Data set extraction method and system for electric energy metering acquisition equipment evaluation model

Abstract

The invention discloses a data set extraction method and system for an electric energy metering and collecting equipment evaluation model, and relates to the field of big data driven electric energy collecting equipment health state evaluation, and the method comprises the steps: retrieving an equipment identification number in a database according to an equipment installation date and a user type; original time sequence data such as three-phase voltage, current, power, power factors and active electric quantity are extracted; and key parameters such as voltage, current fluctuation ratio, current reverse polarity and electric quantity difference rate are calculated based on the cleaned data, and forward active increment and total reverse active data are expanded. And aligning and combining all data according to a time sequence, and performing slicing processing and label endowing based on a health state evaluation rule. According to the method, the problems of data acquisition time sequence splitting, low data quality and no label are solved, the data quality is improved through a scientific index system, a structured data set efficiently supporting deep learning model training is formed, and intelligent evaluation of the health state of the electric energy metering acquisition equipment is assisted.

Description

Technical Field

[0001] The present invention relates to the field of big data-driven health status assessment of power acquisition equipment, and specifically to a method and system for extracting a data set for an evaluation model of power metering and acquisition equipment. Background Art

[0002] With the continuous advancement of the power system towards intelligent automation, traditional evaluation methods are difficult to meet the requirements of modern power systems for real-time monitoring, accurate evaluation, and efficient management of equipment. The continuous development and improvement of big data analysis technologies, such as algorithms like machine learning, deep learning, and association rule mining, can process and analyze massive, multi-source, and heterogeneous data, mine equipment fault characteristics, performance degradation laws, etc., to achieve accurate evaluation and prediction of the health status of equipment, help enterprises formulate reasonable maintenance plans, optimize equipment configuration, and improve asset utilization rate, thereby realizing the transformation from traditional extensive management to refined management.

[0003] Since the power consumption data records of power collection equipment and systems for each user group are complex and large in quantity, some of the collected data are fragmented in time series and are all unlabeled data. Machine learning and deep learning models require high-quality, data-structured training data sets. Using the original collected data for simple processing as the model training data set will cause the model to be difficult to learn the desired mapping relationship and deep-level features of the data, thereby reducing the performance of the model. Therefore, the original collection equipment needs to be scientifically analyzed and transformed in a patterned manner into an effective data training set for the health status evaluation model of power collection equipment. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: how to efficiently extract, clean, calculate features, and expand a large amount of original time-series data of power metering and acquisition equipment to generate a high-quality training data set, so as to accurately evaluate the health status of power metering and acquisition equipment and improve the accuracy and reliability of equipment fault diagnosis and status prediction.

[0006] To solve the above technical problem, the present invention provides the following technical solution: A method for extracting a data set for an evaluation model of power metering and acquisition equipment, which includes the following steps,

[0007] Classify and retrieve the identification numbers of metering and acquisition equipment in the database storing the detailed information of power metering and acquisition equipment and the detailed information of users according to the installation date of the power metering and acquisition equipment and the type of the affiliated user;

[0008] In the database storing the identification numbers of the metering and acquisition devices, retrieve the original time-series data corresponding to the device identification number for three-phase voltage, current, power, total power, power factor, starting meter reading for forward active power, and total forward and reverse active power in the tables, and export it to a working table according to the retrieved identification number of the metering and acquisition device.

[0009] Preprocess the data collected by the exported metering and acquisition devices.

[0010] Based on the preprocessed data, calculate the parameters of three-phase voltage and current waveform rate, current reverse polarity, and power difference rate, and perform data expansion for power difference rate, starting meter reading increment form of forward active power, and total reverse active power.

[0011] Align and combine the original time-series data, calculated parameters, and expanded data in time series.

[0012] Slice the combined data in the time dimension at a fixed time length, and assign corresponding labels according to the health status evaluation rules of the power metering and acquisition devices.

[0013] As a preferred solution of the data set extraction method for the power metering and acquisition device evaluation model described in the present invention, wherein: the classification and retrieval of the identification numbers of the metering and acquisition devices according to the installation date of the power metering and acquisition devices and the type of the affiliated users in the database storing the detailed information of the power metering and acquisition devices and the detailed information of the users includes,

[0014] According to the installation date of the metering and acquisition devices covered by the data in the database storing the detailed information of the power metering and acquisition devices and the detailed information of the users and the time interval of the historical time-series data of the metering and acquisition devices as X1 year Y1 month Z1 day - X2 year Y2 month Z2 day, retrieve the identification numbers of all metering and acquisition devices with the installation date on or before X1 year Y1 month Z1 day in the table recording the detailed information of the devices in the database.

[0015] In the database, query the corresponding information for the retrieved device identification numbers in the table recording the detailed information of the devices and the table recording the information of the users to which the devices belong, classify the metering and acquisition devices of low-voltage users and special transformer users according to the type of the users to which the devices belong, and classify the device identification numbers of the devices installed within XX month of XXXX year according to the installation date time, and store the comprehensive magnification values of the metering and acquisition devices in the table recording the detailed information of the devices.

[0016] As a preferred solution of the dataset extraction method for the power metering and acquisition device evaluation model of the present invention, wherein: retrieving the original time-series data corresponding to the device identification number in the database storing the user's electricity consumption situation and recording three-phase voltage, current, power, total power, power factor, forward active starting meter reading, and total forward and reverse active meter readings and exporting them to a working table includes,

[0017] recording the three-phase voltage V in the database with the device identification number retrieved by classification a 、V b 、V c , the three-phase current I a 、I b 、I c tables, recording the three-phase power P a 、P b 、P c 、total power P Z 、power factor tables, recording the total forward active Q Z 、reverse active Q F 、forward active starting meter reading Code Z in the tables to perform a combined query on the historical time-series data corresponding to the device identification number;

[0018] During the combined query, if the time-series data corresponding to the device identification number cannot be found in any of the tables, the device identification number is discarded. Conversely, the retrieved data is sorted in ascending order of time and aligned and combined in the time dimension;

[0019] Export the combined data to a working table and name the current working table with the power metering and acquisition device identification number.

[0020] As a preferred solution of the dataset extraction method for the power metering and acquisition device evaluation model of the present invention, wherein: the preprocessing of the exported power metering and acquisition device acquisition data includes,

[0021] In the exported historical time-series data, the three-phase voltage V a 、V b 、V c , the three-phase current I a 、I b 、I c , recording the three-phase power P a 、P b 、P c , total power P Z , forward active starting meter reading Code ZMultiply the values at all times by the recorded comprehensive magnification value to overwrite the original data, and save the working table in the named folder path corresponding to the user type and installation date;

[0022] Perform singular value checks on each data column except the time column in the processed working table. If the three-phase voltage data V a 、V b 、V c The corresponding values at time t exceed the set threshold multiple coefficient of the current column average value then it is regarded as a singular value, and the average value of the current column data is used to replace the singular value in the data;

[0023] Perform time integrity checks on the time data column in the exported working table;

[0024] If the data column with a daily collection frequency of N is not continuous in time and the data columns missing within the set n moments are filled with 0 values. Samples that exceed the set moment quantity n moments and are within m months are not filled, and if it exceeds m months, the collected data of the electricity metering collection device is discarded;

[0025] If the data column recorded once a day is not continuous in time and the missing data is within p days. If the missing data is the total positive and negative active power value, it is filled with 0 value. If the missing data is the starting meter code of the positive active power, it is filled using the linear interpolation rule;

[0026] The filling rule is:

[0027] The starting meter code value of the positive active power on the day before the missing data is The starting meter code value of the positive active power on the day after is The number of missing days is P, and the starting meter code value of the positive active power on the i-th day of the missing data, the expression is:

[0028]

[0029] If the missing data exceeds p days and is within m months, no interpolation is performed. If it exceeds m months, the collected data of the electricity metering collection device is discarded;

[0030] Subtract the value in the first row from all the values in the starting meter code data of the positive active power in the exported data to convert it into the form of relative increment. The expression is:

[0031]

[0032] Among them, is the increment form of the starting meter code of the positive active power, is the value of all rows of the starting meter code of the positive active power, It is the value of the first row of the starting meter reading data for positive active power.

[0033] As a preferred solution of the dataset extraction method for the power metering and acquisition equipment evaluation model described in the present invention, wherein: calculating the three-phase voltage and current waveform rates, current reverse polarity, and power difference rate parameters based on the preprocessed data, and the data expansion of the power difference rate, the incremental form of the starting meter reading for positive active power, and the total reverse active power includes

[0034] Calculating the volatility of the three-phase voltage and current from the preprocessed data. The calculation expression for the voltage volatility at time t is:

[0035] Voltage volatility of phase A

[0036] Voltage volatility of phase B

[0037] Voltage volatility of phase C

[0038] Wherein, the voltage value at time t is The average voltage of the T-th day to which time t belongs is They are the voltage volatilities of phase A, phase B, and phase C respectively;

[0039] The calculation expression for the current volatility at time t is:

[0040] Current volatility of phase A

[0041] Current volatility of phase B

[0042] Current volatility of phase C

[0043] Wherein, the current value at time t is The average current of the T-th day to which time t belongs is They are the current volatilities of phase A, phase B, and phase C respectively;

[0044] The calculation expression for the current reverse polarity from the preprocessed data is:

[0045]

[0046] Wherein, the three-phase power at time t is The total power is

[0047] Data expansion is performed on the data at the non-zero point time of the starting meter reading for positive active power in the preprocessed data, and the expression is:

[0048]

[0049] Among them, is the starting meter reading of positive active power at 0 o'clock, is the total power at the i-th moment of the T-th day;

[0050] For the preprocessed data, calculate the daily power difference rate value and expand the daily power difference data. The expression of the power difference rate on the T-th day is:

[0051]

[0052] The expression of the power difference rate at the i-th moment of the T-th day is:

[0053] D ti = D T ÷N

[0054] Among them, the total positive active power of a certain day is The total power at the i-th moment of the T-th day is N is the data acquisition frequency of one day;

[0055] For the preprocessed data, perform total reverse active power data expansion. The expression is:

[0056]

[0057] Among them, the total reverse active power value on the T-th day is

[0058] As a preferred scheme of the method for extracting a data set of the evaluation model for electric energy metering and acquisition equipment according to the present invention, wherein: the alignment and combination of the original time series data, calculation parameters, and expanded data in time series includes,

[0059] Align and combine the preprocessed data, three-phase voltage, current waveform rate, current reverse polarity, calculated values of power difference rate parameters, and power difference rate, the incremental form of the starting meter reading of positive active power, and the expanded value of total reverse active power by each hour to form a data column combination. The order of the combined data columns is three-phase voltage, current, power, total power, power factor, voltage and current volatility, interpolation expansion of the starting meter reading of positive active power, current reverse polarity, data expansion of power difference rate, and data expansion of total reverse active power.

[0060] As a preferred scheme of the method for extracting a data set of the evaluation model for electric energy metering and acquisition equipment according to the present invention, wherein: the slicing process of the combined data in the time dimension according to a fixed time length and assigning corresponding labels according to the evaluation rules of the health state of the electric energy metering and acquisition equipment includes,

[0061] Slice the combined data columns in the time dimension with a specified time length of X days. The sampling frequency of the metering acquisition device per day is N. After slicing, the data segment dimension is X×N rows and Y columns;

[0062] Find the corresponding time of the acquisition device identification number corresponding to the sliced data in the database storing device fault work orders, the number of fault records of the acquisition device in XX month of XXXX year, and assign labels to the corresponding health status stages of the metering acquisition device in combination with the detailed information of the metering acquisition device and the data conditions of each feature column in the current sliced data.

[0063] The purpose of the present invention is to provide a training dataset extraction system for evaluating power metering acquisition devices, which can realize the rapid generation of training datasets through efficient retrieval, cleaning and preprocessing, feature parameter calculation and expansion, data alignment and combination, and labeling of the original time series data of power metering acquisition devices, and solves the problems of scattered data, low processing efficiency, insufficient feature parameters, and unlabeled data in the prior art.

[0064] To solve the above technical problems, the present invention provides the following technical solutions: A training dataset extraction system for evaluating power metering acquisition devices, including: a data extraction and preliminary classification module, a data preprocessing module, a feature parameter calculation and data expansion module, a data combination and health status label assignment module;

[0065] The data extraction and preliminary classification module classifies and retrieves in the database storing the detailed information of power metering acquisition devices and user information according to the installation date and user type, extracts the device identification numbers that meet the conditions, classifies and stores the device numbers according to the user type and installation date, uses the classified device identification numbers to jointly query the time series data storing user power consumption data, arranges them in ascending order of time, aligns and combines them, exports the results to a working table, and names it with the device identification number;

[0066] The data preprocessing module cleans and preliminarily processes the exported original data;

[0067] The feature parameter calculation and data expansion calculates the three-phase voltage and current volatility, expands the form of the forward active starting meter code increment, expands the total reverse active power, and calculates the daily power difference rate;

[0068] The data combination and health status label assignment module aligns and combines the preprocessed data, calculated parameters, and expanded data in time series, and assigns health status labels to the sliced data according to the device fault work order records and device health status evaluation rules, in combination with the numerical distribution of the feature column data and the historical number of faults.

[0069] A computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the data set extraction method for the power metering acquisition device evaluation model described above are implemented.

[0070] A computer-readable storage medium stores a computer program thereon. When the computer program is executed by a processor, the steps of the data set extraction method for the power metering acquisition device evaluation model described above are implemented.

[0071] Advantages of the present invention: The present invention converts the temporal fragmentation and non-uniformity of some data in the collected data into data with a unified time series, increases electrical index parameters related to the healthy life of the power acquisition device on the basis of the original data to improve data quality, which is beneficial for the deep learning model to learn the internal feature correlation relationship, and solves the problem of unlabeled original data by formulating scientific and reasonable rules for the health status evaluation system of the power metering acquisition device and labeling the structured data. Thus, it realizes the conversion of complex unlabeled original collected time series data into a training data set for the deep learning model for evaluating the health status of the power metering acquisition device. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. Among them:

[0073] Figure 1 It is the overall flowchart of the data set extraction method for the power metering acquisition device evaluation model provided by the first embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0074] To make the above objects, features, and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are some embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0075] Example 1, referring to Figure 1 An embodiment of the present invention provides a data set extraction method for a power metering acquisition device evaluation model, including:

[0076] Taking the data collected by the metering and acquisition devices of some users in a certain county bureau from June 1, 2023 to May 31, 2024 as an example, this method will be illustrated. In this example, the historical acquisition data of a special transformer user is selected to explain the data set extraction method. The specific operation method is to use Python software programming to operate the database and work sheets according to the above step logic to implement an automated program to extract and produce the data set.

[0077] S1. In the database storing the detailed information of electric energy metering and acquisition devices and user detailed information, retrieve the identification numbers of metering and acquisition devices by classifying according to the installation date of the electric energy metering and acquisition devices and the user types they belong to;

[0078] S11. According to the installation date of the metering and acquisition devices covered by the data in the database storing the detailed information of electric energy metering and acquisition devices and user detailed information and the historical time series data time range of the metering and acquisition devices from June 1, 2023 to May 31, 2024, retrieve in the database the identification numbers of all metering and acquisition devices with installation dates on or before May 31, 2023 in the table recording the device detailed information. The earliest installation date of the retrieved metering and acquisition devices can be traced back to June 1, 2015;

[0079] S12. Query the corresponding information in the table recording the device detailed information and the table recording the user information to which the device belongs in the database with the retrieved device identification numbers, and based on this, classify the metering and acquisition devices of special transformer users according to the user types they belong to, and classify the devices installed within February 2019 according to the installation date time. And store and record the comprehensive magnification value B = 40.0 of the metering and acquisition devices in the table recording the device detailed information;

[0080] S2. According to the metering and acquisition device identification numbers retrieved in S1, query the original time series data of the corresponding device identification numbers in the tables recording three-phase voltage, current, power, total power, power factor, starting meter reading of forward active power, and total positive and reverse active power in the database storing the user electricity consumption situation and export them to the work sheet;

[0081] S21. Use the classified and retrieved device identification numbers to record three-phase voltage V a 、V b 、V c in the tables of the database, record three-phase current I a 、I b 、I c of the tables, record three-phase power P a 、P b 、P c 、total power P Z 、power factor of the tables, record total forward active power Q Z 、reverse active power QF , the starting code of active power in the forward direction Code Z Perform a joint query in the table to obtain the historical time-series data corresponding to the device identification number where t = 0, 1, 2,..., 23, and a, b, c are the codes for the three-phase lines;

[0082] S22. During the joint query in S21, if the time-series data corresponding to the device identification number cannot be found in any of the tables, the device identification number is abandoned and no subsequent operations are performed. For all the data obtained by querying the device identification number in S21, the query results are sorted in ascending order of time and aligned and combined in the time dimension;

[0083] S23. Export the data processed in step S22 to a working table, and name the working table with the metering acquisition device identification number 4000100530348336;

[0084] S3. Perform preprocessing operations on the exported data of the metering acquisition device, including selecting data groups with complete dimensions, trimming singular values, filling a small number of missing values, converting the comprehensive magnification, and converting the starting code data of active power in the forward direction into an incremental form;

[0085] S31. For the three-phase voltages V a , V b , V c , the three-phase currents I a , I b , I c , record the three-phase powers P a , P b , P c , the total power P Z , and the starting code of active power in the forward direction Code Z , multiply the values at all times by the comprehensive magnification value B = 40.0 recorded in S12 and overwrite the original data, and then save the working table in the folder path named according to the corresponding user type and installation date: "Special transformer user\Electric energy metering acquisition device installed in February 2019\4000100530348336";

[0086] S32. Check the singular values of each data column except the time column in the working table processed in S31. If the corresponding values of the three-phase voltage data V a , V b , V c at time t exceed the set threshold multiple coefficient c of the average value of this column , they are regarded as singular values, and the average value of this column data Replace the singular value in the data. In the exported work sheet data, there is a singular value "290.4" in the row of 9:00:00 on June 1, 2023 in the "CDY" column of the C-phase voltage. Replace the singular value in the C-phase voltage data column with the average value "241.9" of this data column;

[0087] S33. Check the time integrity of the time data column in the exported work sheet: If the data column with a daily collection frequency of N is not continuous in time and the data columns missing within the set n moments in the middle are filled with 0 values, samples that exceed the set number of moments n and are within m months are not filled, and if it exceeds m months, the collection data of this power metering collection device is discarded. In the exported work sheet data, the data at the moments of 19:00:00 on June 1, 2023 and 20:00:00 on June 1, 2023 are missing. Fill the discontinuous three-phase voltage data column with the average value on June 1, 2014 of the Tth day, and fill the three-phase current, power, and total power by interpolation;

[0088] If the data column with a daily record once is not continuous in time and the missing data is within p days in the middle, if the missing data is the total positive and negative active power value, it is filled with 0 value, and if the missing data is the starting meter code of the positive active power, it is filled with the linear interpolation rule. The interpolation rule is:

[0089] The starting meter code value of the positive active power on the day before the missing data is The starting meter code value of the positive active power on the day after is The number of missing days is P, and the starting meter code value of the positive active power on the ith day of the missing data If the missing data exceeds p days and is within m months, no interpolation is performed. If it exceeds m months, the collection data of this power metering collection device is discarded. In the exported data, there is no discontinuity in time in the data column with a daily record once, so directly proceed to the subsequent steps;

[0090] S34. Subtract the value in the first row from the value in the starting meter code data of the positive active power in the exported data, Convert it into the form of relative increment

[0091] S4. Calculate the three-phase voltage, current waveform rate, current reverse polarity, and power difference rate parameters for the data processed according to S3, and perform data expansion for the power difference rate, the incremental form of the starting meter code of the positive active power, and the total reverse active power;

[0092] S41. Calculate the three-phase voltage and current calculation volatility for the data processed by S3. The calculation rules are as follows:

[0093] The voltage value at time t is The average value of the voltage on the Tth day to which this t moment belongs is Then the calculation formula for the voltage volatility at time t is: the voltage volatility of phase A The voltage volatility of phase B The voltage volatility of phase C The current value at time t is The average value of the current on the T-th day to which time t belongs is Then the calculation formula for the current volatility at time t is the current volatility of phase A The current volatility of phase B The current volatility of phase C

[0094] S42. Calculate the current reverse polarity for the data after S3 processing, and the calculation rules are as follows:

[0095] The three-phase power at time t is The total power is Calculate the current reverse polarity at time t

[0096] S43. Perform data expansion on the data at the non-zero point time of the forward active starting meter reading in the data after S3 processing, and the expansion rules are as follows:

[0097] The forward active starting meter reading at 0:00 on a certain day is The total power at the i-th moment on the T-th day is Then the forward active starting meter reading at the i-th point on the T-th day is

[0098] S44. Calculate the daily power difference rate value and the data expansion of the daily power difference data for the data after S3 processing, and the calculation and expansion rules are as follows:

[0099] The total forward active power on a certain day is The total power at the i-th moment on the T-th day is If the data acquisition frequency is N = 24 times a day, then the power difference rate on the T-th day is The power difference rate D at the i-th moment on the T-th day ti = D T ÷ N;

[0100] S45. Perform total reverse active power data expansion on the data after S3 processing, and the expansion rules are as follows:

[0101] The total reverse active power value on the T-th day of T is If the data acquisition frequency is N = 24 times a day, then the total reverse active power value at the i-th moment on the T-th day

[0102] S5. Align the data processed in step S3 and the data processed in step S4 hour by hour for data column combination. The combined feature dimensions are three-phase voltage, current, power, total power, power factor, voltage and current volatility, interpolation expansion of the starting meter reading of forward active power, current reverse polarity, data expansion of the electricity quantity difference rate, and data expansion of the total reverse active power;

[0103] S6. Perform slicing processing on the data combined in step S5 in the time dimension with a specified time length of X = 30 days. The daily sampling frequency of the metering acquisition device is N = 24. The dimension of the sliced data segment is X × N rows and Y columns (Y is the number of data columns). The dimension of the sliced data segment is 720 rows and 21 columns;

[0104] S7. Find the corresponding time in the database storing device fault work orders for the acquisition device identification number corresponding to the sliced data (belonging to February 2019) processed in step S6: The number of fault records G = 0 for this acquisition device in February 2019. Combine the detailed information of the metering acquisition device and the data conditions of each feature column in this sliced data to assign labels to the health status stage of the metering acquisition device. The specific rules for assigning health status labels are as follows:

[0105] The installation date of the metering acquisition device to which the sliced data segment belongs is within G year H month (February 2019). The number of fault work order records queried within the time range J year K month (June 2023) of the data segment is G = 0. The daily data acquisition frequency of the metering acquisition device is N = 24 times, and the time length covered by the sliced data is X = 30;

[0106] The average value of the three-phase voltage volatility in the sliced data segment is The average value of the current reverse polarity R t' is The power factor The average value The average value of the electricity quantity difference rate The average value of the three-phase voltage volatility is The average value of the current reverse polarity The average value of the power factor The average value of the electricity quantity difference rate where t' = 1, 2, 3,......, 719, 720;

[0107] Then the estimation formula for the estimated remaining life (L) of the metering acquisition device:

[0108]

[0109] L0 = 96 - [(J - G) × 12 + (K - H)]

[0110] Among them, L0 is the original remaining life, α and β are weight coefficients, and K1, K2, K3, K4, K5, a, b, c, and d are life degradation coefficients, which can be adjusted according to the actual situation. The reference values of the weights and degradation coefficients are as follows:

[0111] Table 1: Reference values of coefficients α and β

[0112]

[0113] Table 2: Reference values of coefficients K1 and a

[0114]

[0115] Table 3: Reference values of coefficients K2 and b

[0116]

[0117]

[0118] Table 4: Reference values of coefficients K3 and c

[0119]

[0120] Table 5: Reference values of coefficients K4 and d

[0121]

[0122] Table 6: Reference value of coefficient K5

[0123] Fault work order record G K5 0 K5＝0 ≤2 K5＝0.7 ≤4 K4＝1.8 ＞4 K4＝2.5

[0124] Among them, the original remaining life L0 is:

[0125] L0 = 96 - [(J - G) × 12 + (K - H)] = 96 - [(2023 - 2019) × 12 + (6 - 2)] = 44 months

[0126] Judging from the power consumption, the metering and acquisition device belongs to the medium-load operation condition. According to the reference value table of weights, α = 0.3 and β = 0.7. According to the reference value table of degradation coefficients, K1 = 0.01, K2 = 1.0, K3 = 2.0, K4 = 0.1, K5 = 0, a = 1.0, b = 2.0, c = 1.8, d = 1.3. Estimate the remaining life L of the metering and acquisition device:

[0127]

[0128] Therefore, the estimated remaining life of the metering and acquisition device with the identification number 4000100530348336 is 41.78 months. Therefore, the label of this data segment is: 41.78.

[0129] In the automated extraction program in the implementation example, the identification of the electric energy metering and collection device is in batches by year. The program retrieves the device identification numbers belonging to different batches in parallel, and the device identification numbers in each batch are sequentially executed for the above steps S1-S7. When the computer hardware permits, the device collection historical data of each year can be extracted in parallel, shortening the extraction time.

[0130] Embodiment 2 is an embodiment of the present invention, which provides a system for the data set extraction method of the electric energy metering and collection device evaluation model, including:

[0131] A data extraction and preliminary classification module, a data preprocessing module, a feature parameter calculation and data expansion module, and a data combination and health status label assignment module;

[0132] The data extraction and preliminary classification module classifies and retrieves in the database storing the detailed information of the electric energy metering and collection device and user information according to the installation date and user type, extracts the device identification numbers that meet the conditions, classifies and stores the device numbers according to the user type and installation date, uses the classified device identification numbers to jointly query the time series data storing the user's electricity consumption data, arranges them in ascending order of time, aligns and combines them, exports the results to a work sheet, and names it with the device identification number;

[0133] The data preprocessing module cleans and preliminarily processes the exported original data;

[0134] The feature parameter calculation and data expansion calculates the three-phase voltage and current volatility, expands the form of the initial meter reading increment of the positive active power, expands the total reverse active power, and calculates the daily electricity difference rate;

[0135] The data combination and health status label assignment module aligns and combines the preprocessed data, calculated parameters, and expanded data in time series, and assigns health status labels to the sliced data according to the device fault work order records and device health status evaluation rules, in combination with the numerical distribution of the feature column data and the historical fault times.

[0136] If the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0137] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or in conjunction with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0138] More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection part with one or more wirings (electronic device), a portable computer disk cartridge (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or otherwise processing it as appropriate, and then storing it in a computer memory.

[0139] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one of the following techniques known in the art or a combination thereof can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0140] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A data set extraction method for an electric energy metering and collection equipment evaluation model, characterized in that: include: In a database storing detailed information of electric energy metering and collecting equipment and detailed information of users, the identification number of the metering and collecting equipment is retrieved according to the installation date of the electric energy metering and collecting equipment and the type of user to which it belongs; According to the identification number of the retrieved metering and collecting equipment, in the database storing the user's electricity consumption, record the three-phase voltage, current, power, total power, power factor, forward active starting table code, total forward and reverse active power, query the original time series data of the corresponding equipment identification number, and export it to the worksheet; Preprocess the exported data collected by the measurement collection equipment; Based on the pre-processed data, the three-phase voltage and current waveform rate, current reverse polarity, and power difference rate parameters are calculated, and the power difference rate, the forward active starting code increment form, and the total reverse active data are expanded; Align and combine the original time series data, calculation parameters, and expanded data in time series; The combined data is sliced ​​according to a fixed time length in the time dimension and assigned corresponding labels according to the health status assessment rules of the electric energy metering and collection equipment.

2. The method for extracting data sets for an evaluation model of electric energy metering and collection equipment according to claim 1, characterized in that: The method of retrieving the identification number of the metering and collecting device according to the installation date of the metering and collecting device and the type of user to which the metering and collecting device belongs in the database storing the detailed information of the metering and collecting device and the detailed information of the user includes: According to the time interval of the installation date of the metering and collecting equipment and the historical time series data of the metering and collecting equipment covered by the data in the database storing the detailed information of the electric energy metering and collecting equipment and the detailed information of the user, which is X1-X2, the identification numbers of all the metering and collecting equipment with the installation date of X1-X2 and before are searched in the table recording the detailed information of the equipment in the database; The retrieved equipment identification number is stored in the database, in the table recording the equipment detailed information and the table recording the equipment user information, and the corresponding information is queried and classified into the metering and collecting equipment of low-voltage users and special transformer users according to the user type to which the equipment belongs. The equipment identification number of the equipment installed within XX month of XXXX is classified according to the installation date and time, and the comprehensive multiplier value of the metering and collecting equipment is stored in the table recording the equipment detailed information.

3. The method for extracting data sets for an evaluation model of electric energy metering and collection equipment according to claim 2, characterized in that: The identification number of the metering and collecting device retrieved is recorded in the database storing the user's electricity consumption, and the three-phase voltage, current, power, total power, power factor, forward active starting table code, total forward and reverse active table are queried for the original time series data of the corresponding device identification number and exported to the work table, including, Record the three-phase voltage V in the database using the equipment identification number retrieved by classification a 、V b 、V c , three-phase current I a ,I b ,I c Table, record the three-phase power P a , P b , P c , total power P Z , Power Factor The table records the total forward active power Q Z , Reverse active power Q F 、Forward active power start code Z Perform a joint query on the historical time series data corresponding to the device identification number in the table; If the time series data corresponding to the device identification number cannot be found in any table during the joint query, the device identification number is abandoned. Otherwise, the queried data is arranged in ascending order of time and aligned in the time dimension; Export the combined data to a worksheet and name the current worksheet with the identification number of the measurement collection equipment.

4. The method for extracting data sets for an evaluation model of electric energy metering and collection equipment according to claim 3, characterized in that: The preprocessing of the exported data collected by the metering collection device includes: The three-phase voltage V in the exported historical time series data a 、V b 、V c , three-phase current I a ,I b ,I c , record the three-phase power P a , P b , P c , total power P Z , Forward active starting code Code Z Multiply the values ​​of all moments by the recorded comprehensive multiplier value to overwrite the original data, and save the worksheet in the folder path named corresponding to the user type and installation date; After processing, the data in the worksheet is checked for singular values ​​in each data column except the time column. If the three-phase voltage data V a 、V b 、V c The corresponding value at time t Exceeds the current column average The threshold multiple coefficient is considered a singular value, and the average value of the current column data is used. Replace the singular values ​​in the data; Perform time integrity check on time data columns in exported worksheets; If the data column with a daily collection frequency of N is discontinuous in time and the data column within the set n moments is missing in the middle, it will be filled with 0 values. If the data column exceeds the set number of moments n and is within m months, it will not be filled. If it exceeds m months, the collected data of the electric energy metering collection device will be abandoned; If the data series recorded once a day is discontinuous in time and the missing days are less than p, if the missing data are the total forward and reverse active values, they are filled with 0 values; if the missing data are the forward active starting table codes, they are filled with linear interpolation rules; The filling rules are: The positive active power start table value of the previous day is missing The positive active power start value for the next day is The number of missing days is P, and the starting table code value of the positive active power on the i-th day of missing data is expressed as: If the missing data exceeds p days and is within m months, no interpolation will be performed. If it exceeds m months, the collected data of the electric energy metering collection device will be abandoned. Subtract the value of the first row from the value of the forward active power start table code in the exported data and convert it into a relative increment. The expression is: in, It is the increment form of the positive active starting code. is the value of all rows of the positive active power start table code, It is the value of the first row of the positive active power start table code data.

5. The method for extracting data sets for an evaluation model of electric energy metering and collection equipment according to claim 4, characterized in that: The calculation of the three-phase voltage, current waveform rate, current reverse polarity, and power difference rate parameters based on the pre-processed data, and the expansion of the power difference rate, the forward active starting table code increment form, and the total reverse active data include: The three-phase voltage and current fluctuation rate are calculated using the preprocessed data. The voltage fluctuation rate calculation expression at time t is: Among them, the voltage value at time t is The average voltage of the Tth day at time t is They are phase A voltage fluctuation rate, phase B voltage fluctuation rate, and phase C voltage fluctuation rate respectively; The calculation expression of the current fluctuation rate at time t is: Among them, the current value at time t is The average value of the current on day T at time t is They are the current fluctuation rate of phase A, the current fluctuation rate of phase B, and the current fluctuation rate of phase C respectively; The current reverse polarity expression calculated from the preprocessed data is: Among them, the three-phase power at time t is The total power is The data expansion of the non-0 time data of the forward active power start table code in the pre-processed data is performed, and the expression is: in, The positive active starting code is 0. is the total power at the i-th moment on the T-th day; The daily power difference rate value and daily power difference data expansion are calculated for the preprocessed data. The power difference rate expression for the Tth day is: The expression of the power difference rate at the i-th moment on the T-th day is: D ti =D T ÷N Among them, the total positive work on a certain day is The total power at the i-th moment on the T-th day is N is the frequency of data collection per day; The total reverse active data expansion is performed on the preprocessed data, and the expression is: Among them, the total reverse active value on day T is 6. The method for extracting data sets for an evaluation model of electric energy metering and collection equipment according to claim 5, characterized in that: The aligning and combining the original time series data, the calculation parameters and the expanded data in time series includes: The preprocessed data, three-phase voltage, current waveform rate, current reversal polarity, calculated values ​​of electric quantity difference rate parameters and electric quantity difference rate, forward active power starting code increment form and total reverse active power data expansion value are aligned and combined into data columns every hour. The order of the combined data columns is three-phase voltage, current, power, total power, power factor, voltage and current fluctuation rate, forward active power starting code interpolation expansion, current reversal polarity, electric quantity difference rate data expansion and total reverse active power data expansion.

7. The method for extracting data sets for an evaluation model of electric energy metering and collection equipment according to claim 6, characterized in that: The combined data is sliced ​​according to a fixed time length in the time dimension, and the corresponding labels are assigned according to the health status assessment rules of the electric energy metering and collection equipment, including: The combined data column is sliced ​​according to the specified time length of X days in the time dimension. The sampling frequency of the measurement collection equipment is N every day. The dimension of the data segment after slicing is X×N rows and Y columns. The corresponding time of the collection equipment identification number corresponding to the slice data is found in the database storing the equipment fault work order, and the number of fault records of the collection equipment in XX month of XXXX year is combined with the detailed information of the metering collection equipment and the data conditions of each feature column in the current slice data to assign a label of the corresponding health status stage of the metering collection equipment.

8. A system using the data set extraction method for an electric energy metering and collection equipment evaluation model as claimed in any one of claims 1 to 7, characterized in that: Packet data extraction and preliminary classification module, data preprocessing module, feature parameter calculation and data expansion module, data combination and health status label assignment module; The data extraction and preliminary classification module is in the database storing the detailed information of the electric energy metering and collection equipment and the user information, and performs classified retrieval according to the installation date and user type, extracts the equipment identification number that meets the conditions, and classifies and stores the equipment number according to the user type and installation date, and uses the classified equipment identification number to jointly query the time series data storing the user's electricity consumption data, arranges it in ascending order of time, aligns and combines it, and exports the result to the work table, and names it with the equipment identification number; The data preprocessing module cleans and preliminarily processes the exported raw data; The characteristic parameter calculation and data expansion are to calculate the three-phase voltage and current fluctuation rate, expand the forward active power starting table code increment form, expand the total reverse active power and calculate the daily power difference rate; The data combination and health status label assignment module aligns and combines the preprocessed data, calculation parameters and expanded data in time series, and assigns health status labels to the sliced ​​data based on the equipment failure work order records and equipment health status assessment rules, combined with the numerical distribution of the feature column data and the number of historical failures.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the data set extraction method for the electric energy metering and collection equipment evaluation model described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the data set extraction method for an electric energy metering and collection equipment evaluation model described in any one of claims 1 to 7 are implemented.

Images