Electricity larceny prevention identification method and system

Abstract

The invention provides an anti-electricity-stealing identification method and system. The method comprises the following steps: acquiring electricity utilization characteristics of each original user in a to-be-inspected transformer area; analyzing the power utilization characteristics of each original user to obtain a data change curve of each power utilization characteristic; identifying whether the original user has an electricity stealing behavior based on the data change curve; marking the original user with the electricity stealing behavior as a target user, and establishing an electricity stealing probability curve of each time period based on the data change curve of the target user; and constructing an electricity stealing user portrait of the target user according to the electricity stealing probability curve. According to the invention, the technical problem of low inspection efficiency in anti-electricity-stealing identification caused by the fact that the electricity stealing behavior is identified only according to the electricity consumption data but cannot be effectively inspected is solved.

Description

Technical Field

[0001] The present invention relates to the field of anti-electricity theft, and in particular to an anti-electricity theft identification method and system. Background Art

[0002] There are unavoidable losses in the transmission and distribution of electricity, including technical line losses in the transmission, transformation and distribution process, as well as theft and theft of electricity. Theft of electricity not only increases the power load of the power system, but also easily causes fires and line failures, seriously threatening the safety of public property.

[0003] The existing anti-electricity theft identification technology mainly focuses on the detection and identification of electricity theft, specifically, analyzing the probability of electricity theft by users through the characteristics related to electricity theft, and issuing timely warnings based on the analysis results. However, the existing technology can only provide assistance for determining whether there is electricity theft, and cannot provide effective assistance for the inspection of electricity theft, resulting in low efficiency in the inspection work of anti-electricity theft identification.

[0004] The existing invention patent application document "An anti-electricity theft inspection and monitoring system and its working method combining man-machine-object" with publication number CN109858679A, the existing system includes: data acquisition and processing module; anti-electricity theft algorithm model library management module; anti-electricity theft warning model customization module; suspected electricity theft judgment module; suspected electricity theft user judgment result storage module; work order generation module; work order processing module; data collection module; work order archiving storage module; data feedback module. And the existing invention patent application document "Method and device for identifying electricity theft users in combination with line loss and abnormal events in the substation area" with publication number CN110824270A, the existing method includes: obtaining the substation area and user data of at least one substation to be inspected, the substation area and user data include substation area line loss data and basic data of each electricity user in the substation area; applying the substation area line loss abnormality detection method and the substation area line loss data of the at least one substation to be inspected, determining the abnormal line loss substation area with suspected electricity theft users in the specified electricity consumption period; for any abnormal line loss substation area with suspected electricity theft users, determining the K-means clustering suspected electricity theft user set, the support vector machine suspected electricity theft user set and the Bayesian algorithm suspected electricity theft user set; and determining the list of suspected electricity theft users in the abnormal line loss substation area after comprehensive evaluation. And the existing public document "Anti-electricity theft detection system based on optimized neural network algorithm", in which, in view of the shortcomings of BP neural network algorithm in detecting electricity theft problems, an anti-electricity theft detection method of optimizing BP neural network algorithm with genetic algorithm is proposed. The genetic algorithm is used to determine the population range and encode it. The adaptive function and selection function are introduced to calculate the fitness value to select the individuals with the highest survival rate and decode them, so as to determine the initial value parameters in the forward propagation process. The forward propagation parameters of the neural network are updated by the gradient descent method and the back propagation method to reduce the number of iterations in the model training process. The simulation experiment is carried out by MATLAB in combination with the test set data of the acquisition system. The simulation results show that this method improves the accuracy of the anti-electricity theft detection model detection, shortens the model training time, and has reliability in the detection of electricity theft problems. The visual detection platform GU is designed and built for visual analysis. The tool used is MATLABR2020b, and the electricity information data is tested by the anti-electricity theft detection system. However, the aforementioned prior art does not fully identify whether the original user has electricity theft behavior based on the data change curve; it is also impossible to determine the target user among the original users with electricity theft behavior. The aforementioned prior art does not further mine and analyze users with electricity theft behavior, and it is impossible to find the best audit time period. The audit efficiency of the aforementioned traditional scheme needs to be improved.

[0005] In summary, the existing technology has the technical problem of only identifying electricity theft based on electricity consumption data, but being unable to effectively audit the electricity theft, resulting in low efficiency of the audit work in anti-electricity theft identification. Summary of the invention

[0006] The technical problem to be solved by the present invention is how to solve the technical problem that the prior art only identifies electricity theft based on electricity consumption data, but cannot effectively audit the electricity theft, resulting in low audit efficiency in anti-electricity theft identification.

[0007] The present invention adopts the following technical solutions to solve the above technical problems: an anti-electricity theft identification method comprises:

[0008] S1. Collect the electricity consumption data of each original user from the area to be audited, and obtain the electricity consumption characteristics;

[0009] S2. Analyze the electricity consumption characteristics to obtain the electricity consumption characteristic data change curve of each original user;

[0010] S3. Construct and use an electricity theft identification model to identify whether the original user has committed electricity theft based on the change curve of electricity consumption characteristic data, and obtain an electricity theft identification result;

[0011] S4. According to the electricity theft identification result, the original user who is identified as the electricity theft behavior is marked as the target user, and based on the change curve of the target user's electricity consumption characteristic data, the electricity theft probability curve of each time period is established;

[0012] S5. Build an electricity theft user portrait of the target user based on the electricity theft probability curve for anti-electricity theft identification.

[0013] This application can determine the best audit time period by further mining and analyzing users who have committed electricity theft, which helps to improve the efficiency of the audit and solves the technical problem that the existing technology only identifies electricity theft based on electricity consumption data but cannot effectively audit electricity theft.

[0014] In a more specific technical solution, in S1, the electricity consumption characteristics include: the original user's actual electricity consumption, meter-displayed electricity consumption, line loss rate, and user electricity theft history data, and the difference between the actual electricity consumption and the meter-displayed electricity consumption is obtained.

[0015] In a more specific technical solution, S1 includes:

[0016] S11, calculating the electricity consumption difference based on the original user's electricity consumption data;

[0017] S12. Calculate the line loss growth rate from the line loss rate in the power consumption characteristics;

[0018] S13. Taking time as the independent variable, according to the power consumption difference and line loss growth rate, construct the power consumption change curve and line loss change curve respectively.

[0019] In more specific technical solutions, S3 includes:

[0020] S31, performing time registration on the power consumption change curve and the time change curve in the data change curve;

[0021] S32, extracting feature groups according to a preset time period;

[0022] S33. Building an electricity theft identification model based on a preset artificial intelligence model;

[0023] S34, preprocessing feature groups, obtaining and inputting preprocessing features into an electricity theft identification model to obtain electricity theft identification results.

[0024] Specifically, the electrical recognition model is based on the artificial intelligence model building process and also includes:

[0025] Acquire standard training data; wherein the standard training data includes standard input data consistent with the content attributes of the feature group, and annotated output data corresponding to the standard input data and used to indicate whether there is electricity theft;

[0026] Construct an artificial intelligence model; train the artificial intelligence model through standard training data, and mark the trained artificial intelligence model as an electricity theft identification model; wherein the artificial intelligence model includes a BP neural network model or a RBF neural network model.

[0027] In a more specific technical solution, the feature group includes: the average of the power consumption difference and the average of the line loss change.

[0028] In a more specific technical solution, S33 also includes:

[0029] S331, obtaining standard training data, and updating the standard training data according to the electricity consumption data; wherein the standard training data includes: standard input data with consistent feature group attributes, and output data with electricity theft behavior annotations;

[0030] S332. Use standard training data to train the artificial intelligence model to obtain and label the artificial intelligence model and determine the electricity theft identification model.

[0031] The standard training data in the present invention will be continuously updated and expanded along with the collected data, and the electricity theft identification model will also be continuously updated and trained through the standard training data to improve the availability and reliability of the electricity theft identification model.

[0032] In a more specific technical solution, S34 includes:

[0033] S341. For a preset time period, obtain the corresponding average power consumption difference and average line loss growth rate;

[0034] S342, integrating the average of the power consumption difference and the average of the line loss growth rate to obtain an integrated array;

[0035] S343, inputting the integrated array into the electricity theft identification model to perform identification analysis and obtain an electricity theft identification result.

[0036] Specifically, several feature data are integrated into a data sequence that can be recognized by the model.

[0037] In a more specific technical solution, S4 includes:

[0038] S41, dividing the data change curve corresponding to the target user into at least 2 data segments according to a preset unit time, and dividing the data segment into at least 2 data groups according to the data change curve;

[0039] S42, dividing the data segments in the data group according to a preset period to obtain data sub-segments, and associating and processing the data sub-segments according to a preset time to generate a sub-segment group;

[0040] Specifically, the data change curves corresponding to the target user for several days are divided into several data segments according to the day, and each data segment includes the target user's daily power consumption difference curve and line loss change curve, so two data groups corresponding to the target user, that is, two curves, can be obtained.

[0041] The target user's daily electricity consumption difference curve and line loss change curve are divided into several data sub-segments by hour, and two data sub-segments with the same time are associated to generate sub-segment groups. The data in each sub-segment group is pre-processed and input into the electricity theft identification model to determine whether there is electricity theft in the corresponding time period.

[0042] S43, using the electricity theft identification model to identify whether each sub-segment group has electricity theft behavior, obtain a sub-segment identification result, and set a sub-segment label according to the sub-segment identification result;

[0043] S44. Based on the sub-segment labels corresponding to the preset periods, determine the electricity theft probability of the target user in each sub-segment, and construct an electricity theft probability curve according to the electricity theft probability.

[0044] Specifically, it is determined whether the target user has committed electricity theft in each hour period of any day. The recognition results of the same hour period of all days are comprehensively analyzed to obtain the probability of electricity theft in that hour period. For example, if electricity theft occurs in a certain hour period on 50 out of 100 days, the probability of electricity theft is marked as 0.5.

[0045] Specifically, the above calculation can be used to obtain the probability of electricity theft of the target user in each hour of the day, and then the electricity theft probability curve can be constructed by fitting.

[0046] In a more specific technical solution, S5 includes:

[0047] S51, setting a probability threshold;

[0048] S52, obtaining the intersection point of the probability threshold and the electricity theft probability curve;

[0049] S53, determining whether the probability value between adjacent intersection points is greater than a preset probability threshold, and whether the time difference is greater than a set time;

[0050] S54, if yes, then use the set time corresponding to the current adjacent intersection as the inspection time;

[0051] S55. Build a profile of the target user who steals electricity according to the audit time, and arrange the audit work according to the profile of the user who steals electricity.

[0052] Specifically, the user portrait is the probability of the user stealing electricity in each time period, which facilitates the audit work.

[0053] The present invention determines whether the target user's electricity theft duration is sufficient, and once the duration is sufficient, it is used as the audit time. The audit task is performed within the audit time, which can greatly improve the audit efficiency.

[0054] In a more specific technical solution, an anti-electricity theft identification system includes:

[0055] The power consumption characteristics collection module is used to collect the power consumption data of each original user from the area to be audited, and obtain the power consumption characteristics accordingly;

[0056] The feature analysis module is used to analyze the power consumption characteristics and obtain the power consumption characteristic data change curve of each original user. The feature analysis module is connected to the power consumption characteristic collection module;

[0057] The model building and electricity theft identification module is used to build and use the electricity theft identification model, identify whether the original user has committed electricity theft based on the change curve of electricity consumption characteristic data, and obtain the electricity theft identification result. The model building and electricity theft identification module is connected with the characteristic analysis module;

[0058] The electricity theft probability curve construction module is used to mark the original user who is identified as the electricity theft behavior as the target user according to the electricity theft identification result, and to establish the electricity theft probability curve for each time period based on the change curve of the target user's electricity consumption characteristic data. The electricity theft probability curve construction module is connected with the model construction and electricity theft identification module;

[0059] The user portrait acquisition module is used to construct the electricity theft user portrait of the target user according to the electricity theft probability curve for anti-electricity theft identification. The user portrait acquisition module is connected to the electricity theft probability curve construction module.

[0060] Compared with the prior art, the present invention has the following advantages:

[0061] This application can determine the best audit time period by further mining and analyzing users who have committed electricity theft, which helps to improve the efficiency of the audit and solves the technical problem that the existing technology only identifies electricity theft based on electricity consumption data but cannot effectively audit electricity theft.

[0062] The standard training data in the present invention will be continuously updated and expanded along with the collected data, and the electricity theft identification model will also be continuously updated and trained through the standard training data to improve the availability and reliability of the electricity theft identification model.

[0063] The present invention determines whether the target user's electricity theft duration is sufficient, and once the duration is sufficient, it is used as the audit time. The audit task is performed within the audit time, which can greatly improve the audit efficiency.

[0064] The present invention identifies whether the original user has committed electricity theft based on the data change curve; marks the original user with electricity theft as the target user, and establishes the electricity theft probability curve for each time period based on the data change curve of the target user; constructs the electricity theft user portrait of the target user according to the electricity theft probability curve; the present application can determine the optimal audit time period by further mining and analyzing the users with electricity theft, which helps to improve the audit efficiency.

[0065] The present invention solves the technical problem in the prior art that electricity theft behavior is only identified based on electricity consumption data, but the electricity theft behavior cannot be effectively audited, resulting in low audit efficiency in anti-electricity theft identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 A schematic diagram of basic steps of an anti-electricity theft identification method according to Embodiment 1 of the present invention;

[0067] Figure 2 This is a schematic diagram of specific steps for obtaining the electricity consumption characteristics of each original user according to Embodiment 1 of the present invention;

[0068] Figure 3 This is a schematic diagram of specific steps for establishing a probability curve of electricity theft in each time period according to Embodiment 1 of the present invention;

[0069] Figure 4 This is a schematic diagram of the specific steps of constructing a profile of an electricity theft user according to Embodiment 1 of the present invention. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0071] Example 1

[0072] like Figure 1 As shown, an anti-electricity theft identification method includes the following basic steps:

[0073] S1. Obtain the electricity consumption characteristics of each original user in the area to be audited;

[0074] S2. Analyze the electricity consumption characteristics of each original user in the area to be audited, and obtain the data change curve of each electricity consumption characteristic;

[0075] In this embodiment, the power usage characteristics include, but are not limited to: power usage data and line loss rate;

[0076] After identifying the target user's electricity theft, how to conduct an audit of the theft is a difficult problem. During the audit process, it is best to show or find relevant evidence on site to make the target user deeply understand the problem. Some users know that electricity theft is illegal, so they will steal electricity in real time late at night or when other auditors are not at work.

[0077] The electricity usage characteristics in this embodiment are mainly the actual electricity usage of the original user and the electricity usage displayed by the meter, and the electricity usage difference is the difference between the two; the electricity usage characteristics also include: line loss rate, mainly using the line loss growth rate as an important indicator for evaluating whether there is electricity theft. However, the electricity usage characteristics can also include other data that can be used to evaluate whether there is electricity theft, such as whether the user has ever engaged in electricity theft.

[0078] In this embodiment, after the power consumption characteristics of each original user are obtained through various detection devices, it is necessary to analyze them and generate a data change curve for later use.

[0079] like Figure 2 As shown, in this embodiment, the step S2 of obtaining the electricity consumption characteristics of each original user also includes the following specific steps:

[0080] S21, calculating the electricity consumption difference based on the electricity consumption data;

[0081] S22. Calculate the line loss growth rate based on the line loss rate;

[0082] S23. Taking time as the independent variable, the electricity consumption difference and the line loss growth rate are used to construct the electricity consumption change curve and the line loss change curve respectively.

[0083] In this embodiment, the power consumption difference curve can be understood as the relationship between the original user's power consumption difference and time. The larger the power consumption difference, the greater the probability of power theft. The line loss change curve represents the change of line loss growth rate over time. The larger the line loss growth rate, the greater the probability of power theft.

[0084] S3, identifying whether the original user has committed electricity theft based on the data change curve;

[0085] In this embodiment, the power consumption change curve and the time change curve in the data change curve are time-aligned, and after the time alignment, the feature group is extracted according to the set time period; specifically, the feature group includes the mean of the power consumption difference and the mean of the line loss change;

[0086] In this embodiment, the feature group is pre-processed and then input into the electricity theft identification model to obtain the electricity theft identification result; specifically, the electricity theft identification model is constructed based on the artificial intelligence model;

[0087] In this embodiment, in the construction operation of the aforementioned electricity theft identification model based on the artificial intelligence model, standard training data is obtained; specifically, the standard training data includes standard input data consistent with the content attributes of the feature group, and labeled output data corresponding to the standard input data and used to indicate whether there is electricity theft behavior; an artificial intelligence model is constructed; specifically, the artificial intelligence model is trained by the standard training data, and the trained artificial intelligence model is marked as an electricity theft identification model; specifically, the artificial intelligence model includes a BP neural network model or a RBF neural network model.

[0088] Standard training data is the electricity consumption data and line loss data collected after the previous electricity theft occurred. After screening and marking these data, standard training data is generated. Standard training data will also be continuously updated and expanded with the collected data, and the electricity theft identification model will be continuously updated and trained through standard training data to improve the availability and reliability of the electricity theft identification model.

[0089] Artificial intelligence models are mainly BP neural network models or RBF neural network models, which have strong linear fitting capabilities. These models are continuously trained through standard training data, and when all data meet the requirements, the electricity theft recognition model can be obtained.

[0090] In this embodiment, an example is given to illustrate the process of determining whether electricity theft occurs:

[0091] Assume that in a certain time period A, the average power consumption difference corresponding to time period A is S1, and the average line loss growth rate is S2; after normalization and other preprocessing, S1 and S2 are integrated into an array [S1, S2];

[0092] The array [S1, S2] is input into the constructed electricity theft identification model for identification analysis to obtain the corresponding identification result. The identification result can be used to assist in determining whether the original user has committed electricity theft.

[0093] S4. According to the identification result of the electricity theft model, the original users who are identified to have committed electricity theft are marked as target users, and the electricity theft probability curves for each time period are established based on the data change curves of the target users;

[0094] like Figure 3As shown, in this embodiment, step S4 of establishing the electricity theft probability curve for each time period also includes the following specific steps:

[0095] S41, dividing the data change curve corresponding to the target user into a plurality of data segments on a daily basis, and dividing the plurality of data segments into two data groups according to the corresponding curves;

[0096] S42, dividing the data segments in the two data groups into a plurality of data sub-segments according to a set period, and associating the data sub-segments in the two data groups according to time to generate sub-segment groups;

[0097] S43, identifying whether each field group has electricity theft behavior through the electricity theft identification model, and setting a sub-segment label according to the identification result; specifically, the period interval can be set to, for example, less than 2 hours;

[0098] S44: constructing a power theft probability curve based on at least two sub-segment tags corresponding to each set period.

[0099] In this embodiment, several sub-segment labels of each set period are extracted in sequence, and the probability of electricity theft within the set period is calculated through no less than 2 sub-segment labels; and an electricity theft probability curve is constructed with each set period as an independent variable and the corresponding electricity theft probability as a dependent variable.

[0100] In this embodiment, the data change curve corresponding to the target user is divided into several data segments on a daily basis, and the several data segments are divided into two data groups according to the corresponding curves; the data segments in the two data groups are divided into several data sub-segments according to the set period, and the data sub-segments in the two data groups are associated according to time to generate sub-segment groups; the electricity theft identification model is used to identify whether there is electricity theft behavior in each field group, and the sub-segment label is set according to the identification result; wherein the set period is less than 2 hours; and the electricity theft probability curve is constructed based on the several sub-segment labels corresponding to each set period.

[0101] In this embodiment, an example is given to illustrate:

[0102] The data variation curves corresponding to the target user for several days are divided into several data segments according to the day, and each data segment includes the target user's daily power consumption difference curve and line loss variation curve, so two data groups corresponding to the target user, that is, two curves, can be obtained.

[0103] The target user's daily electricity consumption difference curve and line loss change curve are divided into several data sub-segments by hour, and two data sub-segments with the same time are associated to generate sub-segment groups. The data in each sub-segment group is pre-processed and input into the electricity theft identification model to determine whether there is electricity theft in the corresponding time period.

[0104] In this way, it can be determined whether the target user has committed electricity theft in each hour period of any day. The probability of electricity theft in the hour period can be obtained by comprehensively analyzing the recognition results of the same hour period in all days. For example, if electricity theft occurs in a certain hour period on 50 days out of 100 days, the probability of electricity theft is marked as 0.5.

[0105] Through the above calculation, the probability of electricity theft by the target user in each hour of the day can be obtained, and then the electricity theft probability curve can be fitted and constructed.

[0106] S5. Construct an electricity theft user profile of the target user based on the electricity theft probability curve.

[0107] like Figure 4 As shown in the capital embodiment, step S5 of constructing a profile of a power theft user also includes the following specific steps:

[0108] S51, setting a probability threshold;

[0109] S52, obtaining the intersection point of the probability threshold and the electricity theft probability curve;

[0110] S53, determining whether the probability value between adjacent intersections is greater than a probability threshold, and whether the time difference is greater than a set time;

[0111] S54, if yes, then the time corresponding to the adjacent intersection is used as the inspection time;

[0112] S55. Build a profile of the target user who steals electricity according to the audit time, and arrange the audit work according to the profile of the user who steals electricity.

[0113] In summary, this application can determine the best audit time period by further mining and analyzing users who have committed electricity theft, which helps to improve the efficiency of the audit and solves the technical problem that the prior art only identifies electricity theft based on electricity consumption data but cannot effectively audit electricity theft.

[0114] The standard training data in the present invention will be continuously updated and expanded along with the collected data, and the electricity theft identification model will also be continuously updated and trained through the standard training data to improve the availability and reliability of the electricity theft identification model.

[0115] The present invention determines whether the target user's electricity theft duration is sufficient, and once the duration is sufficient, it is used as the audit time. The audit task is performed within the audit time, which can greatly improve the audit efficiency.

[0116] The present invention solves the technical problem in the prior art that electricity theft behavior is only identified based on electricity consumption data, but the electricity theft behavior cannot be effectively audited, resulting in low audit efficiency in anti-electricity theft identification.

[0117] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An anti-electricity theft identification method, characterized in that: The method comprises: S1. Collect the electricity consumption data of each original user from the area to be audited, and obtain the electricity consumption characteristics; S2. Analyze the electricity consumption characteristics to obtain a change curve of the electricity consumption characteristic data of each of the original users; S3, constructing and using an electricity theft identification model, based on the electricity consumption characteristic data change curve, identifying whether the original user has committed electricity theft, and obtaining an electricity theft identification result; S4. According to the electricity theft identification result, the original user identified as the electricity theft behavior is marked as a target user, and based on the electricity consumption characteristic data change curve of the target user, an electricity theft probability curve for each time period is established; S5. Build an electricity theft user portrait of the target user based on the electricity theft probability curve for anti-electricity theft identification.

2. The anti-electricity theft identification method according to claim 1, characterized in that: In S1, the electricity usage characteristics include: the actual electricity usage, the meter-displayed electricity usage, the line loss rate and the user's electricity theft history data of the original user, and the electricity usage difference between the actual electricity usage and the meter-displayed electricity usage is obtained.

3. The anti-electricity theft identification method according to claim 1, characterized in that: The S1 includes: S11, calculating a power consumption difference based on the power consumption data of the original user; S12, calculating the line loss growth rate from the line loss rate in the power consumption characteristics; S13, taking time as an independent variable, and constructing a power consumption change curve and a line loss change curve according to the power consumption difference and the line loss growth rate, respectively.

4. The anti-electricity theft identification method according to claim 1, characterized in that: The S3 includes: S31, performing time registration on the power consumption change curve and the time change curve in the data change curve; S32, extracting feature groups according to a preset time period; S33, constructing the electricity theft identification model based on a preset artificial intelligence model; S34, preprocessing the feature group, obtaining and inputting preprocessed features into an electricity theft identification model, and obtaining the electricity theft identification result.

5. The anti-electricity theft identification method according to claim 4, characterized in that: In S32, the feature group includes: a mean value of power consumption difference and a mean value of line loss change.

6. The anti-electricity theft identification method according to claim 4, characterized in that: The S33 further includes: S331, obtaining standard training data, and updating the standard training data according to the electricity consumption data; wherein the standard training data includes: standard input data with consistent feature group attributes, and output data with electricity theft behavior annotations; S332: Use the standard training data to train the artificial intelligence model to obtain and mark the artificial intelligence model and determine the electricity theft identification model.

7. The anti-electricity theft identification method according to claim 4, characterized in that: The S34 includes: S341. For a preset time period, obtain the corresponding average power consumption difference and average line loss growth rate; S342, integrating the average of the power consumption difference and the average of the line loss growth rate to obtain an integrated array; S343: input the integrated array into the electricity theft identification model to perform identification analysis to obtain the electricity theft identification result.

8. The anti-electricity theft identification method according to claim 1, characterized in that: The S4 includes: S41, dividing the data change curve corresponding to the target user into at least two data segments according to a preset unit time, and dividing the data segment into at least two data groups according to the data change curve; S42, dividing the data segments in the data group according to a preset period to obtain data sub-segments, and associatively processing the data sub-segments according to a preset time to generate a sub-segment group; S43, using the electricity theft identification model to identify whether each sub-segment group has the electricity theft behavior, obtain a sub-segment identification result, and set a sub-segment label according to the sub-segment identification result; S44. Determine the electricity theft probability of the target user in each of the sub-segments based on the sub-segment labels corresponding to each preset period, and construct the electricity theft probability curve according to the electricity theft probability.

9. The anti-electricity theft identification method according to claim 1, characterized in that: The S5 includes: S51, setting a probability threshold; S52, obtaining the intersection point of the probability threshold and the electricity theft probability curve; S53, determining whether the probability value between the adjacent intersection points is greater than a preset probability threshold, and whether the time difference is greater than a set time; S54, if yes, taking the set time corresponding to the current adjacent intersection as the inspection time; S55. Construct the electricity theft user portrait of the target user according to the audit time, and arrange the audit work according to the electricity theft user portrait. 10.An anti-electricity theft identification system, characterized in that: The system comprises: The power consumption characteristics collection module is used to collect the power consumption data of each original user from the area to be audited, and obtain the power consumption characteristics accordingly; A feature analysis module, used to analyze the power consumption features and obtain a power consumption feature data change curve of each of the original users, the feature analysis module is connected to the power consumption feature acquisition module; A model building and electricity theft identification module is used to build and use an electricity theft identification model to identify whether the original user has committed electricity theft based on the electricity consumption characteristic data change curve, and obtain an electricity theft identification result. The model building and electricity theft identification module is connected to the characteristic analysis module; An electricity theft probability curve construction module is used to mark the original user identified as the electricity theft behavior as the target user according to the electricity theft identification result, and to establish the electricity theft probability curve for each time period based on the electricity consumption characteristic data change curve of the target user. The electricity theft probability curve construction module is connected to the model construction and electricity theft identification module; The user portrait acquisition module is used to construct an electricity theft user portrait of the target user according to the electricity theft probability curve for anti-electricity theft identification. The user portrait acquisition module is connected to the electricity theft probability curve construction module.

Images