Big data-based electricity larceny detection method and system

Abstract

The invention provides an electricity stealing detection method and system based on big data, and the method comprises the steps: obtaining power utilization data and line loss data corresponding to a target user through a detection device, and extracting the power utilization credit data of the target user; wherein the power consumption data comprises actual power consumption data and theoretical power consumption data, and the line loss data is used for calculating a line loss growth rate; constructing a power utilization error curve according to the power utilization data, and constructing a line loss change curve according to the line loss data; and performing time registration on the electricity utilization error curve, the line loss change curve and the user credit data, extracting electricity larceny characteristics from a registration result, and inputting the electricity larceny characteristics into an electricity larceny detection model to perform electricity larceny identification. According to the invention, the technical problem that the accuracy of the difference between the primary side power consumption data and the secondary side power consumption data is low and the judgment precision of the electricity stealing behavior is restricted is solved.

Description

Technical Field

[0001] The present invention relates to the field of electricity theft prevention, and in particular to an electricity theft detection method and system based on big data. Background Art

[0002] As the demand for electricity continues to increase, the electricity sales market continues to expand. In order to conveniently and accurately measure electricity and calculate electricity charges, smart meters are widely used in the power sector. However, some electricity users, driven by profit, use various means to steal electricity, resulting in a large difference between the amount of electricity read in the smart meter and the actual amount of electricity used, which seriously interferes with the normal operation of the power grid.

[0003] The invention patent application with publication number CN117110702A discloses a power theft detection system, which includes: an electric energy data acquirer arranged on the primary side of the power user's transformer, used to acquire the primary side power consumption data of the power user's transformer, and send it to the electric energy data receiver arranged on the secondary side of the power user's transformer through wireless transmission; a collection terminal, used to collect the primary side power consumption data received by the power data receiver, and to acquire the secondary side power consumption data of the power user's transformer collected by the power meter, and upload it to the main control machine through wireless transmission; the main control machine is used to determine whether the power user has committed power theft according to the received primary side power consumption data and secondary side power consumption data.

[0004] The above system determines whether there is electricity theft by comparing the primary-side electricity consumption data with the secondary-side electricity consumption data. However, electricity needs to pass through various power equipment during the transmission process, and the power equipment will be affected by the environment, and the degree of influence is not the same. This will affect the accuracy of the difference between the primary-side electricity consumption data and the secondary-side electricity consumption data, and thus cannot guarantee whether the judgment of electricity theft is accurate.

[0005] The existing invention patent application document "Anti-electricity theft intelligent early warning method, system, electronic device and storage medium" with publication number CN116862116A, the existing method includes: according to the historical electricity theft user electricity consumption data of the selected area, the historical electricity theft user electricity consumption data is classified to obtain typical electricity theft patterns and corresponding electricity consumption data; the electricity consumption data of the users to be tested in the area are compared with the electricity consumption data of each typical electricity theft pattern, and the users who meet the preset similarity conditions are determined as abnormal users; the abnormal user electricity consumption data is input into the encoding and decoding model obtained by training the normal user electricity consumption data to obtain the corresponding decoding data, and the users whose abnormal user electricity consumption data and the corresponding decoding data meet the preset difference conditions are determined as verification objects.

[0006] The existing invention patent application document with publication number CN109146705A, "A method for detecting electricity theft by reducing the dimension of electricity characteristic indicators and using an extreme learning machine algorithm", includes: analyzing the historical electricity consumption data of users, extracting characteristic indicators for dimension reduction, and combining local outlier factors and an extreme learning machine algorithm to detect electricity theft. First, the user's load data is classified, four indicators for measuring the load curve are proposed, and characteristic variables are obtained; secondly, the extracted characteristic variables are reduced in dimension based on the classification, and the local outlier factors are used to screen out abnormal electricity users; finally, the electricity theft discrimination indicators are used to extract the principal components, and the electricity theft discrimination indicator data after the principal components of abnormal electricity users are extracted is used as a training sample to input the model.

[0007] The existing invention patent application document with publication number CN116933118A is "A method and system for identifying electricity theft based on an improved parasitic predation optimization algorithm". The existing method includes: exporting electricity consumption data of all users in the substation area, extracting the typical daily load curve of each user, classifying the typical daily load curve of each user to form a typical daily load curve for each type of user, extracting the daily load curve of the user in the time period to be detected and the typical daily load curve of the class to which it belongs, calculating the matching degree, preliminarily determining users with a matching degree below the threshold as electricity theft users, using the improved parasitic predation optimization algorithm for detection, obtaining a set of optimal electricity theft coefficient sequences, setting the electricity theft coefficient threshold, and further determining the electricity theft users.

[0008] The existing public document "Electricity Theft Behavior Diagnosis Technology Considering User Credit Rating" first analyzes the principles of electricity metering and common electricity theft methods, and summarizes some common means of electricity theft, including undercurrent, phase shift, differential expansion, borrowing zero, high-tech electricity theft methods such as high-power wireless interference and remote control electricity theft, providing a theoretical basis for the subsequent construction of an evaluation system. Secondly, based on the concept of data mining, the user's electricity consumption indicators are analyzed, and normal indicators and abnormal indicators are selected. Through these indicators, abnormal users can be identified and analyzed whether the user steals electricity or leaks electricity. Then a credit rating evaluation index system is established, which includes the user's payment credit and electricity regulation credit, and the user's credit rating is evaluated by fuzzy hierarchical analysis. Finally, based on the analysis of electricity theft indicators and credit rating evaluation, a deep neural network model is used to diagnose users for electricity theft, and it is found that considering the user's credit rating can significantly improve the recognition rate of electricity theft.

[0009] The aforementioned existing solutions do not consider both the line loss data and the user credit data of the target user at the same time; the electricity theft data processing sequence generated by them considers a single factor and does not integrate factors such as electricity usage errors, line loss growth, and user credit, resulting in the detection model used to obtain electricity theft identification results with low accuracy.

[0010] In summary, the existing technology has a technical problem that the difference accuracy between the primary side power consumption data and the secondary side power consumption data is low, which restricts the accuracy of judging the power theft behavior. Summary of the invention

[0011] The technical problem to be solved by the present invention is: how to solve the technical problem in the prior art that the difference accuracy between primary side power consumption data and secondary side power consumption data is low, which restricts the accuracy of judging power theft behavior.

[0012] The present invention adopts the following technical solutions to solve the above technical problems: A method for detecting electricity theft based on big data includes:

[0013] S1. Use pre-installed detection equipment to collect and obtain the target user's electricity consumption data and line loss data, and extract the target user's electricity consumption and credit data;

[0014] S2. construct an electricity consumption error curve based on the electricity consumption data, and construct a line loss change curve based on the line loss data; perform time registration processing on the electricity consumption error curve, the line loss change curve and the user credit data to obtain a registration result;

[0015] S3. Build an electricity theft detection module based on the artificial intelligence model, extract electricity theft features from the registration results according to the preset time frame, input the electricity theft features into the electricity theft detection model, and process to obtain the electricity theft identification results.

[0016] The present invention detects electricity theft based on big data. It introduces line loss data and user credit data of target users on the basis of traditional electricity consumption data, replacing the traditional technology of comparing primary and secondary electricity consumption data to determine electricity theft. It fully considers the environmental impact on electric energy when passing through power equipment, quantitatively studies the above-mentioned impact degree, and ensures the accuracy of the electricity theft judgment operation.

[0017] In a more specific technical solution, in S1, the pre-installed detection equipment includes: power sensor, third-party data acquisition device.

[0018] In a more specific technical solution, the operation of S2 constructing the power consumption error curve also includes:

[0019] S21, extracting actual power consumption data and theoretical power consumption data from the power consumption data, and aligning the corresponding time of the actual power consumption data and the theoretical power consumption data;

[0020] Specifically, in the aforementioned configuration process, time registration is performed for different types of data, and the collection times of various data are aligned;

[0021] S22, calculating the actual and theoretical differences between the actual power consumption data and the theoretical power consumption data at each moment, and marking them as power consumption errors;

[0022] S23. Taking time as the independent variable and electricity consumption error as the dependent variable, construct an electricity consumption error curve.

[0023] Before constructing the power consumption error curve, the present invention needs to time-align the actual power consumption data and the theoretical power consumption data in the power consumption data. This is mainly because the different collection times of the two affect the reliability and accuracy of the power consumption error calculation.

[0024] In a more specific technical solution, the operation of S2 constructing the line loss change curve also includes:

[0025] S21', calculating the line loss rate according to the line loss data;

[0026] S22', calculating the line loss growth rate per preset unit time according to the line loss rate;

[0027] S23', taking time as the independent variable and line loss growth rate as the dependent variable, construct the line loss change curve.

[0028] 1. A method for detecting electricity theft based on big data according to claim 4, characterized in that S21' further comprises:

[0029] S211', extracting the line loss power SD and power supply GD at each moment from the line loss data;

[0030] S212', according to the line loss power SD and the power supply GD, the line loss rate XSL is calculated using the following logic:

[0031] XSL=SD / GD×100%.

[0033] In a more specific technical solution, S22' also includes:

[0034] S221', extracting the line loss rate according to the preset unit time, so as to obtain the ratio of the difference between adjacent line loss rates and the previous line loss rate as the line loss growth rate;

[0035] S222', associating the line loss growth rate with the corresponding time of the next line loss rate in the adjacent line loss rates.

[0036] In a more specific technical solution, in S3, the artificial intelligence models include: BP neural network model and RBF neural network model.

[0037] In a more specific technical solution, in S3, the preset time frame is greater than or equal to the preset unit time.

[0038] In view of the fact that line loss data does not change frequently, the present invention reduces the amount of data processing and improves the efficiency of electricity theft detection by setting a unit time.

[0039] In a more specific technical solution, the operation of extracting the electricity theft feature from the registration result in S3 also includes:

[0040] S31, setting a preset time frame;

[0041] S32, extracting the mean value of power consumption error, the mean value of line loss growth, and the user credit label from the registration results according to the preset time frame;

[0042] S33, integrating and processing the mean value of electricity usage error, the mean value of line loss growth and the user credit label to obtain an electricity theft assessment sequence, and inputting the electricity theft assessment sequence into an electricity theft detection model to obtain an electricity theft identification result.

[0043] Specifically, the aforementioned average power consumption error, average line loss growth and user credit label are integrated and input into the power theft detection model; based on the current credit data and the detected power consumption data, it is determined whether there is power theft. The power consumption data includes but is not limited to: power consumption difference and line loss.

[0044] The present invention extracts the mean value of electricity usage error, the mean value of line loss growth and the user credit label from the registration results according to the set time frame, and integrates them to generate an electricity theft assessment sequence, which is input into the electricity theft detection model to obtain the electricity theft identification result; the present invention takes into account various factors, overcomes the accuracy problem of identifying electricity theft behavior through single electricity usage data, and comprehensively improves the identification result of electricity theft behavior.

[0045] The standard training data of the electricity theft detection model used in the present invention is derived from the real situation, that is, the electricity consumption data and line loss growth rate data corresponding to the electricity theft behavior that has occurred are statistically saved. After data screening and labeling, standard training data is generated. In addition, noise is appropriately introduced during the training process to improve the reliability and accuracy of the electricity theft detection model.

[0046] In a more specific technical solution, a big data-based electricity theft detection system includes:

[0047] The user data collection module is used to collect the target user's power consumption data and line loss data by using the pre-installed detection equipment, and extract the target user's power consumption and credit data;

[0048] The data registration module is used to construct an electricity consumption error curve according to the electricity consumption data and to construct a line loss change curve according to the line loss data; the electricity consumption error curve, the line loss change curve and the user credit data are subjected to time registration processing to obtain a registration result. The data registration module is connected to the user data acquisition module;

[0049] The electricity theft model detection module is used to build an electricity theft detection module based on the artificial intelligence model. According to the preset time frame, the electricity theft features are extracted from the alignment results, the electricity theft features are input into the electricity theft detection model, and the electricity theft identification results are obtained through processing. The electricity theft model detection module is connected to the data alignment module.

[0050] Compared with the prior art, the present invention has the following advantages:

[0051] The present invention detects electricity theft based on big data. It introduces line loss data and user credit data of target users on the basis of traditional electricity consumption data, replacing the traditional technology of comparing primary and secondary electricity consumption data to determine electricity theft. It fully considers the environmental impact on electric energy when passing through power equipment, quantitatively studies the above-mentioned impact degree, and ensures the accuracy of the electricity theft judgment operation.

[0052] Before constructing the power consumption error curve, the present invention needs to time-align the actual power consumption data and the theoretical power consumption data in the power consumption data, taking into full consideration the reliability and accuracy of the power consumption error calculation affected by the different collection times of the two.

[0053] In view of the fact that line loss data does not change frequently, the present invention reduces the amount of data processing and improves the efficiency of electricity theft detection by setting a unit time.

[0054] The present invention extracts the mean value of electricity usage error, the mean value of line loss growth and the user credit label from the registration results according to the set time frame, and integrates them to generate an electricity theft assessment sequence, which is input into the electricity theft detection model to obtain the electricity theft identification result; the present invention takes into account various factors, overcomes the accuracy problem of identifying electricity theft behavior through single electricity usage data, and comprehensively improves the identification result of electricity theft behavior.

[0055] The standard training data of the electricity theft detection model used in the present invention is derived from the real situation, that is, the electricity consumption data and line loss growth rate data corresponding to the electricity theft behavior that has occurred are statistically saved. After data screening and labeling, standard training data is generated. In addition, noise is appropriately introduced during the training process to improve the reliability and accuracy of the electricity theft detection model.

[0056] The present invention introduces line loss data and user credit data of target users; extracts the mean value of electricity usage error, the mean value of line loss growth and the user credit label from the registration result according to the set time frame, and integrates them to generate an electricity theft assessment sequence, which is input into the electricity theft detection model to obtain the electricity theft identification result; the present application takes many factors into consideration, overcomes the accuracy problem of identifying electricity theft behavior through single electricity usage data, and comprehensively improves the identification result of electricity theft behavior.

[0057] The present invention solves the technical problem in the prior art that the difference between primary-side power consumption data and secondary-side power consumption data has low accuracy, which restricts the accuracy of judging power theft. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is a schematic diagram of basic steps of a method for detecting electricity theft based on big data according to Embodiment 1 of the present invention;

[0059] Figure 2 This is a schematic diagram of specific steps for constructing a power consumption error curve according to Embodiment 1 of the present invention;

[0060] Figure 3 This is a schematic diagram of specific steps for constructing a line loss variation curve according to Embodiment 1 of the present invention;

[0061] Figure 4 It is a schematic diagram of specific steps of extracting electricity theft features from the registration results according to Embodiment 1 of the present invention. DETAILED DESCRIPTION

[0062] 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.

[0063] Example 1

[0064] like Figure 1 As shown, a method for detecting electricity theft based on big data includes the following basic steps:

[0065] S1. Obtain the target user's corresponding electricity consumption data and line loss data through the detection equipment, and extract the target user's electricity consumption and credit data;

[0066] In this embodiment, the electricity consumption data includes but is not limited to: actual electricity consumption data and theoretical electricity consumption data, and the line loss data is used to calculate the line loss growth rate.

[0067] S2. Construct an electricity consumption error curve based on the electricity consumption data, and construct a line loss change curve based on the line loss data; perform time registration on the electricity consumption error curve, the line loss change curve, and the user credit data to obtain a registration result;

[0068] In this embodiment, electricity theft features are extracted from the registration results and input into the electricity theft detection model for electricity theft identification;

[0069] like Figure 2 As shown, in this embodiment, the operation of constructing the power consumption error curve also includes the following specific steps:

[0070] S21, extracting actual power consumption data and theoretical power consumption data from the power consumption data, and aligning the corresponding time between the two;

[0071] S22, after time alignment, calculate the difference between the actual power consumption data and the theoretical power consumption data at each moment, and mark it as power consumption error;

[0072] S23. Construct an electricity consumption error curve with time as the independent variable and electricity consumption error as the dependent variable.

[0073] Before constructing the power consumption error curve, it is necessary to time-align the actual power consumption data and the theoretical power consumption data in the power consumption data. This is mainly because the different collection times of the two affect the reliability and accuracy of the power consumption error calculation. After time alignment, the difference between the actual power consumption data and the theoretical power consumption data at each moment is calculated as the power consumption error. The power consumption error curve can be constructed by combining the time corresponding to each power consumption error.

[0074] In this embodiment, the actual power consumption data and the theoretical power consumption data in the power consumption data are measured at different locations. The actual power consumption data refers to the amount of electricity actually used by the target user, and the theoretical power consumption data refers to the data measured by the electric meter at the target user end. If the difference between the two exceeds the normal floating range, the target user may have taken measures to steal electricity at the power consumption end. The detection equipment is used to collect power consumption data and line loss data, which can be various types of power sensors or qualified third-party data collection equipment that has passed quality inspection.

[0075] In this embodiment, the target user's historical electricity theft behavior is also an important consideration for determining whether electricity is currently being stolen, so this application also extracts the target user's electricity consumption and credit data. The electricity consumption and credit data are mainly set based on whether the target user has a history of electricity theft behavior. If the target user has a history of electricity theft, the target user's electricity consumption and credit data is set to 1, otherwise it is set to 0; of course, the credit score can also be used as the user's credit data.

[0076] like Figure 3 As shown, in this embodiment, the operation of constructing the line loss change curve also includes the following specific steps:

[0077] S21', calculating the line loss rate according to the line loss data;

[0078] In this embodiment, the line loss power and power supply at each moment are extracted from the line loss data and marked as SD and GD respectively; the line loss rate XSL is calculated by the following formula:

[0079] XSL=SD / GD×100%

[0080] Among them, SD and GD are line loss and power supply respectively;

[0081] S22', calculating the line loss growth rate per unit time based on the line loss rate;

[0082] In this embodiment, the line loss rate is extracted according to unit time, and the ratio of the difference between adjacent line loss rates and the previous line loss rate is calculated as the line loss growth rate; the line loss growth rate is associated with the corresponding time of the next line loss rate in the adjacent line loss rates.

[0083] S23', construct a line loss variation curve with time as the independent variable and line loss growth rate as the dependent variable.

[0084] In this embodiment, after the line loss rate is calculated, the ratio of the difference between adjacent line loss rates and the previous line loss rate is used as the line loss growth rate. The line loss change curve can be constructed in combination with time.

[0085] The setting of the unit time in this embodiment is mainly to reduce the amount of data processing, and the line loss data does not change very frequently. Therefore, the unit time in this embodiment can be set to 1s, and the line loss growth rate per unit time can be understood as the ratio of the difference between the line loss rate after one second and the line loss rate before one second to the line loss rate before one second.

[0086] S3. Extract electricity theft features from the registration results according to the set time frame, and input the electricity theft features into the electricity theft detection model to obtain electricity theft identification results; specifically, the electricity theft detection model is constructed based on an artificial intelligence model.

[0087] In this embodiment, in the operation of extracting electricity theft features from the registration results, a time frame is set, and the mean electricity usage error, the mean line loss growth, and the user credit label are respectively extracted from the registration results according to the time frame; wherein the duration corresponding to the time frame is greater than or equal to the unit time; the mean electricity usage error, the mean line loss growth, and the user credit label are integrated into an electricity theft evaluation sequence, and the electricity theft evaluation sequence is input into the electricity theft detection model to obtain the electricity theft identification result.

[0088] In the process of constructing the electricity theft detection model of this embodiment, standard training data is obtained; specifically, the standard training data includes a standard input sequence having the same content attributes as the electricity theft assessment sequence, and a standard output sequence for representing the electricity theft identification result; the artificial intelligence model constructed by training the standard training data is obtained to obtain the electricity theft identification sequence; wherein the artificial intelligence model includes a BP neural network model or an RBF neural network model.

[0089] like Figure 4 As shown, in this embodiment, step S3 of extracting electricity theft features from the registration results further includes the following specific steps:

[0090] S31, setting a time frame, and extracting the mean value of power consumption error, the mean value of line loss growth, and the user credit label from the registration result according to the time frame;

[0091] In this embodiment, the duration corresponding to the time frame is greater than or equal to the unit time;

[0092] S32, integrating the average value of electricity usage error, the average value of line loss growth and the user credit label into an electricity theft evaluation sequence, and inputting the electricity theft evaluation sequence into an electricity theft detection model to obtain an electricity theft identification result;

[0093] In this embodiment, a time frame is set, which is greater than or equal to the unit time, for example, 2s. If it is necessary to determine whether the target user currently has electricity theft behavior, the current moment is taken as the end moment of the time frame, a time period is determined, and the mean value of the electricity error and the mean value of the line loss growth in the time period are extracted from the electricity error curve and the line loss change curve. Of course, the user credit label of the target user (in the time period) also needs to be extracted.

[0094] S33, integrating the mean value of electricity usage error, the mean value of line loss growth, and the user credit label into a data sequence so as to be input into the electricity theft detection model;

[0095] S34, inputting the data sequence into the electricity theft detection model to obtain the corresponding recognition result, if the output label is 1, it is determined that the target user has committed electricity theft, and if the output label is 0, there is no electricity theft.

[0096] In this embodiment, since the power consumption error curve and line loss change curve of the present application are continuously updated and correspond to a long period of time, it is possible to identify at any time whether the target user has electricity theft positioning within the time range corresponding to the curve.

[0097] In this embodiment, the user credit tag can be unchanged. For example, if the target user has stolen electricity, the user credit tag is set to 1, otherwise the user credit tag is set to 0. Of course, it can also be variable. For example, if the target user has stolen electricity in the third quarter, if the identified time is the third quarter, the user credit tag is set to 1, otherwise it is set to 0. The user credit tag setting method is selected according to the actual situation.

[0098] In summary, the present invention detects electricity theft based on big data. On the basis of traditional electricity consumption data, it introduces line loss data and user credit data of target users, replacing the traditional technology of comparing primary and secondary electricity consumption data to determine electricity theft. It fully considers the environmental impact on electric energy when passing through power equipment, quantitatively studies the degree of the aforementioned impact, and ensures the accuracy of the electricity theft judgment operation.

[0099] Before constructing the power consumption error curve, the present invention needs to time-align the actual power consumption data and the theoretical power consumption data in the power consumption data, taking into full consideration the reliability and accuracy of the power consumption error calculation affected by the different collection times of the two.

[0100] In view of the fact that line loss data does not change frequently, the present invention reduces the amount of data processing and improves the efficiency of electricity theft detection by setting a unit time.

[0101] The present invention extracts the mean value of electricity usage error, the mean value of line loss growth and the user credit label from the registration results according to the set time frame, and integrates them to generate an electricity theft assessment sequence, which is input into the electricity theft detection model to obtain the electricity theft identification result; the present invention takes into account various factors, overcomes the accuracy problem of identifying electricity theft behavior through single electricity usage data, and comprehensively improves the identification result of electricity theft behavior.

[0102] The standard training data of the electricity theft detection model used in the present invention is derived from the real situation, that is, the electricity consumption data and line loss growth rate data corresponding to the electricity theft behavior that has occurred are statistically saved. After data screening and labeling, standard training data is generated. In addition, noise is appropriately introduced during the training process to improve the reliability and accuracy of the electricity theft detection model.

[0103] The present invention solves the technical problem in the prior art that the difference between primary-side power consumption data and secondary-side power consumption data has low accuracy, which restricts the accuracy of judging power theft.

[0104] 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. A method for detecting electricity theft based on big data, characterized in that: The method comprises: S1. Using pre-installed detection equipment, collect and obtain the target user's electricity consumption data and line loss data, and extract the target user's electricity consumption and credit data; S2. constructing an electricity usage error curve according to the electricity usage data, and constructing a line loss change curve according to the line loss data; performing time registration processing on the electricity usage error curve, the line loss change curve and the user credit data to obtain a registration result; S3. Building an electricity theft detection module based on the artificial intelligence model, extracting electricity theft features from the alignment results according to a preset time frame, inputting the electricity theft features into the electricity theft detection model, and processing to obtain electricity theft identification results.

2. The method for detecting electricity theft based on big data according to claim 1, characterized in that: In S1, the preset detection equipment includes: a power sensor and a third-party data acquisition device.

3. The method for detecting electricity theft based on big data according to claim 1, characterized in that: The operation of constructing the power consumption error curve in S2 further includes: S21, extracting actual power consumption data and theoretical power consumption data from the power consumption data, and aligning and processing corresponding times of the actual power consumption data and the theoretical power consumption data; S22, calculating the actual and theoretical differences between the actual power consumption data and the theoretical power consumption data at each moment, and marking them as power consumption errors; S23. Taking time as the independent variable and the power consumption error as the dependent variable, construct the power consumption error curve.

4. The method for detecting electricity theft based on big data according to claim 1, characterized in that: The operation of constructing the line loss variation curve in S2 further includes: S21', calculating the line loss rate according to the line loss data; S22', calculating the line loss growth rate per preset unit time according to the line loss rate; S23', constructing the line loss variation curve by taking time as the independent variable and the line loss growth rate as the dependent variable.

5. The method for detecting electricity theft based on big data according to claim 4, characterized in that: The S21' further includes: S211', extracting the line loss power SD and power supply GD at each moment from the line loss data; S212', according to the line loss power SD and the power supply GD, the line loss rate XSL is calculated using the following logic: XSL=SD / GD×100%.

6. The method for detecting electricity theft based on big data according to claim 4, characterized in that: The S22' further includes: S221', extracting the line loss rate according to the preset unit time, so as to obtain the ratio of the difference between adjacent line loss rates and the previous line loss rate as the line loss growth rate; S222', associating the line loss growth rate with the corresponding time of the next line loss rate among the adjacent line loss rates.

7. The method for detecting electricity theft based on big data according to claim 1, characterized in that: In S3, the artificial intelligence model includes: a BP neural network model and a RBF neural network model.

8. The method for detecting electricity theft based on big data according to claim 1, characterized in that: In S3, the preset time frame is greater than or equal to a preset unit time.

9. The method for detecting electricity theft based on big data according to claim 1, characterized in that: The operation of extracting the electricity theft feature from the registration result in S3 further includes: S31, setting the preset time frame; S32, extracting the mean value of power consumption error, the mean value of line loss growth and the user credit label from the registration result according to the preset time frame; S33, integrating and processing the mean value of the electricity usage error, the mean value of the line loss growth, and the user credit label to obtain an electricity theft evaluation sequence, and inputting the electricity theft evaluation sequence into the electricity theft detection model to detect and obtain the electricity theft identification result.

10. A power theft detection system based on big data, characterized in that: The system comprises: A user data collection module is used to collect and obtain the target user's electricity consumption data and line loss data by using pre-installed detection equipment, and extract the target user's electricity consumption and credit data; A data registration module is used to construct an electricity consumption error curve according to the electricity consumption data, and to construct a line loss change curve according to the line loss data; the electricity consumption error curve, the line loss change curve and the user credit data are subjected to time registration processing to obtain a registration result, and the data registration module is connected to the user data acquisition module; The electricity theft model detection module is used to build an electricity theft detection module based on the artificial intelligence model, extract electricity theft features from the alignment results according to a preset time frame, input the electricity theft features into the electricity theft detection model, and process to obtain electricity theft identification results. The electricity theft model detection module is connected to the data alignment module.

Images