and trend of the data missing segment for data filling, and compared with trend analysis and data filling of adjacent data, the method can be more accurate and close to the power consumption feature of an actual missing time period; therefore, a more accurate data completion result is obtained.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of electronic data processing, and in particular to a method and system for correcting and supplementing incomplete data of a smart electric meter. Background Art

[0002] With the continuous development of the power system, traditional meters are gradually replaced by more economical and intelligent smart meters. As an indispensable component of the construction of power systems or energy consumption systems, the integrity and correctness of electricity consumption data have become the most concerned objects of the system. The meter data is mainly collected by the system according to the collection frequency. The smart meter data is persisted in the back-end database, and the hourly consumption is calculated based on the collected data. In addition to the reasons such as the damage of the communication module and the counting chip, the smart meter may also fail to collect the original data of the meter due to the downtime of the application acquisition module and the failure of the acquisition task. The currently commonly used data supplementation method adopts the single-threaded adjacent value method. The missing data supplemented by this method has a large error and cannot reflect the real energy consumption curve. Especially for electricity consumption data with periodic characteristics, the electricity consumption characteristics of the time period with missing data may be different from the electricity consumption characteristics of the adjacent time period with data. The data obtained by supplementing the adjacent data does not conform to the actual data characteristics of the missing time period, and the supplemented data is inaccurate.

[0003] The "User Missing Data Filling Method, Device and Equipment" disclosed in the Chinese patent literature has a publication number of CN113051256A and a publication date of 2021-06-29, including: extracting power data from the original power data matrix to be filled through a preset sliding time window to form a time window data matrix; based on an interpolation algorithm, pre-filling the missing data in the time window data matrix to obtain a pre-filled matrix; determining the first user in the pre-filled matrix, wherein the first user is the target user with missing power data; for each first user, searching the pre-filled matrix for multiple second users whose power data is similar to the first user, and forming a similarity matrix corresponding to the first user with the power data of the first user and the power data of the multiple second users, and based on a singular value threshold algorithm, determining the filling value of the missing data of the first user in the similarity matrix corresponding to the first user, and filling it into the corresponding position in the original power data matrix, which can improve the filling accuracy of the user's missing data. However, this technology still fills the gaps with data adjacent to the missing data. For electricity consumption data with time period characteristics, there may be significant differences in the electricity consumption characteristics between the time period of missing data and the adjacent time period. Filling the missing data with the characteristic trends of the adjacent data will result in large errors, and the filled data will be inaccurate. Summary of the invention

[0004] The present invention aims to overcome the problem in the prior art that for electricity consumption data with time period characteristics, data filling based on adjacent data of missing data will result in large errors and inaccurate supplemented data, and provides a method and system for correcting and supplementing incomplete data of a smart meter.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: A method for correcting and supplementing incomplete data of a smart meter, comprising: Get the time period corresponding to the data missing segment of the data missing day; Filter out complete data of the same time period from historical data, and form several similar day data sets based on their correlation with the days with missing data; Filter out the similar day data set with the highest data similarity, and calculate the data features in the similar day data set; The missing data segment is filled based on the data range corresponding to the missing data segment and the data features.

[0006] The present invention searches for a large amount of electricity consumption data in the same time period as the missing data based on the acquired historical data, and groups the electricity consumption data according to the characteristics or influencing factors of the day when the data is missing, screens out several similar day data sets that are highly correlated with the day when the data is missing, and performs similarity analysis on the data in each similar day data set to find a group of data sets with the highest similarity. The electricity consumption trends and characteristics of the data sets are used to represent the electricity consumption characteristics and trends of the time period corresponding to the missing data, thereby filling in the missing data. Compared with trend analysis and data filling of adjacent data, it can be more accurate and close to the electricity consumption characteristics of the actual missing time period, thereby obtaining a more accurate data completion result.

[0007] Preferably, the plurality of similar day data sets at least include: The same-day-in-a-month data set contains complete data for the same time period of several days that are on the same day but in a different month than the missing day; the same-day-in-a-week data set contains complete data for the same time period of several days that are on the same day but in a different week than the missing day; the consecutive-day data set contains complete data for the same time period of several consecutive days adjacent to the missing day.

[0008] Preferably, the screening out of similar day data sets with the highest data similarity comprises: For any similar day data set, calculate the element similarity between any two elements in the similar day data set; The similarity value of the dataset is obtained by averaging the similarities of all elements in the similar day dataset; The dataset with the smallest similarity value is selected as the similar day dataset with the highest data similarity.

[0009] Preferably, the time period corresponding to the data missing segment includes several sampling intervals; Calculate the proportion of electricity consumption data in each sampling interval based on the similar day data set with the highest data similarity; Calculate the total electricity consumption of the data missing segment based on the data range corresponding to the data missing segment; The total electricity consumption is allocated according to the proportion of electricity consumption data, and the data missing segments are filled in according to the sampling interval.

[0010] Preferably, the process of calculating the proportion of electricity consumption data in each sampling interval includes: Calculate the sum of energy usage of all elements in the same time period based on similar day data sets; and The power consumption data of all elements in the same sampling interval; The proportion of electricity consumption data in each sampling interval is calculated by dividing the electricity consumption data by the total energy consumption.

[0011] Preferably, the method of screening out the similar day data set with the highest data similarity also includes: For any similar day data set, the time period corresponding to the missing data segment is extended forward and backward to form a new similar day data set; the element similarity between any two elements in the similar day data set is calculated; The similarity value of the dataset is obtained by averaging the similarities of all elements in the similar day dataset; The dataset with the smallest similarity value is selected as the similar day dataset with the highest data similarity.

[0012] Preferably, the calculating the element similarity between any two elements in the similar day data set comprises: Where X and Y are two elements in the similar day data set, each element contains a number of electricity consumption data arranged in the order of sampling time; n represents the number of sampling moments in the time period corresponding to the data missing segment, and the subscript i represents the electricity consumption data at the i-th sampling moment.

[0013] Preferably, the calculating the element similarity between any two elements in the similar day data set comprises: Where X and Y are two elements in the similar day data set, each element contains several electricity consumption data arranged in the order of sampling time; α i is the weight coefficient, α i ∈[0, 1); m represents the number of sampling moments in the extended time period; the subscript i represents the electricity consumption data at the i-th sampling moment, and i∈[p, q] represents the sampling moment in the time period corresponding to the data missing segment.

[0014] A system for correcting and supplementing incomplete data of a smart meter, comprising: Historical database, which stores complete historical electricity consumption data; The similar day data set generation module selects the complete data of the same time period according to the time period corresponding to the data missing segment of the data missing day, and forms several similar day data sets; Matching module, which screens similar day datasets for data similarity; The data filling module fills the missing data segments according to the data range corresponding to the missing data segments and the data characteristics of similar daily data sets.

[0015] As a preference, it also includes: A data missing detection module, which detects the data missing segments and corresponding time periods of the sampled data; Timing module, set the interval time for data missing detection and the interval time for data filling.

[0016] The present invention has the following beneficial effects: finding a similar day data set that is most similar to the electricity consumption trend of the data missing segment of the data missing day to simulate the data characteristics and trends of the data missing segment for data filling. Compared with trend analysis and data filling of adjacent data, it can be more accurate and close to the electricity consumption characteristics of the actual missing time period, thereby obtaining a more accurate data completion result. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of the method for correcting and supplementing incomplete data of a smart meter in the present invention. DETAILED DESCRIPTION

[0018] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0019] Missing data filling is an important part of data processing. It aims to reasonably estimate and fill the missing values ​​in the data set through certain methods and techniques to improve the accuracy and reliability of data analysis. Existing data filling methods include regression analysis-based methods: by establishing a regression model, using known data to predict missing values, which is suitable for situations where there is a linear relationship between variables in the data set; neighbor value interpolation method: using adjacent data to fill in the difference, etc. Existing data filling methods all use the adjacent data of the missing data to fill in the gaps, and simulate the trend law of the missing data according to the trend law of the adjacent data. However, this method of filling in data gaps is not accurate for electricity consumption data with time characteristics. When there are great differences in the trend law of electricity consumption in adjacent time periods, there will be great errors when using existing data filling methods. For example, the electricity consumption characteristics of an office building during the working hours and the off-get off work hours are completely different, and the characteristics of the peak electricity consumption time period and the adjacent flat or low peak time period are also different. It is impossible to use adjacent data to accurately fill in the gaps.

[0020] Based on the above problems, the present invention provides Figure 1 A method for correcting and recording incomplete data of a smart meter is shown, comprising: Get the time period corresponding to the data missing segment of the data missing day; Filter out complete data of the same time period from historical data, and form several similar day data sets based on their correlation with the days with missing data; Filter out the similar day data set with the highest data similarity, and calculate the data features in the similar day data set; The missing data segment is filled based on the data range corresponding to the missing data segment and the data features.

[0021] It should be noted that the present invention searches for a large amount of electricity consumption data in the same time period as the missing data based on the acquired historical data, and groups these electricity consumption data according to the characteristics or influencing factors of the day when the data is missing, and screens out several similar day data sets with a high correlation with the day when the data is missing, and performs similarity analysis on the data in each similar day data set to find a group of data sets with the highest similarity, and uses the electricity consumption trends and characteristics of the data sets to represent the electricity consumption characteristics and trends of the time period corresponding to the missing data, thereby filling in the missing data. Compared with trend analysis and data filling of adjacent data, it can be more accurate and close to the actual electricity consumption characteristics of the missing time period, thereby obtaining more accurate data completion results.

[0022] It is worth noting that each similar day data set contains complete electricity consumption data for the same time period on several different dates. There is a certain correlation between the dates in each similar day data set and the days with missing data, such as similar date features or similar electricity consumption influencing factors, so that the data in each similar day data set has certain similar patterns. In the present invention, a similarity comparison is performed on the electricity consumption data curves of the same time period in each similar day data set to determine the similarity of the electricity consumption data between each date in the similar day data set. The higher the similarity, the closer the electricity consumption trend and pattern of the corresponding similar day data set is to the trend pattern of the data missing segment of the data missing day, so that the electricity consumption trend and pattern with the highest degree of fit to the data missing segment can be found to fill in the data.

[0023] As a specific embodiment, the plurality of similar day data sets at least include: The same-day-in-a-month data set contains complete data for the same time period of several days that are on the same day but in a different month than the missing day; the same-day-in-a-week data set contains complete data for the same time period of several days that are on the same day but in a different week than the missing day; the consecutive-day data set contains complete data for the same time period of several consecutive days adjacent to the missing day.

[0024] It should be noted that there is a certain correlation between the date in each similar day data set and the day with missing data, such as similar date features or similar factors affecting electricity consumption, so that the data in each similar day data set has certain similar patterns. For the same day data set in a month, the date features of the same day in different months are considered for data integration, which can mainly distinguish the differences between different times such as the beginning, middle and end of the month; for the same day data set in a week, the date features of the same day in different weeks are considered for data integration, which can mainly distinguish the differences between weekdays and weekends and different weekdays; for the continuous day data set, the weather in adjacent consecutive days is considered to be relatively similar and the factors affecting electricity consumption are relatively consistent.

[0025] Specifically, if June 23, 2024 is a Sunday and there is a data missing segment, the data on the same day of the month for the missing data day can be selected from the same day of different months before the missing data day, such as May 23, 2024, April 23, 2024, March 23, 2024, and February 23, 2024, etc. to form the same day of the month data set. The data on the same day of the week for the missing data day can be selected from the same day of different weeks before the missing data day, such as June 16, 2024, June 9, 2024, June 2, 2024, and May 26, 2024, etc. to form the same day of the week data set. The data on the consecutive days of the missing data day can be selected from the previous days, such as June 22, 2024, June 21, 2024, June 20, 2024, and June 19, 2024, etc. to form the consecutive day data set.

[0026] It is worth noting that in addition to the above-mentioned monthly same-day data set, weekly same-day data set and continuous day data set, data sets with other characteristics can also be selected or added according to actual needs, such as the annual same-day data set, which considers the date characteristics of the same day of each year for data integration, and can mainly compare the electricity consumption of specific holidays or special dates of each year; the same odd / even day data set, which considers the data of the even-numbered days adjacent to the data missing day (the date of the data missing day is an even number) for integration, or considers the data of the odd-numbered days adjacent to the data missing day (the date of the data missing day is an odd number) for integration.

[0027] The electricity consumption data in the present invention is updated every day, that is, the electricity consumption data of each day is accumulated from zero at the beginning of the day until the end of the day and then restarted from a new day; for the electricity consumption data accumulated daily in the electric energy meter, the electricity consumption data at each sampling moment of the day is subtracted from the sampling data at the last moment of the previous day to obtain the electricity consumption data at each moment of the day as the data required for the similar day data set of the present invention.

[0028] Embodiment 1: For specific data collection of smart meters, a sampling interval is generally set, and data collection is performed every other sampling interval to obtain a full-day electricity consumption data sequence. In this embodiment, sampling is performed at intervals of fifteen minutes to obtain a historical database that stores complete historical electricity consumption data.

[0029] For the electricity consumption data on the data-missing day, the data-missing segments of the electricity consumption data on the data-missing day and the time periods corresponding to the data-missing segments are detected. Specifically, taking the electricity consumption data collected on June 23, 2024 as an example, the electricity consumption data is missing in the time period between 12:00 and 14:00, which corresponds to the missing electricity consumption data at the seven sampling times of 12:15, 12:30, 12:45, 13:00, 13:15, 13:30 and 13:45. However, when actually calculating the time period corresponding to the missing data segment, two additional sampling times of 12:00 and 14:00 are required, for a total of nine sampling times for analysis. The first and last data of the data-missing segment on the data-missing day are not missing, and are used to determine the specific range of the data in the data-missing segment.

[0030] The complete electricity consumption data of the same time period on different dates are screened out from the historical database, that is, the electricity consumption data of the nine sampling times between 12:00 and 14:00 every day are screened out; and the corresponding similar day data sets are formed according to the correlation with the data missing days, including the monthly same-day data set, the weekly same-day data set and the continuous day data set.

[0031] Specifically, the complete electricity consumption data of the same day and the same time period in the six months before the data missing date can be selected to form a monthly same-day data set [A m , Bm , C m , D m , E m , F m ]; You can select the complete electricity consumption data of the same day and the same time period in the 6 weeks before the data missing date to form a weekly same day data set [A w , B w , C w , D w , E w , F w ]; You can select the complete electricity consumption data of the same time period for 6 consecutive days before the data missing date to form a continuous day data set [A d , B d , C d , D d , E d , F d ]. The specific number of options is not fixed and can be set according to actual needs.

[0032] Furthermore, after obtaining several similar day data sets, the similar day data sets with the highest data similarity are screened out.

[0033] For any similar day data set, calculate the element similarity between any two elements in the similar day data set, and calculate the element similarity between any two elements in the similar day data set by the Euclidean distance calculation method: Where X and Y are two elements in the same similar day data set, each element contains a number of electricity consumption data arranged in the order of sampling time; n represents the number of sampling moments in the time period corresponding to the data missing segment (in this embodiment, there are 9 sampling moments between 12:00 and 14:00), and the subscript i represents the electricity consumption data at the i-th sampling moment.

[0034] Specifically, take the same day in the month dataset [A m , B m , C m , D m , E m , F m ] is used as an example to illustrate the calculation of element A m The element similarity d between the corresponding 9 data sequences and the data sequences corresponding to the following five elements AB , d AC , d AD , d AE , d AF ; Calculate element B m The element similarity d between the corresponding 9 data sequences and the data sequences corresponding to the following four elementsBC , d BD , d BE , d BF ; Calculation element C m The element similarity d between the corresponding 9 data sequences and the data sequences corresponding to the following three elements CD , d CE , d CF ; Until the calculation element E m The element similarity d between the corresponding 9 data sequences and the data sequence corresponding to the next element EF , get the similarity result d of all elements in the same day of the month dataset AB , d AC , d AD , d AE , d AF , d BC , d BD , d BE , d BF , d CD , d CE , d CF , d DE , d DF , d EF The similarity results of all elements of the weekly same day data set and the similarity results of all elements of the consecutive day data set are calculated in the same way.

[0035] Optionally, in addition to the Euclidean distance method used in this embodiment to calculate data similarity, other existing similarity calculation methods can also be used to calculate the similarity between any two elements in the similar day data set; including Manhattan distance calculation, cosine similarity calculation represented by sampling time and corresponding electricity consumption data as vectors, Pearson correlation coefficient calculation between two elements, etc.

[0036] The similarity value of the dataset is obtained by averaging the similarities of all elements in the similar day dataset.

[0037] Specifically, for all element similarities of the same-day-in-month dataset, the sum of all element similarities is divided by the number of element similarities to obtain the average dataset similarity value of the same-day-in-month dataset: The similarity value of the dataset of the same day of the week is calculated in the same way: The dataset similarity value of the continuous day dataset is

[0038] The dataset with the smallest similarity value is selected as the similar day dataset with the highest data similarity.

[0039] Specifically, when calculating the similarity of data using the Euclidean distance, the smaller the calculation result, the higher the similarity between the two elements. Therefore, for the similar day data set with the smallest data set similarity value, the electricity consumption data of different days and the same period within it have a high trend and feature similarity with the electricity consumption data of the data missing period on the data missing day, which can be used to represent the electricity consumption trend and features of the data missing period. Corresponding similar day dataset.

[0040] As an optional embodiment, after obtaining a similar day data set with the highest data similarity to the data missing segment, data feature calculation of the similar day data set is performed. The data feature in this embodiment is the proportion of electricity consumption data in each sampling interval.

[0041] Specifically, the time period corresponding to the data missing segment includes several sampling intervals; that is, there are 9 sampling moments between the data missing segment from 12:00 to 14:00, corresponding to 8 sampling intervals. The electricity consumption data corresponding to the jth element at the i-th sampling moment in the similar day data set is D ij , j∈[1,n] corresponds to each sampling moment, and j∈[1,m] corresponds to each element in the data set arranged in sequence.

[0042] The proportion of electricity consumption data in each sampling interval is calculated based on the similar day data set with the highest data similarity. The process of calculating the proportion of electricity consumption data in each sampling interval includes: Calculate the sum of energy usage of all elements in the same time period based on similar day data sets; and The power consumption data of all elements in the same sampling interval; Calculate the proportion of electricity consumption data in each sampling interval by dividing the electricity consumption data by the total energy consumption First, calculate the total energy consumption of all elements in the similar day data set in the same time period: That is, the difference between the power consumption data at the last sampling moment and the power consumption data at the first sampling moment in the time period corresponding to the missing data segment is calculated, and the difference of all elements is summed.

[0043] For the n sampling moments in the data missing segment, there are n-1 corresponding sampling intervals. The power consumption data of each sampling interval is divided by the total energy consumption to obtain the proportion of power consumption data of the corresponding sampling interval: That is, the power consumption data of the i-th sampling interval is the difference between the power consumption data at the i+1-th sampling moment and the power consumption data at the i-th sampling moment, and the difference of all elements is summed. The power consumption data proportion is the power consumption data of the i-th sampling interval divided by the total energy consumption.

[0044] The total power consumption of the data missing segment is calculated based on the data range corresponding to the data missing segment; the power consumption data of the n sampling moments of the data missing segment can be expressed as: 1 Data D 1 , the second sampling time T 2 Data D 2 , until the nth sampling time T n Data D n The data D at the first sampling time in the data missing segment 1 and the data D at the nth sampling moment n is known, and the data in the middle needs to be filled, so the total power consumption of the data missing segment is Q = D n -D 1 .

[0045] The total electricity consumption is allocated according to the proportion of electricity consumption data, and the data missing segments are filled in according to the sampling interval.

[0046] Specifically, the proportion of electricity consumption data corresponding to each sampling interval can be expressed as (T 1 , R 1 ), (T 2 , R 2 ),…,(T n-1 , R n-1 ). In the process of filling the missing data segment, the i-th sampling time T i The electricity consumption data filled in is D i =D i-1 +Q·R i-1 , i∈[2,n-1]. This completes the missing data filling.

[0047] Embodiment 2: The prerequisites in this embodiment are the same as those in Embodiment 1, except that the selection of data in the similar day data set is improved.

[0048] In this embodiment, after obtaining a plurality of similar day data sets through the same process as in the first embodiment, the similar day data sets with the highest data similarity are screened out.

[0049] For any similar day data set, the time period corresponding to the missing data segment is extended forward and backward to form a new similar day data set.

[0050] Specifically, in some cases, the missing time of the collected data may be relatively short, and the number of missing sampling moments is relatively small. In this case, when the time period corresponding to the data missing segment is used to screen the similar day data set for similarity calculation, the final similarity result calculated is not ideal because the number of data sequences in each group is too small. For example, if the actual data missing segment is between 12:00 and 13:00, the missing data is the data of the three sampling moments of 12:15, 12:30, and 12:45. Although the sampling moments of the two boundaries of 12:00 and 13:00 will still be included in the selection of the specific time period, when the number of data points is small, it will have a certain impact on the final similarity calculation result. In this case, it is appropriate to consider relaxing a certain time period based on the missing data segment and increasing the number of sampling moments, such as relaxing it to between 11:30 and 13:30, including the known data of the four adjacent time periods of 11:30, 11:45, 13:15, and 13:30 and the data missing segment between 12:00 and 13:00.

[0051] Calculate the element similarity between any two elements in the similar day data set; calculate the element similarity between any two elements in the similar day data set using the Euclidean distance calculation method: Where X and Y are two elements in the similar day data set, each element contains several electricity consumption data arranged in the order of sampling time; α i is the weight coefficient, α i ∈[0, 1); m represents the number of sampling moments in the extended time period; the subscript i represents the electricity consumption data at the i-th sampling moment, and i∈[p, q] represents the sampling moment in the time period corresponding to the data missing segment.

[0052] In the process of calculating the similarity of elements, the data of the time period corresponding to the data missing segment is still sampled in the normal similarity calculation method, and for the data of the additional extended time period, the weight coefficient α is multiplied in the calculation process when calculating the similarity i , weight coefficient α iIt can be expressed by the correlation between the data of the corresponding time period of the data missing segment and the data of the adjacent extended time period in the same day in the historical data. For example, in this embodiment, the Pearson correlation coefficient between the data missing segments 12:00, 12:15, 12:30, 12:45, 13:00 and the adjacent extended time periods 11:30, 11:45, 12:00 can be calculated. If the number of sampling moments of the adjacent extended time period is inconsistent with the number of sampling moments of the data missing segment, the adjacent extended time period before the data missing segment can be filled in when the length is consistent with the data missing segment, and then the weight coefficient of the extended time period before the data missing segment is calculated. Similarly, the weight coefficient of the extended time period after the data missing segment can be calculated.

[0053] Specifically, take the same day in the month dataset [A m , B m , C m , D m , E m , F m ] is used as an example to illustrate the calculation of element A m The element similarity d between the corresponding 9 data sequences and the data sequences corresponding to the following five elements AB , d AC , d AD , d AE , d AF ; Calculate element B m The element similarity d between the corresponding 9 data sequences and the data sequences corresponding to the following four elements BC , d BD , d BE , d BF ; Calculation element C m The element similarity d between the corresponding 9 data sequences and the data sequences corresponding to the following three elements CD , d CE , d CF ; Until the calculation element E m The element similarity d between the corresponding 9 data sequences and the data sequence corresponding to the next element EF , get the similarity result d of all elements in the same day of the month dataset AB , d AC , d AD , d AE , d AF , d BC , d BD , d BE , d BF , d CD , d CE , d CF , d DE , d DF, d EF The similarity results of all elements of the weekly same day data set and the similarity results of all elements of the consecutive day data set are calculated in the same way.

[0054] Optionally, in addition to the Euclidean distance method used in this embodiment to calculate data similarity, other existing similarity calculation methods can also be used to calculate the similarity between any two elements in the similar day data set; including Manhattan distance calculation, cosine similarity calculation represented by sampling time and corresponding electricity consumption data as vectors, Pearson correlation coefficient calculation between two elements, etc.

[0055] The similarity value of the dataset is obtained by averaging the similarities of all elements in the similar day dataset.

[0056] Specifically, for the similarities of all elements in the same-day-in-month dataset, the sum of all element similarities is divided by the number of element similarities to obtain the average dataset similarity value of the same-day-in-month dataset: The similarity value of the dataset of the same day of the week is calculated in the same way: The dataset similarity value of the continuous day dataset is

[0057] The dataset with the smallest similarity value is selected as the similar day dataset with the highest data similarity.

[0058] Specifically, when calculating the similarity of data using the Euclidean distance, the smaller the calculation result, the higher the similarity between the two elements. Therefore, for the similar day data set with the smallest data set similarity value, the electricity consumption data of different days and the same period within it have a high trend and feature similarity with the electricity consumption data of the data missing period on the data missing day, which can be used to represent the electricity consumption trend and features of the data missing period. Corresponding similar day dataset.

[0059] When the similarity is calculated by other existing similarity calculation methods, a judgment is made based on the corresponding similarity calculation result and the related similarity high and low relationship, and the similar day data set with the highest similarity is selected.

[0060] As an optional embodiment, after obtaining a similar day data set with the highest data similarity to the data missing segment, the data feature calculation of the similar day data set is performed. When calculating the data feature of the similar day data set, it is no longer necessary to consider the time period after the data missing segment is expanded, and only the power consumption data of the time period corresponding to the data missing segment in the screened similar day data set needs to be considered. The data feature in this embodiment is the proportion of power consumption data in each sampling interval.

[0061] Specifically, the time period corresponding to the data missing segment includes several sampling intervals; that is, there are 5 sampling moments between the data missing segment from 12:00 to 13:00, corresponding to 4 sampling intervals. The electricity consumption data corresponding to the jth element at the i-th sampling moment in the similar day data set is D ij , i∈[1,n] corresponds to each sampling moment, and j∈[1,m] corresponds to each element in the data set arranged in sequence.

[0062] The proportion of electricity consumption data in each sampling interval is calculated based on the similar day data set with the highest data similarity. The process of calculating the proportion of electricity consumption data in each sampling interval includes: Calculate the sum of energy usage of all elements in the same time period based on similar day data sets; and The power consumption data of all elements in the same sampling interval; Calculate the proportion of electricity consumption data in each sampling interval by dividing the electricity consumption data by the total energy consumption First, calculate the total energy consumption of all elements in the similar day data set in the same time period: That is, the difference between the power consumption data at the last sampling moment and the power consumption data at the first sampling moment in the time period corresponding to the missing data segment is calculated, and the difference of all elements is summed.

[0063] For the n sampling moments in the data missing segment, there are n-1 corresponding sampling intervals. The power consumption data of each sampling interval is divided by the total energy consumption to obtain the proportion of power consumption data of the corresponding sampling interval: That is, the power consumption data of the i-th sampling interval is the difference between the power consumption data at the i+1-th sampling moment and the power consumption data at the i-th sampling moment, and the difference of all elements is summed. The power consumption data proportion is the power consumption data of the i-th sampling interval divided by the total energy consumption.

[0064] The total power consumption of the data missing segment is calculated based on the data range corresponding to the data missing segment; the power consumption data of the n sampling moments of the data missing segment can be expressed as: 1 Data D 1 , the second sampling time T 2 Data D 2 , until the nth sampling time T n Data D n The data D at the first sampling time in the data missing segment 1 and the data D at the nth sampling moment n is known, and the data in the middle needs to be filled, so the total power consumption of the data missing segment is Q = D n -D 1.

[0065] The total electricity consumption is allocated according to the proportion of electricity consumption data, and the data missing segments are filled in according to the sampling interval.

[0066] Specifically, the proportion of electricity consumption data corresponding to each sampling interval can be expressed as (T 1 , R 1 ), (T 2 , R 2 ),…,(T n-1 , R n-1 ). In the process of filling the missing data segment, the i-th sampling time T i The electricity consumption data filled in is D i =D i-1 +Q·R i-1 , i∈[2,n-1]. This completes the missing data filling.

[0067] In addition to providing a method for correcting and re-recording incomplete data of a smart meter, the present invention also provides a system for correcting and re-recording incomplete data of a smart meter, including: Historical database, which stores complete historical electricity consumption data; The similar day data set generation module selects the complete data of the same time period according to the time period corresponding to the data missing segment of the data missing day, and forms several similar day data sets according to the correlation with the data missing day; The matching module selects the similar day data sets with the highest data similarity; The data filling module fills the missing data segments according to the data range corresponding to the missing data segments and the data characteristics of similar daily data sets; A data missing detection module, which detects the data missing segments and corresponding time periods of the sampled data; The timing module sets the interval time for data missing detection and the interval time for data filling, so that data missing detection and data filling can be performed regularly.

[0068] The above embodiments are further elaborations and illustrations of the present invention for ease of understanding, and are not limitations of the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for correcting and supplementing incomplete data of a smart meter, characterized in that: include: Get the time period corresponding to the data missing segment of the data missing day; Filter out complete data of the same time period from historical data, and form several similar day data sets based on their correlation with the days with missing data; Filter out the similar day data set with the highest data similarity, and calculate the data features in the similar day data set; The missing data segment is filled based on the data range corresponding to the missing data segment and the data features.

2. A method for correcting and recording incomplete data of a smart meter according to claim 1, characterized in that: The plurality of similar day data sets at least include: Same-day-in-month data set: complete data for the same time period for several days that are in different months but on the same day as the missing day; Same-day-of-week data set: complete data for the same time period on several days that are in different weeks but on the same day as the missing day; Continuous day data set, which contains complete data for the same time period for several consecutive days adjacent to the day with missing data.

3. A method for correcting and supplementing incomplete data of a smart meter according to claim 1 or 2, characterized in that: The similar day data sets with the highest data similarity screened out include: For any similar day data set, calculate the element similarity between any two elements in the similar day data set; The similarity value of the dataset is obtained by averaging the similarities of all elements in the similar day dataset; The dataset with the smallest similarity value is selected as the similar day dataset with the highest data similarity.

4. A method for correcting and supplementing incomplete data of a smart meter according to claim 1 or 2, characterized in that: The time period corresponding to the data missing segment includes a number of sampling intervals; Calculate the proportion of electricity consumption data in each sampling interval based on the similar day data set with the highest data similarity; Calculate the total electricity consumption of the data missing segment based on the data range corresponding to the data missing segment; The total electricity consumption is allocated according to the proportion of electricity consumption data, and the data missing segments are filled in according to the sampling interval.

5. A method for correcting and supplementing incomplete data of a smart meter according to claim 4, characterized in that: The process of calculating the proportion of power consumption data in each sampling interval includes: Calculate the sum of energy usage of all elements in the same time period based on similar day data sets; and The power consumption data of all elements in the same sampling interval; The proportion of electricity consumption data in each sampling interval is calculated by dividing the electricity consumption data by the total energy consumption.

6. A method for correcting and supplementing incomplete data of a smart meter according to claim 1 or 2, characterized in that: The similar day data set with the highest data similarity screened out also includes: For any similar day data set, the time period corresponding to the missing data segment is extended forward and backward to form a new similar day data set; the element similarity between any two elements in the similar day data set is calculated; The similarity value of the dataset is obtained by averaging the similarities of all elements in the similar day dataset; The dataset with the smallest similarity value is selected as the similar day dataset with the highest data similarity.

7. A method for correcting and supplementing incomplete data of a smart meter according to claim 3, characterized in that: The calculation of the element similarity between any two elements in the similar day data set includes: Where X and Y are two elements in the similar day data set, each element contains a number of electricity consumption data arranged in the order of sampling time; n represents the number of sampling moments in the time period corresponding to the data missing segment, and the subscript i represents the electricity consumption data at the i-th sampling moment.

8. A method for correcting and supplementing incomplete data of a smart meter according to claim 6, characterized in that: The calculation of the element similarity between any two elements in the similar day data set includes: Where X and Y are two elements in the similar day data set, each element contains several electricity consumption data arranged in the order of sampling time; α i is the weight coefficient, α i ∈[0,1); m represents the number of sampling moments in the extended time period; the subscript i represents the electricity consumption data at the i-th sampling moment, and i∈[p,q] represents the sampling moment in the time period corresponding to the data missing segment.

9. A system for correcting and supplementing incomplete data of a smart meter, applicable to the correction and supplementing method according to any one of claims 1 to 8, characterized in that: include: Historical database, which stores complete historical electricity consumption data; The similar day data set generation module selects the complete data of the same time period according to the time period corresponding to the data missing segment of the data missing day, and forms several similar day data sets; The matching module selects the similar day data sets with the highest data similarity; The data filling module fills the missing data segments according to the data range corresponding to the missing data segments and the data characteristics of similar daily data sets.

10. A system for correcting and supplementing incomplete data of a smart meter according to claim 9, characterized in that: include: A data missing detection module, which detects the data missing segments and corresponding time periods of the sampled data; Timing module, set the interval time for data missing detection and the interval time for data filling.

Citation Information

Cited By

  • Data calibration system and method based on electric power acquisition terminal

    CN119375809A

  • Prepayment early warning method and system for intelligent electric meter

    CN119904945A

  • Missing load recovery method based on similar block embedding and first-order polynomial interpolation

    CN120011148A

  • Multi-source heterogeneous data fusion-based power utilization data processing method for power transfer and supply main body

    CN121144296A