A method for accurately locating users with abnormal electricity consumption or electricity theft in a power system

Abstract

A method for accurately locating users with abnormal electricity consumption or electricity theft in an electric power system, wherein each meter box is equipped with an intelligent measuring switch to obtain the electricity consumption of the meter box; the difference between the electricity consumption of the intelligent measuring switch and the electricity consumption of all sub-meters is used as the electricity loss to determine abnormal electricity consumption or electricity theft; the regression coefficient of each sub-meter is calculated through a multivariate linear regression model, and the abnormal electricity consumption or electricity theft status of each sub-meter is identified based on the comparison between the regression coefficient and the set threshold. The method of the present invention can accurately identify users with abnormal electricity consumption or electricity theft, and the judgment accuracy rate reaches more than 98%.

Description

Technical Field

[0001] The present invention belongs to the technical field of electricity theft detection in a smart grid, and in particular relates to a method for accurately locating users with abnormal electricity consumption or electricity theft in a power system. Background Art

[0002] Electricity theft will cause serious losses to the national economy and seriously affect the safe operation of the power grid. It is also an illegal act. However, electricity theft cannot be eliminated. An efficient and accurate electricity theft detection method can effectively and timely prevent the various hazards caused by electricity theft.

[0003] Most previous electricity theft detection algorithms used big data analysis to identify outliers and identify suspected electricity theft users, which were then confirmed on-site by staff. Although effective, they were time-consuming, inaccurate in judgment, and required a high level of manual involvement. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the above-mentioned scheme, such as poor timeliness of electricity theft analysis and judgment, insufficient accuracy of judgment, and high degree of manual participation, and to provide a method for efficiently and accurately locating users with abnormal electricity consumption or electricity theft in a power system.

[0005] The technical solution of the present invention is:

[0006] The present invention provides a method for accurately locating users with abnormal electricity consumption or electricity theft in an electric power system. Each meter box is equipped with an intelligent measuring switch to obtain the electricity consumption of the meter box; the difference between the electricity consumption of the intelligent measuring switch and the electricity consumption of all sub-meters is used as the electricity loss to determine the abnormal electricity consumption or electricity theft; the regression coefficient of each sub-meter is calculated through a multivariate linear regression model, and the abnormal electricity consumption or electricity theft status of each sub-meter is identified by comparing the regression coefficient with a set threshold.

[0007] Further, the method comprises the following steps:

[0008] S1. Configure an intelligent measurement switch at the external power supply end of the meter box, set the error threshold Tve, the abnormal threshold TVa of the abnormal power consumption identification event of the electric energy meter, and the theft threshold TVs of the power theft identification event of the electric energy meter; the intelligent measurement switch communicates with each sub-meter, and establishes a meter library according to the communication address and sub-meter information of each sub-meter;

[0009] S2. Let the number of sub-meters be M. The intelligent measuring switch collects the power consumption of each sub-meter in the meter library and its own power at a fixed time. The power collected from M sub-meters once constitutes a sampling point. When the number of sampling points N is twice the number of sub-meters M, the power theft analysis is performed:

[0010] S3, calculate the power loss according to the power value of the intelligent measuring switch at the Nth sampling point on the day and the sum of the power consumption of each sub-meter on the day, and obtain the total error according to the power loss. If the total error is greater than the set error threshold Tve, it is determined that there is an abnormal power consumption or power theft user in the meter box, and go to S4, otherwise, return to S2 to continue sampling;

[0011] S4, establish a multiple linear regression model, remove the sub-table data with abnormal data in N sampling points, the number of valid sub-tables after removing the abnormal data is M', substitute the M' sub-table sampling data of N sampling points into the multiple linear regression model, and obtain a multiple linear regression equation group consisting of N linear equations;

[0012] The multivariate linear regression algorithm based on the least squares method is used to calculate and obtain a set of regression coefficients β i , as the abnormal or electricity theft coefficient of the sub-table, i represents the sub-table number, i∈[1,M'].

[0013] S5, the regression coefficients β are sequentially i Each sub-table corresponds to each other, and each regression coefficient is compared with the set abnormal threshold TVa and electricity theft threshold TVs respectively:

[0014] If the regression coefficient β i If it is less than the abnormal threshold TVa, it is judged as a normal user;

[0015] If the regression coefficient β i If the value is greater than or equal to the abnormal threshold TVa and less than the electricity theft threshold TVs, the user is judged as an abnormal electricity user;

[0016] If the regression coefficient β i If the user's value is greater than or equal to the electricity theft threshold TVs, the user is considered to be an electricity theft user.

[0017] The intelligent measurement switch obtains the judgment information of each sub-meter, records the corresponding events and reports them to the main station system through the collector.

[0018] Furthermore, the establishment of the meter library in S1 is specifically as follows: the intelligent measuring switch automatically searches the meter after power is turned on, that is, sends a communication address request frame to each sub-meter of the meter box, receives the communication address of each sub-meter, and forms a meter library with the communication address and sub-meter information of each sub-meter.

[0019] Furthermore, the intelligent measuring switch in S2 collects the electric quantity of each sub-meter at a fixed time, removes invalid data and saves it, wherein the invalid data refers to data of unsuccessful meter reading or abnormal communication.

[0020] Furthermore, in S3, the power loss P is calculated using the following formula: L :

[0021]

[0022] Among them, P G is the current power value of the smart measuring switch, P i is the daily power consumption value of each sub-meter in the meter box, M is the total number of sub-meters in the meter box, and i represents the sub-meter number;

[0023] Calculate the total error based on power loss: P L / P G .

[0024] Furthermore, in S4, if the power consumption data of a certain sub-table is always 0, the data corresponding to the sub-table is deleted.

[0025] Furthermore, in S4, the multivariate linear regression model is as follows:

[0026]

[0027] Where: t represents the time, i.e. the sampling point number, t∈[1, N], y error (t) represents the power loss at time t, that is, the difference between the power consumption of the intelligent measuring switch and the sum of the power consumption of all sub-meters at time t, δ represents the error term, i represents the sub-meter number, i∈[1, M'], M' represents the number of valid sub-meters after eliminating abnormalities, β i Indicates the abnormal or power theft coefficient of the subtable numbered i, x i (t) represents the power consumption of the submeter numbered i;

[0028] Substitute the data of M' subtables of N sampling points into the multivariate linear regression model, and obtain the N linear regression equations as follows:

[0029]

[0030] Where: x tM′ β M′ represents the loss contributed by the M'th block sub-table at the tth sampling point;

[0031] Call the multivariate linear regression algorithm regress to calculate the linear regression coefficient β 1 ,β 2 ,…,β M′ , as the abnormality or electricity theft coefficient of each sub-meter.

[0032] Furthermore, the error threshold Tve<abnormal threshold TVa<electricity theft threshold TVs.

[0033] Furthermore, the error threshold Tve<abnormal threshold TVa<electricity theft threshold TVs; the electricity theft threshold TVs is 10%.

[0034] Beneficial effects of the present invention:

[0035] The present invention provides a method for efficiently and accurately locating users with abnormal electricity usage or electricity theft in an electric power system; by calculating the line loss in the meter box, analyzing whether there are users with abnormal electricity usage or electricity theft, and then calculating the corresponding error coefficient of each meter through multivariate linear regression analysis, and comparing them with the set threshold one by one, accurately identifying users with abnormal electricity usage or electricity theft, and the judgment accuracy rate reaches more than 98%.

[0036] When the method of the present invention is applied, the intelligent measuring switch automatically searches the meter after the system is powered on, and no external equipment such as a master station or a host computer is required to perform file configuration work. The intelligent measuring switch automatically searches the meter, accurately obtains all the meter information in the meter box, and forms a meter library for subsequent collection of meter data. The entire process is automatically executed without external interference, achieving true intelligent automation.

[0037] Other features and advantages of the present invention will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The above and other objects, features and advantages of the present invention will become more apparent through a more detailed description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, wherein like reference numerals generally represent like components throughout the exemplary embodiments of the present invention.

[0039] Figure 1 A schematic diagram of an application scenario of the present invention is shown.

[0040] Figure 2 The flowchart of electricity theft analysis of the present invention is shown. DETAILED DESCRIPTION

[0041] The preferred embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the preferred embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0042] The method of the present invention accurately and efficiently locates the electricity stealing user in the meter box of the low-voltage station area. After power-on, the intelligent measuring switch actively searches the meter and reads the meter. The power consumption of the electric energy meter and its own power are recorded regularly, and records are saved. After the sampling requirements are met, the power consumption of the intelligent measuring switch and the power consumption of all sub-meters and the calculation error are calculated. After comparing and analyzing with the set threshold, it is preliminarily judged whether there is an abnormality or electricity stealing user, and then the regression coefficient is calculated through multivariate linear regression analysis, the error of each meter is calculated, and then compared with the set threshold one by one, the abnormal power consumption or electricity stealing users are accurately identified, and the judgment accuracy rate reaches more than 98%.

[0043] First, for power theft or abnormal power consumption behind the meter, the metered value and the true value approximately satisfy the proportional relationship. Therefore, given the readings x(t) and y(t) of the sub-meter and the measuring switch in the meter box, the two satisfy the following equation:

[0044]

[0045] Among them: δ represents the error term, β i represents the abnormality or electricity theft coefficient of the sub-meter, M represents the number of sub-meters in the meter box, and the linear regression model formula is obtained by subtracting the sum of the sub-meters on both sides of the equation:

[0046]

[0047] Therefore, solving this problem can be transformed into solving the problem of linear regression. To solve such problems, a sufficient sample size is required, which is at least twice the number of independent variables, i.e., the number of meters.

[0048] Substitute the collected data into the linear regression model formula to obtain the multivariate linear regression equation group:

[0049]

[0050] Call the linear regression algorithm regress() function to calculate the linear regression coefficient β 1 ,β 2 ,…,β M′ , that is, the abnormality or electricity theft coefficient of each sub-meter is obtained. Then, it can be judged whether it is abnormal according to the threshold of the electricity theft coefficient.

[0051] In the method of the present invention, after the system is powered on, the intelligent measuring switch automatically searches the meter, and no external equipment such as a master station or a host computer is required to perform file configuration work. The intelligent measuring switch automatically searches the meter, and finally accurately obtains all the meter information in the meter box, and forms a meter library for subsequent collection of meter data. The entire process is automatically executed without external interference, achieving true intelligent automation.

[0052] After the meter search is completed, read the power data of each sub-meter and the power data of the intelligent measuring switch itself at regular intervals, and record and save the collected data. When the number of collection points reaches twice the number of sub-meters, that is, when the sampling points meet the calculation requirements, calculate the power loss of all sub-meters in the meter box:

[0053]

[0054] Among them, P G is the current power value of the smart measuring switch, P i is the daily power consumption value of each sub-meter in the meter box, M is the total number of sub-meters in the meter box, and i represents the sub-meter number; P LThe power loss is the power consumption of the intelligent measuring switch minus the power consumption of each sub-meter;

[0055] Calculate the total error based on power loss: P L / P G ;

[0056] If the total error is greater than or equal to the set error threshold, it is determined that there are users with abnormal electricity usage or electricity theft in the meter box, and in-depth electricity theft analysis is required.

[0057] If electricity theft analysis is required, read the frozen data at different sampling points from the storage device, and preliminarily screen the data to remove abnormal points. M' represents the number of valid sub-tables after removing abnormal points, and substitutes it into the multivariate linear regression model formula:

[0058]

[0059] Substituting into the formula, we can get the multivariate linear regression equation group, and then use the multivariate linear regression algorithm regress based on the least squares method to calculate the regression coefficient. The obtained regression coefficient is the abnormal or electricity theft coefficient, and the regression coefficient corresponds to each sub-table in order. Then, compare each regression coefficient with the set threshold one by one. If the regression coefficient is greater than the abnormal electricity meter identification event threshold and lower than the electricity theft event reporting threshold, it is determined to be an abnormal electricity user; if the regression coefficient exceeds the electricity theft event reporting threshold, it is determined to be an electricity theft user. After the intelligent measurement switch obtains the corresponding meter serial number, it records the corresponding event and reports it to the collector, which then reports it to the main station system.

[0060] In this way, the system can accurately locate abnormal users or electricity theft users without the need for manual operation, and the accuracy can reach over 98%.

[0061] When implementing:

[0062] A method for accurately locating electricity theft users based on multiple linear regression, comprising the steps of:

[0063] 1) After the intelligent measuring switch is powered on, it automatically searches the meter and sends a wildcard address request frame to each sub-meter in the meter box. According to the received return data, it accurately obtains the communication address and other relevant information of each sub-meter in the meter box and forms a meter library;

[0064] 2) According to the meter information in the meter database, the power of each sub-meter in the meter box and the power of the intelligent measuring switch itself are collected regularly and saved;

[0065] 3) According to the number of sub-meters M in the meter box, determine the number of sampling points (starting points) N. When the number of sampling points is more than twice the number of meters, the electricity theft analysis can be started;

[0066] 4) According to the power consumption of the intelligent measuring switch and the sum of the power consumption of each user's electricity meter, the total error, that is, the power loss, is calculated. The total error is obtained based on the power loss. If the total error is greater than the set error threshold, it is determined that there is an abnormal power consumption or power theft user in the meter box, and further detailed analysis is required. Otherwise, there is no need to conduct power theft analysis;

[0067] 5) Based on the judgment result of 4), if electricity theft analysis is required, read N groups of recorded data from the storage device, and filter out abnormal points in each group of data;

[0068] 6) Based on the preprocessed data obtained in 5), substituting into the multiple linear regression model formula, a multiple linear regression equation group consisting of N linear equations is obtained;

[0069] 7) Using the least squares method based multivariate linear regression algorithm regress to calculate a set of regression coefficients, i.e., the abnormality of the meter or the electricity theft system;

[0070] 8) The order of regression coefficients corresponds to the order of each sub-table, and each regression coefficient is compared with the set threshold. If the regression coefficient is greater than the abnormal electricity consumption electric energy meter identification event threshold and lower than the electricity theft event reporting threshold, it is determined to be an abnormal electricity user; if the regression coefficient exceeds the electricity theft event reporting threshold, it is determined to be an electricity theft user. After the intelligent measurement switch obtains the corresponding meter number, it records the corresponding event and reports it to the collector, which then reports it to the main station system.

[0071] 9) At this point, the system has completed the accurate positioning of abnormal users or electricity theft users.

[0072] The embodiments of the present invention have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.

Claims

1. A method for accurately locating users with abnormal electricity consumption or electricity theft in a power system. It is characterized in that Each meter box is equipped with an intelligent measuring switch to obtain the power consumption of the meter box; the difference between the power consumption of the intelligent measuring switch and the power consumption of all sub-meters is used as the power loss to determine abnormal power consumption or power theft; the regression coefficient of each sub-meter is calculated through a multivariate linear regression model, and the abnormal power consumption or power theft status of each sub-meter is identified by comparing the regression coefficient with a set threshold; the method includes the following steps: S1. Configure an intelligent measurement switch at the external power supply end of the meter box, set the error threshold Tve, the abnormal threshold TVa of the abnormal power consumption identification event of the electric energy meter, and the theft threshold TVs of the power theft identification event of the electric energy meter; the intelligent measurement switch communicates with each sub-meter, and establishes a meter library according to the communication address and sub-meter information of each sub-meter; S2. Let the number of sub-meters be M. The intelligent measuring switch collects the power consumption of each sub-meter in the meter library and its own power at a fixed time. The power collected from M sub-meters once constitutes a sampling point. When the number of sampling points N is twice the number of sub-meters M, the power theft analysis is performed: S3, calculate the power loss according to the power value of the intelligent measuring switch at the Nth sampling point on the day and the sum of the power consumption of each sub-meter on the day, and obtain the total error according to the power loss. If the total error is greater than the set error threshold Tve, it is determined that there is an abnormal power consumption or power theft user in the meter box, and go to S4, otherwise, return to S2 to continue sampling; S4, establish a multiple linear regression model, remove the sub-table data with abnormal data in N sampling points, the number of valid sub-tables after removing the abnormal data is M', substitute the M' sub-table sampling data of N sampling points into the multiple linear regression model, and obtain a multiple linear regression equation group consisting of N linear equations; The multivariate linear regression algorithm based on the least squares method is used to calculate and obtain a set of regression coefficients β i , as the abnormal or electricity theft coefficient of the sub-table, i represents the sub-table number, i∈[1,M']; S5, the regression coefficients β are sequentially i Each sub-table corresponds to each other, and each regression coefficient is compared with the set abnormal threshold TVa and electricity theft threshold TVs respectively: If the regression coefficient β i If it is less than the abnormal threshold TVa, it is judged as a normal user; If the regression coefficient β i If the value is greater than or equal to the abnormal threshold TVa and less than the electricity theft threshold TVs, the user is judged as an abnormal electricity user; If the regression coefficient β i If the user's value is greater than or equal to the electricity theft threshold TVs, the user is considered to be an electricity theft user. The intelligent measurement switch obtains the judgment information of each sub-meter, records the corresponding events and reports them to the main station system through the collector.

2. According to claim 1, a method for accurately locating users with abnormal electricity consumption or electricity theft in a power system, Features The specific steps of establishing the meter library in S1 are as follows: the intelligent measuring switch automatically searches the meter after being powered on, that is, sends a communication address request frame to each sub-meter in the meter box, receives the communication address of each sub-meter, and forms a meter library with the communication address and sub-meter information of each sub-meter.

3. According to claim 1, a method for accurately locating users with abnormal electricity consumption or electricity theft in a power system, Features The intelligent measuring switch in S2 collects the electric quantity of each sub-meter at a fixed time, removes invalid data and saves it, wherein the invalid data refers to data of unsuccessful meter reading or abnormal communication.

4. According to claim 1, a method for accurately locating users with abnormal electricity consumption or electricity theft in a power system, Features In S3, the power loss P is calculated using the following formula: L : Among them, P G is the current power value of the smart measuring switch, P i is the daily power consumption value of each sub-meter in the meter box, M is the total number of sub-meters in the meter box, and i represents the sub-meter number; Calculate the total error based on power loss: P L / P G .

5. According to claim 1, a method for accurately locating users with abnormal electricity consumption or electricity theft in a power system, Features In S4, if the power consumption data of a sub-table is always 0, the data corresponding to the sub-table is deleted.

6. A method for accurately locating users with abnormal electricity consumption or electricity theft in a power system according to claim 1, Features In S4, the multiple linear regression model is as follows: Where: t represents the time, i.e. the sampling point number, t∈[1, N], y error (t) represents the power loss at time t, that is, the difference between the power consumption of the intelligent measuring switch and the sum of the power consumption of all sub-meters at time t, δ represents the error term, i represents the sub-meter number, i∈[1, M'], M' represents the number of valid sub-meters after eliminating abnormalities, β i Indicates the abnormal or power theft coefficient of the subtable numbered i, x i (t) represents the power consumption of the submeter numbered i; Substitute the data of M' subtables of N sampling points into the multivariate linear regression model, and obtain the N linear regression equations as follows: Where: x tM′ β M′ represents the loss contributed by the M'th block sub-table at the tth sampling point; Call the multivariate linear regression algorithm regress to calculate the linear regression coefficient β 1 ,β 2 ,…,β M′ , as the abnormality or electricity theft coefficient of each sub-meter.

7. A method for accurately locating users with abnormal electricity consumption or electricity theft in a power system according to claim 1, Features The error threshold Tve<abnormal threshold TVa<electricity theft threshold TVs.

8. A method for accurately locating users with abnormal electricity consumption or electricity theft in a power system according to claim 7, Features The electricity theft threshold TVs is 10%.

Images