Low-voltage user electricity stealing identification system and method based on correlation association search

Abstract

The invention discloses a low-voltage user electricity stealing identification system based on correlation association search, and the system comprises a data obtaining module which is used for obtaining the line loss data in a target time period and the electricity consumption data of each user; the data preprocessing module is used for preprocessing the acquired data; the correlation association search module is used for calculating a correlation coefficient between the normalized line loss data and the electricity consumption data of each user, and identifying users strongly correlated with the line loss data; the association search identification module is used for finding a user combination with strong correlation between the accumulated electricity consumption sequence and the line loss sequence; the sliding window detection module is used for calculating a mean value of correlation coefficients in different time periods to confirm electricity stealing suspicion; and the output module is used for outputting a final suspected electricity stealing user list. The invention further discloses a corresponding method. According to the invention, the electricity larceny behavior of the low-voltage user can be effectively identified, the efficiency and accuracy of electricity larceny detection are improved, and the detection cost is reduced.

Description

Technical Field

[0001] The present invention relates to the field of electricity fee management, and in particular to a method for identifying electricity theft by low-voltage users in industrial parks or urban power grids when multiple enterprises share transformers. The present invention uses data analysis technology to effectively identify electricity theft by low-voltage users, thereby improving the efficiency and accuracy of electricity fee management. Background Art

[0002] With the continuous expansion of my country's power supply network and the popularization of household appliances, electricity theft incidents occur frequently, causing economic losses to power companies and threatening the safety and stability of the power supply network. The number of low-voltage users is large and increasing, and the dimensions of the available data of low-voltage users are limited. It is necessary to obtain additional data from many regions, such as voltage, current, meter cover opening and closing events, etc. Acquiring this data not only increases the burden on personnel, but also has limited data storage time. In addition, the quality of meter reading equipment is uneven, resulting in a large amount of data loss. This situation poses a huge challenge to anti-electricity theft work. Traditional manual on-site inspection methods are inefficient, costly, and the number of electricity theft cases captured is limited. Therefore, it is urgent to establish an effective model based on data analysis to detect electricity theft.

[0003] The main method for power supply companies to detect electricity theft is to check the user's electricity meter on site. This method is inefficient, labor-intensive and resource-intensive, and has a limited inspection scope. Therefore, the number of electricity theft cases caught is very small. In order to detect electricity theft more effectively, many electricity theft detection methods have been proposed at home and abroad in recent years, which can be mainly divided into two categories: expert system and machine learning. Expert systems contain manually formulated decision rules, which are mainly based on the physical model and energy balance equation of the power system. This usually requires network topology information, which is difficult to apply in areas where the network topology changes frequently and the metering infrastructure is not perfect. In contrast, data-driven machine learning methods can use existing electricity theft samples and electricity consumption data to train different detection models. They have the advantages of flexible use, low cost and high accuracy, and have become the main development direction of electricity theft detection.

[0004] Specialized electricity theft equipment or technology makes anti-electricity theft more difficult. Traditional analysis methods not only waste manpower, but also rely on human quality. The detection efficiency is too low to timely warn of electricity theft. Many electricity theft detection methods require manual analysis and interpretation of power grid data. This method has the risk of human error and subjective judgment, and may require a lot of human resources and time, especially in large-scale power grids. At the same time, there are problems with data collection and access difficulties. Electricity theft detection requires the acquisition and analysis of a large amount of power data, including current, voltage, power data, etc. However, access and collection of this data may be hindered, especially in distributed power grids or private property areas. Therefore, it may be a challenge to obtain enough data for accurate electricity theft detection. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a system and method for identifying electricity theft by low-voltage users based on correlation association search. The method analyzes the line loss and electricity consumption data in the same low-voltage area, constructs a correlation association search model and an association search recognition model in a feature selection algorithm, thereby realizing the detection of electricity theft by low-voltage users in urban power grids and improving the detection efficiency and accuracy.

[0006] In order to solve the above technical problems, as a first aspect of the present invention, a low-voltage user electricity theft identification system based on correlation association search is provided, which comprises:

[0007] A data acquisition module is used to obtain line loss data and power consumption data of each user within a target time period from a power collection system;

[0008] A data preprocessing module is used to preprocess the acquired data to filter out the users of key concern; the preprocessing includes: cleaning, screening and normalization processing;

[0009] A correlation association search module is used to calculate the correlation coefficient between the normalized line loss data and the power consumption data of each user by using a correlation association search model, and identify users with strong correlation with the line loss data;

[0010] The correlation search and identification module is used to find the user combination whose cumulative power consumption sequence is strongly correlated with the line loss sequence among all users belonging to the same transformer area by using the correlation search and identification model;

[0011] A sliding window detection module is used to perform sliding window detection on the identified users or user combinations, and confirm the suspicion of electricity theft by calculating the mean of the correlation coefficient in different time periods;

[0012] The output module is used to output the final list of suspected electricity theft users.

[0013] Optionally, the data preprocessing module further includes:

[0014] The user screening unit is used to focus on users whose daily power consumption accounts for a large proportion of the power supply in the transformer area, and to eliminate users with low and stable power consumption;

[0015] The data normalization unit processes the power consumption data of the remaining users using a zero-mean normalization method.

[0016] Optionally, the correlation association search model uses the Pearson correlation coefficient as an evaluation indicator, and obtains the correlation coefficient by calculating the covariance and standard deviation between the line loss sequence and the power consumption sequence.

[0017] Optionally, the association search recognition model adopts an add-L, remove-R selection algorithm to search for an optimal user combination by iteratively adding and removing users so that the correlation coefficient between the combination and the line loss sequence is maximized.

[0018] Optionally, the sliding window detection module divides the target time period into multiple sub-time periods, calculates the correlation coefficient in each sub-time period, and then obtains the average; if the average is greater than a preset threshold, the corresponding user or user combination is regarded as the final suspect.

[0019] As another aspect of the present invention, a method for identifying electricity theft by low-voltage users based on correlation association search is also provided, which comprises the following steps:

[0020] Step S1, obtaining line loss data of low-voltage areas and power consumption data of each user from a power collection system;

[0021] Step S2, preprocessing the acquired data to screen out key users; the preprocessing includes: cleaning, screening and normalization processing;

[0022] Step S3, using the correlation association search model to calculate the correlation coefficient between the normalized line loss data and the power consumption data, and identifying users with strong correlation with line loss;

[0023] Step S4, using the association search recognition model to find a user combination whose cumulative power consumption sequence is strongly correlated with the line loss sequence among all users belonging to the same transformer area;

[0024] Step S5, performing sliding window detection on the identified user or user combination, and confirming the suspicion of electricity theft by calculating the average of the correlation coefficients in different time periods;

[0025] Step S6: output the final list of suspected electricity theft users.

[0026] Optionally, the step S2 further includes:

[0027] Screen out users whose daily electricity consumption accounts for a large proportion of the power supply in the substation as key focus objects;

[0028] Eliminate users with low and stable electricity consumption and users with zero electricity consumption;

[0029] The electricity consumption data of the remaining users are normalized using zero mean.

[0030] Optionally, in step S13, the Pearson correlation coefficient is used as an evaluation index, and the correlation coefficient is obtained by calculating the covariance and standard deviation between the line loss sequence and the power consumption sequence.

[0031] Optionally, in step S4, an L-adding, R-removing selection algorithm is used to search for the optimal user combination by iteratively adding and removing users so that the correlation coefficient between the combination and the line loss sequence is maximized to find a user combination that is highly correlated with the line loss data.

[0032] Optionally, in step S5, the target time period is divided into multiple sub-time periods, and the correlation coefficient in each sub-time period is calculated, and then the mean is taken as the final correlation coefficient for judgment. If the mean is greater than a preset threshold, the user or user combination is regarded as the final suspect.

[0033] Optionally, it also includes:

[0034] Step S7: conduct on-site verification of the final list of suspected electricity theft users, confirm the electricity theft behavior, and take corresponding measures to deal with it.

[0035] The implementation of the embodiments of the present invention has the following beneficial effects:

[0036] The present invention provides a low-voltage user electricity theft identification system and method based on correlation association search. On the basis of analyzing line loss and power consumption data, combined with the characteristics and laws of electricity theft behavior, a correlation association search model and an association search identification model are established to search for suspected electricity theft users, and then detect the users through a sliding window inspection method. By implementing the present invention, features with a strong correlation with electricity theft behavior can be screened out from line loss and power consumption data; compared with the current mainstream data analysis method, this method can achieve a higher identification accuracy by using only line loss and power consumption data; at the same time, it can effectively detect the electricity theft behavior of low-voltage users, greatly narrow the search range of electricity theft users, and reduce line losses to a certain extent.

[0037] This method can achieve high recognition accuracy using only line loss and power consumption data, breaking the data access barriers of current mainstream data analysis methods. It does not require additional acquisition of user detailed data or access to peripheral data. It can reduce the workload of staff and improve the level of informatization of anti-electricity theft work.

[0038] The present invention can quickly screen out suspected electricity theft users through data analysis methods, greatly improving the efficiency of electricity theft identification. Compared with the traditional manual on-site inspection method, the present invention reduces the investment of manpower and material resources and reduces the cost of electricity theft detection. Through correlation analysis and sliding window detection, the present invention can more accurately identify electricity theft behavior and reduce misjudgment and missed judgment. The present invention is suitable for low-voltage users of different sizes and types and has good scalability and applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, without paying creative labor, other drawings obtained based on these drawings still belong to the scope of the present invention.

[0040] Figure 1 A schematic diagram of the relationship between normalized line loss and power consumption involved in the present invention;

[0041] Figure 2 A schematic diagram of the main process of an embodiment of low-voltage user electricity theft identification based on correlation association search provided by the present invention;

[0042] Figure 3 for Figure 2 A more detailed flow chart of step S4 in FIG.

[0043] Figure 4 A schematic diagram of the structure of an embodiment of a low-voltage user electricity theft identification system based on correlation association search provided by the present invention;

[0044] Figure 5 for Figure 4 Schematic diagram of the structure of the data preprocessing module. DETAILED DESCRIPTION

[0045] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings.

[0046] In order to understand the technical solution of the present invention, the technical terms involved in the present invention are first introduced as follows:

[0047] Electricity theft: Electricity theft refers to the illegal acquisition and use of electricity resources without settlement and payment in accordance with regulations. This behavior has seriously affected the normal power supply and operation order, and has caused a series of negative impacts on the power industry and social economy. Electricity theft includes various forms: illegal wiring of public power facilities, private wiring, tampering with electricity meters, collusion between insiders and outsiders to steal electricity, etc.

[0048] Low-voltage users: Low-voltage users refer to users with lower voltage levels in the power system. In the power system, electricity is transmitted and transformed from power plants through substations before being delivered to users. According to the different electricity demand and electricity usage locations, the power system divides users into different voltage levels, among which low-voltage users refer to users with lower voltage. The voltage level of low-voltage users is usually 220V or 230V (depending on the electricity standards of the region and country), which is a common voltage level for homes, commercial buildings, small industries, and small offices. Compared with high-voltage users, low-voltage users have a smaller power load and require relatively lower power capacity.

[0049] Correlation: Correlation refers to the degree of association or the strength of the relationship between two or more variables. Two variables are considered to be correlated when their changing trends are similar or clearly related. Correlation can have different forms and degrees. Commonly used methods to measure correlation include correlation coefficient and covariance. Correlation coefficients (such as the Pearson correlation coefficient) can measure the strength of the linear relationship between two variables, ranging from -1 to +1, where -1 indicates a perfect negative correlation, +1 indicates a perfect positive correlation, and 0 indicates no correlation. Covariance measures the overall trend of change of two variables, depending on the difference between the variables and the direction of correlation.

[0050] Sliding Window Detection: Sliding Window Detection is a method for detecting specific events or patterns in time series data or signals. It performs detection by sliding a fixed-size window over the data sequence and analyzing and processing the data within the window. In sliding window detection, the window is usually a continuous subsequence of a fixed size. The window slides over time with a fixed step size, sliding one data point or a small piece of data each time. The size and step size of the sliding window can be determined based on specific applications and requirements. Sliding window detection can be applied to a variety of data analysis and pattern recognition tasks.

[0051] Line loss: Line loss in the power grid refers to the loss of electric energy caused by the resistance, inductance, capacitance and other factors of wires and cables during the power transmission process. Line loss consists of two parts: technical loss and non-technical loss. Technical loss is the loss of electric energy due to the resistance of the wire material. When the current passes through the wire, the wire itself will generate resistance, causing the electric energy to be converted into heat energy and lost. The resistance of the wire is related to the length, cross-sectional area, material resistivity and current of the wire. Technical loss can be reduced by selecting appropriate wire materials, optimizing the design of transmission lines and reducing current load. Non-technical loss refers to the loss of electric energy due to factors such as operating errors, power theft, metering errors, equipment failures, etc. These losses are usually unavoidable and need to be controlled and reduced by measures such as improving power system management, strengthening monitoring and maintenance, and improving equipment efficiency.

[0052] It is understandable that, considering a power consumption area on the low-voltage user side, line loss includes management line loss and technical line loss. Technical line loss is mainly related to the parameters of the power supply equipment and is an inevitable part, generally accounting for about 3%-5% of the line loss. Assuming that the technical line loss remains unchanged, the change in the power consumption of normal users will hardly cause a large change in line loss, while the power consumption of power-stealing users is related to line loss. According to the principle of power balance:

[0053]

[0054] In the above formula, P L (t), P T (t) are the management factor power loss and technical power loss at time t, respectively, P G (t) is the power supply at time t, P i (t) is the meter reading of user i at time t. When the line is normal, P T (t) is a normal constant.

[0055] When user i steals electricity, P i (t) and P L (t) decreases and increases respectively, resulting in an increase in line loss power, and the greater the amount of stolen electricity, the greater the line loss power. This results in a significant correlation between the electricity consumption sequence of the electricity stealing user and the line loss sequence. That is, as the amount of stolen electricity from the user increases with the increase in his or her electricity consumption, the line loss will also increase accordingly, and vice versa. From a mathematical point of view, this means that line loss is positively correlated with the electricity consumption of the electricity stealing user, and their correlation can be evaluated by calculating the correlation coefficient of the two sequences. From a commercial point of view, this same-direction change reflects a form of electricity theft that is achieved by slowing down or stopping the meter through modification (that is, the meter does not fully measure the actual electricity consumption and has the characteristics of proportional electricity theft). Figure 1 A schematic diagram of the relationship between normalized line loss and power consumption is shown, and the technical solution of the present invention will be set accordingly based on the correlation.

[0056] like Figure 2 As shown, a main flow diagram of an embodiment of low-voltage user electricity theft identification based on correlation association search provided by the present invention is shown; Figure 2 As shown, in this embodiment, the method at least includes the following steps:

[0057] Step S1, obtaining line loss data of low-voltage areas and power consumption data of each user from a power collection system;

[0058] Step S2, preprocessing the acquired data to screen out key users; the preprocessing includes: cleaning, screening and normalization processing;

[0059] Step S3, using the correlation association search model to calculate the correlation coefficient between the normalized line loss data and the power consumption data, and identifying users with strong correlation with line loss;

[0060] Step S4, using the association search recognition model to find a user combination whose cumulative power consumption sequence is strongly correlated with the line loss sequence among all users belonging to the same transformer area;

[0061] Step S5, performing sliding window detection on the identified user or user combination, and confirming the suspicion of electricity theft by calculating the average of the correlation coefficients in different time periods;

[0062] Step S6: output the final list of suspected electricity theft users.

[0063] The following will be combined Figure 2 The above steps are described in detail.

[0064] Specifically, in step S1, the line loss rate is compared with the average line loss rate of the same period. If the line loss fluctuation exceeds a certain threshold, the line loss and user power consumption within 30 days before and after the abnormal date are extracted from the collection system. For example, in one example, assuming that the line loss rate is found to increase suddenly on the 15th of a month and exceeds the preset threshold, the line loss data and daily power consumption data of all users from the 1st to the 30th of the month (including 15 days before and after the abnormal date) are extracted.

[0065] Specifically, in step S2, it also includes:

[0066] Screen out users whose daily electricity consumption accounts for a large proportion of the power supply in the substation as key focus objects;

[0067] Eliminate users with low and stable power consumption and users with zero power consumption. In this step, by calculating the sequence variance, if the variance is less than 5%, the power consumption is considered stable and the possibility of power theft is low, and then they are eliminated; at the same time, for users with zero power consumption, the correlation coefficient between their power consumption sequence and line loss cannot be calculated, so this user is eliminated and analyzed separately.

[0068] The power consumption data of the remaining users is normalized using zero mean, and the normalization formula is as follows:

[0069]

[0070] Where:

[0071] z: normalized value;

[0072] x: original data value;

[0073] μ: the mean of the data, that is, the average value of all data points;

[0074] σ: standard deviation of the data, indicating the degree of dispersion of the data.

[0075] Specifically, in step S3, the Pearson correlation coefficient is used as an evaluation index, and the correlation coefficient is obtained by calculating the covariance and standard deviation between the line loss series and the power consumption series.

[0076] It can be understood that based on the analysis of line loss and power consumption, a correlation search model is established to identify power theft by searching for power consumption sequences with strong correlation with line loss during the target period. The higher the correlation, the greater the possibility that user i steals power. This algorithm evaluates the correlation between line loss X and power consumption sequence Y by calculating its Pearson correlation coefficient U. The calculation formula is:

[0077]

[0078] Where:

[0079] ρ: Pearson correlation coefficient, used to measure the degree of linear correlation between two sequences;

[0080] cov(X,Y): the covariance of sequences X and Y, indicating the degree to which the two sequences vary together;

[0081] σx and σy: are the standard deviations of sequences X and Y, respectively, indicating the degree of discreteness of the sequences;

[0082] N: the number of elements in the sequences X and Y, i.e. the number of data points;

[0083] Xi and Yi: are the values ​​of the sequences X and Y at the i-th data point, respectively.

[0084] For example, in one example, the normalized power consumption series and line loss series of user C are calculated, and the Pearson correlation coefficient between the two is 0.85, indicating that there is a strong positive correlation between the power consumption of user C and the line loss.

[0085] In step S3, for users with a high ranking and a correlation coefficient greater than a threshold, the suspected power theft cycle can be shifted backward by 1 to form a new sequence, and the Pearson correlation coefficient between the normalized line loss sequence and the power consumption sequence can be calculated. If the threshold is exceeded three times in a row, the user is included in the suspect list.

[0086] Specifically, in step S4, an L-adding, R-removing selection algorithm is used to search for the optimal user combination by iteratively adding and removing users so that the correlation coefficient between the combination and the line loss sequence is maximized to find a user combination that is highly correlated with the line loss data.

[0087] It is understandable that in recent years, low-voltage users in urban power grids have also experienced group electricity theft characterized by large-scale, full-region linkage. An analysis of the electricity consumption data corresponding to the historical electricity theft meters found that when electricity theft occurred, the line loss sequence of a single electricity theft user was not significantly correlated with the electricity consumption sequence, but the cumulative electricity consumption of multiple electricity theft users was significantly correlated with the line loss sequence, which could not be identified by the above-mentioned association search model. Based on the above analysis of this electricity theft behavior, it is necessary to find a user combination whose cumulative electricity consumption sequence is strongly correlated with the line loss sequence among all users belonging to the same transformer area. If all user combinations are listed and verified one by one to detect whether there is a strong correlation with the line loss, the amount of calculation will be huge. If the number of users in the low-voltage area is N, and the computational complexity is 2 N -(N+1).

[0088] Therefore, the present invention proposes an associated search identification model to identify users who steal electricity in groups. The model defines the correlation association function as an evaluation function. The search process of user combinations is guided by calculating the correlation coefficient between the line loss and the cumulative power consumption sequence of the user combination. When the correlation coefficient obtained by the search no longer increases, the search stops. When the correlation coefficient exceeds a certain threshold, the corresponding combination is more suspected of stealing electricity. The input is the line loss sequence and power consumption sequence of all users in the low-voltage area within the target time period, and the output is a list of suspected power stealing users. The model adopts a selection algorithm of adding L and removing R to obtain a combination of users. Among them,

[0089] L: The number of users added to subset X from among the unselected users in each round.

[0090] R: The number of users removed from subset X.

[0091] X: The current user subset, used to calculate the evaluation function value (ie, correlation coefficient).

[0092] The algorithm starts from an empty set, and selects L users from the unselected users in each round and adds them to the subset X in turn. When the evaluation function value corresponding to the subset X no longer increases, the addition stops.

[0093] Then remove R users from subset X. When the evaluation function value corresponding to subset X no longer increases, stop deleting and obtain subset X.

[0094] When the evaluation function value corresponding to the subset X is greater than a certain threshold, the subset X is a user group suspected of stealing electricity.

[0095] For the specific process, please refer to Figure 3 For example, in one example, the correlation coefficients of all users are sorted from high to low, and it is found that the correlation coefficients of users C, D, and E are all greater than 0.8, so they are included in the group of users suspected of stealing electricity.

[0096] Specifically, in step S5, the target time period is divided into multiple sub-time periods, and the correlation coefficient in each sub-time period is calculated, and then the mean is taken as the final correlation coefficient for judgment. If the mean is greater than a preset threshold, the user or user combination is regarded as the final suspect.

[0097] It is understandable that through the analysis of line loss and power consumption data, it is found that the correlation coefficient between user (set) power consumption and line loss in some target time periods is relatively high, which improves the overall correlation coefficient. This situation affects the recognition rate of suspected power theft. If the target time is 30 days, these 30 days are divided into 3 sections and the correlation is calculated separately.

[0098] In one example, the correlation coefficients of each segment are shown in Table 1. If the correlation coefficients of some segments are too high or too low, they will affect the overall correlation coefficient. For example, for the data of user A, the high correlation coefficient of the third part improves the overall correlation coefficient. In actual operation, the user's electricity consumption characteristics only match the line loss in the later period of the target time, but the suspicion of electricity theft is small.

[0099] Table 1

[0100]

[0101] In order to solve this problem, this method uses a sliding window detection method to verify the obtained suspect list. First, the target time period is divided into three segments using a sliding window, and the mean correlation coefficient of the three segments is calculated. like The user (set) is considered the final suspect.

[0102] For example, for user E, a sliding window detection is performed, and the 30-day electricity consumption data is divided into three segments, each of which is 10 days. The correlation coefficients between the electricity consumption and line loss of user C in these three periods are calculated respectively, and three values ​​are obtained: 0.927, 0.985, and 0.907. The mean of these three values ​​is calculated to obtain ρ-=(0.927+0.985+0.907) / 3=0.940. Since ρ is greater than the preset threshold of 0.9, user E is regarded as the final suspect.

[0103] Optionally, the method provided by the present invention further comprises:

[0104] Step S7: Conduct an on-site check on the final list of suspected electricity theft users, confirm the electricity theft, and take appropriate measures to deal with it. Individual electricity theft suspects and group electricity theft suspects of user groups are included in the final suspect list. Staff can be arranged to conduct a detailed investigation of the electricity usage of these users or user groups. Take appropriate measures based on the investigation results, such as collecting electricity bills, punishing electricity theft, etc.

[0105] In practical applications, a total of 165 low-voltage transformer areas and 8,958 users in a power supply station of the urban power grid were selected to verify the effectiveness of the method provided by the present invention. The data comes from the electricity theft situation in the existing low-voltage transformer area in 2019. The results show that 80% of the electricity theft cases can be detected by the proposed recognition model. Although some users in the suspect list have been confirmed not to have stolen electricity, this has greatly narrowed the search scope of electricity theft cases. The model has obvious advantages in searching for low-voltage electricity theft users. It can not only improve work efficiency, but also further improve the informatization level of anti-electricity theft work. The method proposed in the present invention has strong practical value.

[0106] like Figure 4 FIG. 1 is a schematic diagram showing a structure of an embodiment of a low-voltage user electricity theft identification system based on correlation association search provided by the present invention; and FIG. Figure 5 As shown, in this embodiment, the low-voltage user electricity theft identification system 1 at least includes:

[0107] The data acquisition module 10 is used to acquire the line loss data and the power consumption data of each user within the target time period from the power collection system;

[0108] The data preprocessing module 11 is used to preprocess the acquired data to filter out the users of key concern; the preprocessing includes: cleaning, screening and normalization processing;

[0109] A correlation association search module 12 is used to calculate the correlation coefficient between the normalized line loss data and the power consumption data of each user by using a correlation association search model, and identify users with strong correlation to the line loss data;

[0110] The correlation search identification module 13 is used to find the user combination whose cumulative power consumption sequence is strongly correlated with the line loss sequence among all users belonging to the same transformer area by using the correlation search identification model;

[0111] A sliding window detection module 14 is used to perform sliding window detection on the identified user or user combination, and confirm the suspicion of electricity theft by calculating the mean of the correlation coefficient in different time periods;

[0112] The output module 15 is used to output the final list of suspected electricity theft users.

[0113] like Figure 5 As shown, in a specific example, the data preprocessing module 11 further includes:

[0114] The user screening unit 110 is used to focus on users whose daily power consumption accounts for a large proportion of the power supply in the transformer area, and to eliminate users with low and stable power consumption;

[0115] The data normalization unit 111 processes the power consumption data of the remaining users by using a zero-mean normalization method.

[0116] More specifically, in the correlation association search module 12, the correlation association search model uses the Pearson correlation coefficient as an evaluation index, and obtains the correlation coefficient by calculating the covariance and standard deviation between the line loss sequence and the power consumption sequence.

[0117] More specifically, in the association search identification module 13, the association search identification model adopts an add-L, remove-R selection algorithm to search for the optimal user combination by iteratively adding and removing users so that the correlation coefficient between the combination and the line loss sequence is maximized.

[0118] Optionally, the sliding window detection module 14 divides the target time period into multiple sub-time periods, calculates the correlation coefficient in each sub-time period, and then obtains the average; if the average is greater than a preset threshold, the corresponding user or user combination is regarded as the final suspect.

[0119] For more details, please refer to and combine the above Figure 2 , Figure 3 The description is not repeated here.

[0120] The implementation of the embodiments of the present invention has the following beneficial effects:

[0121] The present invention provides a low-voltage user electricity theft identification system and method based on correlation association search. On the basis of analyzing line loss and power consumption data, combined with the characteristics and laws of electricity theft behavior, a correlation association search model and an association search identification model are established to search for suspected electricity theft users, and then detect the users through a sliding window inspection method. By implementing the present invention, features with a strong correlation with electricity theft behavior can be screened out from line loss and power consumption data; compared with the current mainstream data analysis method, this method can achieve a higher identification accuracy by using only line loss and power consumption data; at the same time, it can effectively detect the electricity theft behavior of low-voltage users, greatly narrow the search range of electricity theft users, and reduce line losses to a certain extent.

[0122] This method can achieve high recognition accuracy using only line loss and power consumption data, breaking the data access barriers of current mainstream data analysis methods. It does not require additional acquisition of user detailed data or access to peripheral data. It can reduce the workload of staff and improve the level of informatization of anti-electricity theft work.

[0123] The present invention can quickly screen out suspected electricity theft users through data analysis methods, greatly improving the efficiency of electricity theft identification. Compared with the traditional manual on-site inspection method, the present invention reduces the investment of manpower and material resources and reduces the cost of electricity theft detection. Through correlation analysis and sliding window detection, the present invention can more accurately identify electricity theft behavior and reduce misjudgment and missed judgment. The present invention is suitable for low-voltage users of different sizes and types and has good scalability and applicability.

[0124] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0125] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1A device that provides the functions specified in a block or multiple blocks.

[0126] The above disclosure is only a preferred embodiment of the present invention, which certainly cannot be used to limit the scope of the present invention. Therefore, equivalent changes made according to the claims of the present invention are still within the scope of the present invention.

Claims

1. A low-voltage user electricity theft identification system based on correlation association search, characterized in that: include: A data acquisition module is used to obtain line loss data and power consumption data of each user within a target time period from a power collection system; The data preprocessing module is used to preprocess the acquired data to filter out the users of key concern; The pretreatment includes: cleaning, screening and normalization; A correlation association search module is used to calculate the correlation coefficient between the normalized line loss data and the power consumption data of each user by using a correlation association search model, and identify users with strong correlation with the line loss data; The correlation search and identification module is used to find the user combination whose cumulative power consumption sequence is strongly correlated with the line loss sequence among all users belonging to the same transformer area by using the correlation search and identification model; A sliding window detection module is used to perform sliding window detection on the identified users or user combinations, and confirm the suspicion of electricity theft by calculating the mean of the correlation coefficient in different time periods; The output module is used to output the final list of suspected electricity theft users.

2. The system according to claim 1, characterized in that The data preprocessing module also includes: The user screening unit is used to focus on users whose daily power consumption accounts for a large proportion of the power supply in the transformer area, and to eliminate users with low and stable power consumption; The data normalization unit processes the power consumption data of the remaining users using a zero-mean normalization method.

3. The system according to claim 2, characterized in that The correlation association search model uses the Pearson correlation coefficient as an evaluation index, and obtains the correlation coefficient by calculating the covariance and standard deviation between the line loss sequence and the power consumption sequence.

4. The system according to claim 3, characterized in that The association search recognition model adopts an add-L, remove-R selection algorithm to search for an optimal user combination by iteratively adding and removing users, so that the correlation coefficient between the combination and the line loss sequence is maximized.

5. The system according to claim 4, characterized in that The sliding window detection module divides the target time period into multiple sub-time periods, calculates the correlation coefficient in each sub-time period, and then obtains the average; if the average is greater than a preset threshold, the corresponding user or user combination is regarded as the final suspect.

6. A method for identifying electricity theft by low-voltage users based on correlation-associated search, characterized in that: The following steps are involved: Step S1, obtaining line loss data of low-voltage areas and power consumption data of each user from a power collection system; Step S2, pre-processing the acquired data to screen out key users; The pretreatment includes: cleaning, screening and normalization; Step S3, using the correlation association search model to calculate the correlation coefficient between the normalized line loss data and the power consumption data, and identifying users with strong correlation with line loss; Step S4, using the association search recognition model to find a user combination whose cumulative power consumption sequence is strongly correlated with the line loss sequence among all users belonging to the same transformer area; Step S5, performing sliding window detection on the identified user or user combination, and confirming the suspicion of electricity theft by calculating the average of the correlation coefficients in different time periods; Step S6: output the final list of suspected electricity theft users.

7. The method according to claim 6, characterized in that The step S2 further comprises: Screen out users whose daily electricity consumption accounts for a large proportion of the power supply in the substation as key focus objects; Eliminate users with low and stable electricity consumption and users with zero electricity consumption; The electricity consumption data of the remaining users are normalized using zero mean.

8. The method according to claim 7, characterized in that In step S13, the Pearson correlation coefficient is used as an evaluation index, and the correlation coefficient is obtained by calculating the covariance and standard deviation between the line loss sequence and the power consumption sequence; In step S4, an L-adding and R-removing selection algorithm is used to search for the optimal user combination by iteratively adding and removing users so that the correlation coefficient between the combination and the line loss sequence is maximized, so as to find a user combination that is highly correlated with the line loss data.

9. The method according to claim 8, characterized in that In step S5, the target time period is divided into multiple sub-time periods, and the correlation coefficient in each sub-time period is calculated, and then the average is taken as the final correlation coefficient for judgment. If the average is greater than a preset threshold, the user or user combination is regarded as the final suspect.

10. The method according to claim 9, characterized in that Also includes: Step S7: conduct on-site verification of the final list of suspected electricity theft users, confirm the electricity theft behavior, and take corresponding measures to deal with it.

Images