Electricity theft detection method and system

Abstract

The present invention discloses a method and system for detecting electricity theft, the method comprising: obtaining electricity consumption data of a user to be detected on the same day; wherein the electricity consumption data comprises: an electricity consumption time series, a first electricity consumption statistical feature and an electricity parameter time series; calculating the similarity between the electricity consumption data of the user to be detected and the reference electricity consumption data of the electricity consumption mode category to which it belongs; when the similarity exceeds a preset range, determining that the user to be detected is an abnormal electricity user; based on the electricity parameter time series of the abnormal electricity user, obtaining characteristic data and operation data of the electrical equipment; inputting the characteristic data into an equipment identification model to obtain the type of the electrical equipment; and determining whether the abnormal electricity user has electricity theft behavior by comparing the operation data with the reference operation data of the type of the electrical equipment. The present invention can effectively improve the accuracy of detecting the user's electricity theft behavior by making multi-level judgments on the user's electricity consumption behavior through multi-dimensional electricity consumption data.

Description

Technical Field

[0001] The present invention relates to the field of electric power technology, and in particular to an electricity theft detection method and system. Background Art

[0002] As the demand for electricity in various industries continues to expand, the problem of illegal electricity theft is becoming increasingly serious. This not only damages the economic interests of power supply companies, but also poses a threat to the safe and stable operation of the power grid. Therefore, how to accurately identify users' electricity theft behavior has become a problem of great concern to power supply companies.

[0003] However, electricity theft and leakage are highly concealed and deceptive, and the traditional single data source detection method has been unable to fully and accurately characterize the user's real electricity usage behavior, resulting in high underreporting and false alarm rates in the identification of violations. For example, simply relying on abnormal changes in electricity consumption to judge electricity theft has a high risk of misjudgment, because in the power system, the electricity usage behavior of normal users itself also has a certain degree of volatility. Summary of the invention

[0004] The purpose of the embodiments of the present invention is to provide a method and system for detecting electricity theft, which can effectively improve the accuracy of detecting electricity theft by making multi-level judgments on the electricity usage behavior of users through multi-dimensional electricity usage data.

[0005] A first aspect of the present invention provides a method for detecting electricity theft, comprising:

[0006] Obtaining the power consumption data of the user to be detected on the day; wherein the power consumption data includes: a power consumption time series, a first power consumption statistical feature and a power parameter time series;

[0007] Calculating the similarity between the power consumption data of the user to be detected and the reference power consumption data of the power consumption mode category to which it belongs;

[0008] When the similarity exceeds a preset range, determining that the user to be detected is an abnormal electricity user;

[0009] Based on the power parameter time series of the abnormal power user, acquiring characteristic data and operation data of the power-consuming equipment;

[0010] Inputting the characteristic data into a device identification model to obtain the type of electrical equipment;

[0011] By comparing the operating data with reference operating data of the type of electrical equipment, it is determined whether the abnormal electricity user has committed electricity theft.

[0012] Optionally, all the power usage mode categories are obtained by the following steps:

[0013] Acquire a first data set of a user group; wherein the first data set includes: geographic areas, electricity consumption types, second electricity consumption statistical characteristics in multiple time periods, average daily electricity consumption time series and average daily power parameter time series of several individual users; the user group includes the user to be detected;

[0014] Based on a preset clustering algorithm and the first data set, all the individual users are clustered and grouped to obtain a plurality of power consumption pattern categories.

[0015] Optionally, all the power usage mode categories are obtained by the following steps:

[0016] Acquire a second data set of a user group; wherein the second data set includes: geographic areas, electricity usage types, electricity voltage levels, third electricity usage statistical characteristics within multiple time periods, and electricity usage time period statistical distribution data of a number of individual users; the user group includes the user to be detected;

[0017] Based on a preset clustering algorithm and the second data set, all the individual users are clustered and grouped to obtain a plurality of power consumption pattern categories.

[0018] Optionally, the reference power consumption data is obtained by the following steps:

[0019] Based on the power usage mode category, determining a corresponding user group;

[0020] According to the specified time period, calculate the candidate power consumption data of the user group under different time types; wherein the time type includes: working day, weekend or holiday; the candidate power consumption data includes: the fourth power consumption statistical feature, the average daily power consumption time series and the average daily power parameter time series;

[0021] Based on the time type of the to-be-detected user on that day, corresponding candidate power usage data is selected as the reference power usage data.

[0022] Optionally, the power parameter time series includes: a power time series, a current time series, a voltage time series and a power load time series.

[0023] Optionally, the acquiring characteristic data of the electric device based on the power parameter time series of the abnormal electric user includes:

[0024] Extracting a current time subsequence corresponding to the operation period of the electrical equipment from the current time series;

[0025] Based on the current time subsequence, obtaining a current change value when the electrical equipment is started, and a current average value and a current standard deviation during steady-state operation;

[0026] Performing harmonic analysis on the current time subsequence to obtain multiple harmonic components and corresponding harmonic amplitudes;

[0027] Characteristic data of the electrical equipment are obtained based on the current change value, the current average value, the current standard deviation, the harmonic component and the harmonic amplitude.

[0028] Optionally, the acquiring operation data of the electric equipment based on the power parameter time series of the abnormal electric user includes:

[0029] Extracting the current time subsequence, voltage time subsequence and power time subsequence corresponding to the electrical equipment in the operation period from the current time series, the voltage time series and the power time series respectively;

[0030] Calculating a phase difference between the current time subsequence and the voltage time subsequence;

[0031] Based on the power time subsequence and the phase difference, calculating the active power and reactive power of the electrical equipment under different operating states;

[0032] Based on the power load time series, obtaining the power load of the power-consuming equipment in different time periods;

[0033] Based on the phase difference, the active power, the reactive power and the power load, the operation data of the power-consuming device is obtained.

[0034] Optionally, the method further comprises:

[0035] Based on the similarity, determining a probability of abnormal power consumption;

[0036] Obtaining payment information and abnormal power usage frequency of the user to be detected, and determining the violation risk level;

[0037] Obtaining the load fluctuation rate and power consumption fluctuation rate of the user to be detected within a preset time period, and determining the power consumption risk level;

[0038] Based on the abnormal power usage probability, the violation risk level and the power usage risk level, a first profile of the user to be detected is established.

[0039] Optionally, the method further comprises:

[0040] Based on the payment information, obtaining corresponding payment characteristics;

[0041] Based on the abnormal power consumption probability, the payment characteristics and the power consumption fluctuation rate, a second portrait of the user to be detected is calculated.

[0042] A second aspect of the present invention provides an electricity theft detection system, comprising:

[0043] An electricity consumption data acquisition module is used to acquire the electricity consumption data of the user to be detected on the day; wherein the electricity consumption data includes: a time series of electricity consumption, a first electricity consumption statistical feature and a time series of electricity parameters;

[0044] A similarity calculation module, used to calculate the similarity between the power consumption data of the user to be detected and the reference power consumption data of the power consumption mode category to which it belongs;

[0045] An abnormal power consumption determination module, configured to determine that the user to be detected is an abnormal power consumption user when the similarity exceeds a preset range;

[0046] An equipment data acquisition module, used to acquire characteristic data and operation data of electric equipment based on the time series of electric power parameters of the abnormal electric user;

[0047] An electric device identification module, used to input the characteristic data into an equipment identification model to obtain the type of electric device;

[0048] The electricity stealing behavior judgment module is used to judge whether the abnormal electricity user has committed electricity stealing behavior by comparing the operation data with reference operation data of the type of electrical equipment.

[0049] Compared with the prior art, the embodiment of the present invention provides a method and system for detecting electricity theft, the method comprising: obtaining electricity consumption data of the user to be detected on the day; wherein the electricity consumption data comprises: electricity consumption time series, first electricity consumption statistical features and electricity parameter time series; calculating the similarity between the electricity consumption data of the user to be detected and the reference electricity consumption data of the electricity consumption mode category to which it belongs; when the similarity exceeds a preset range, determining that the user to be detected is an abnormal electricity user; based on the electricity parameter time series of the abnormal electricity user, obtaining characteristic data and operation data of the electrical equipment; inputting the characteristic data into the equipment identification model to obtain the type of the electrical equipment; and determining whether the abnormal electricity user has electricity theft behavior by comparing the operation data with the reference operation data of the type of the electrical equipment. The present invention can effectively improve the accuracy of detecting the user's electricity theft behavior by making multi-level judgments on the user's electricity consumption behavior through multi-dimensional electricity consumption data. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a flow chart of an embodiment of the electricity theft detection method provided by the present invention;

[0051] Figure 2 is a schematic diagram of an embodiment of obtaining characteristic data of an electric device provided by the present invention;

[0052] Figure 3is a schematic diagram of an embodiment of obtaining operation data of an electric device provided by the present invention;

[0053] Figure 4 It is a structural schematic diagram of an embodiment of the electricity theft detection system provided by the present invention. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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 technical field without creative work are within the scope of protection of the present invention.

[0055] See also Figure 1 , is a flow chart of an embodiment of the electricity theft detection method provided by the present invention.

[0056] The first embodiment of the present invention provides a method for detecting electricity theft, including steps S1 to S6, which are specifically as follows:

[0057] Step S1: obtaining the power consumption data of the user to be detected on the day; wherein the power consumption data includes: power consumption time series, first power consumption statistical characteristics and power parameter time series;

[0058] Step S2: Calculate the similarity between the power consumption data of the user to be detected and the reference power consumption data of the power consumption mode category to which it belongs;

[0059] Step S3: when the similarity exceeds a preset range, determining that the user to be detected is an abnormal electricity user;

[0060] Step S4: based on the power parameter time series of the abnormal power user, acquiring characteristic data and operation data of the power-consuming equipment;

[0061] Step S5: inputting the characteristic data into the equipment identification model to obtain the type of electrical equipment;

[0062] Step S6: By comparing the operating data with reference operating data of the type of electrical equipment, it is determined whether the abnormal electricity user has committed electricity theft.

[0063] It should be noted that the electricity consumption data in the above step S1 is obtained after a series of data cleaning operations, that is, it is normalized electricity consumption data obtained after deleting abnormal values, filling missing values, and data standardization, which can provide effective data support for subsequent analysis of electricity consumption anomalies and judgment of electricity theft.

[0064] The embodiment of the present invention can obtain the user's daily electricity consumption time series through the smart meter to further obtain the first electricity consumption statistical characteristics; wherein the first electricity consumption statistical characteristics at least include: the total electricity consumption, average electricity consumption, electricity consumption standard deviation, maximum electricity consumption and minimum electricity consumption of the day. The embodiment of the present invention can also obtain the user's daily power parameter time series through the smart circuit breaker; wherein the power parameter time series at least includes: the power time series, current time series, voltage time series and power load time series of the daily electricity consumption. Compared with the traditional single data source detection method, the embodiment of the present invention integrates multi-source data to obtain multi-dimensional electricity consumption data, which can comprehensively characterize the user's electricity consumption behavior characteristics.

[0065] In addition, in addition to obtaining the user's electricity usage data, the embodiment of the present invention also obtains the user's basic information, such as user number, user name, electricity usage address, electricity usage type (such as residential electricity usage, commercial electricity usage, and industrial electricity usage, etc.), geographical area, payment information, and abnormal electricity usage frequency.

[0066] It is worth noting that the embodiment of the present invention can identify abnormal electricity users for the first time by calculating the similarity between the electricity consumption data of the user to be detected and the reference electricity consumption data provided by the electricity consumption mode category to which it belongs. In order to further determine whether the abnormal electricity user has committed electricity theft, the characteristic data and operation data of the electrical equipment are first extracted from the power parameter time series. Then, the characteristic data is input into the equipment recognition model constructed based on the classification algorithm to effectively identify the type of electrical equipment. Then, the operation data of the current electrical equipment is compared with the reference operation data of the type of electrical equipment under normal electricity consumption conditions to determine whether the user to be detected has committed electricity theft. If the deviation between the two is greater than the preset threshold, it indicates that the user to be detected has committed electricity theft. Therefore, the embodiment of the present invention can effectively improve the accuracy of detecting the user's electricity theft by making multi-level judgments on the user's electricity consumption behavior through multi-dimensional electricity consumption data.

[0067] In an optional embodiment, all the power usage mode categories are obtained by the following steps:

[0068] Acquire a first data set of a user group; wherein the first data set includes: geographic areas, electricity consumption types, second electricity consumption statistical characteristics in multiple time periods, average daily electricity consumption time series and average daily power parameter time series of several individual users; the user group includes the user to be detected;

[0069] Based on a preset clustering algorithm and the first data set, all the individual users are clustered and grouped to obtain a plurality of power consumption pattern categories.

[0070] It should be noted that the user group includes multiple individual users, one of which is a user to be detected. Therefore, once the user group is clustered and grouped, the power consumption pattern category of the user to be detected can be obtained. The second power consumption statistical characteristics within multiple time periods include: total power consumption, average power consumption, maximum power consumption, minimum power consumption and standard deviation for each month, quarter and year. If the amount of calculation is to be reduced, only the second power consumption statistical characteristics for each quarter and year can be used, which is of course also applicable to other embodiments of the present invention. Similarly, the average daily power consumption time series and the average daily power parameter time series are also calculated in different time periods (such as month, quarter and year). The preset clustering algorithm in the present invention is at least any one of K-means and DBSCAN.

[0071] It is worth noting that the geographical area can reflect the economic development level and regional industrial structure of the user's region to a certain extent; the use of both quarterly data and geographical areas can reflect the climate characteristics of the user's region.

[0072] The embodiment of the present invention inputs the first data set into a clustering model constructed by a preset clustering algorithm, and the clustering model divides a number of individual users into different groups, i.e., a plurality of power consumption pattern categories, according to the similarity values ​​(such as Euclidean distance) between the individual user data. Each power consumption pattern category represents a group of users with similar power consumption behaviors / patterns, such as a group of low-energy-consuming residential users and a group of high-energy-consuming users of small and medium-sized industrial enterprises.

[0073] The embodiment of the present invention will also continuously monitor and update the first data set of the user group, and interact with users through APP push, telephone follow-up, etc., to correct user individual data and information to adjust and optimize the clustering results to cope with changes and migrations in user power consumption patterns.

[0074] In an optional embodiment, all the power usage mode categories are obtained by the following steps:

[0075] Acquire a second data set of a user group; wherein the second data set includes: geographic areas, electricity usage types, electricity voltage levels, third electricity usage statistical characteristics within multiple time periods, and electricity usage time period statistical distribution data of a number of individual users; the user group includes the user to be detected;

[0076] Based on a preset clustering algorithm and the second data set, all the individual users are clustered and grouped to obtain a plurality of power consumption pattern categories.

[0077] It should be noted that the types of electricity consumption include: residential electricity consumption, commercial electricity consumption or industrial electricity consumption. The electricity voltage level can reflect the needs of the user's electrical equipment; the third electricity consumption statistical characteristics within multiple time periods include: monthly, quarterly, and annual total electricity consumption, average electricity consumption, maximum electricity consumption, minimum electricity consumption, and electricity consumption standard deviation. The statistical distribution data of electricity consumption time periods within multiple time periods include: the average electricity consumption distribution of peak electricity consumption periods, trough electricity consumption periods, and level electricity consumption periods corresponding to the month, quarter, and year. The data of each individual user in the above second data set can preliminarily determine the industry type (or industry characteristics) and electricity consumption behavior habits of the user. The reference electricity consumption data for the electricity mode category can characterize the electricity consumption behavior benchmark of the industry.

[0078] It is worth noting that the different time periods in the present invention can be determined according to actual needs or computer computing power. For example, in order to save computer resources, only the quarterly and annual electricity consumption statistical characteristics and electricity consumption time period statistical distribution data of individual users are calculated.

[0079] In an optional embodiment, the reference power consumption data is obtained by the following steps:

[0080] Based on the power usage mode category, determining a corresponding user group;

[0081] According to the specified time period, calculate the candidate power consumption data of the user group under different time types; wherein the time type includes: working day, weekend or holiday; the candidate power consumption data includes: the fourth power consumption statistical feature, the average daily power consumption time series and the average daily power parameter time series;

[0082] Based on the time type of the to-be-detected user on that day, corresponding candidate power usage data is selected as the reference power usage data.

[0083] It should be noted that the specified time period is at least one of monthly, quarterly and annual, and the degree of refinement of the corresponding candidate power consumption data can be selected according to actual needs. The average daily power parameter time series within the specified time period can better understand the operating characteristics of the user's power equipment.

[0084] Specifically, an embodiment of the present invention first determines a matching user group based on the power consumption pattern category of the user to be detected; then, a specified time period is selected based on actual needs; then, the power consumption data of the user group is calculated according to the specified time period and different time types (such as weekdays, weekends or holidays) to obtain corresponding candidate power consumption data; finally, based on the time type to which the user to be detected currently belongs, the corresponding candidate power consumption data is selected as reference power consumption data. For example, if the user to be detected is on a weekday, the fourth power consumption statistical feature, the average daily power consumption time series and the average daily power parameter time series of the user group on the weekday are selected as reference power consumption data.

[0085] It is worth noting that after calculating the similarity values ​​of each dimension between the power consumption data of the user to be detected and the reference power consumption data, the following two methods can be used to determine whether the user to be detected is an abnormal power user:

[0086] (1) Calculate the overall similarity between the two by using weighted average algorithm and geometric mean algorithm, and then determine whether the overall similarity exceeds the preset range to determine whether the user to be detected is an abnormal electricity user;

[0087] (2) A corresponding preset range is set for each dimension. If the calculated similarity values ​​of all dimensions are within the corresponding preset range, the user to be detected is a normal electricity user; otherwise, it is an abnormal electricity user. The similarity measurement methods of each dimension include: Euclidean distance, cosine similarity, deviation rate calculated based on mean square error, and DTW value calculated based on dynamic time warping.

[0088] In an optional embodiment, the power parameter time series includes: a power time series, a current time series, a voltage time series and a power load time series.

[0089] It is worth noting that when electrical equipment is started and shut down, fluctuations in power, current and load are observed, and the fluctuations in these parameters can clearly reflect the operating period of the electrical equipment. For example, when the equipment is started, the starting current will increase instantaneously and then stabilize at the current level of normal operation. The power consumption and load of the equipment are different in different operating states (such as standby state, full load operation, etc.), and these differences can be used to determine the different operating states of the equipment from the time series. In addition, current fluctuations can also reflect the periodic working characteristics of the equipment, such as the working cycle of the compressor.

[0090] See also Figure 2 , is a schematic diagram of an embodiment of obtaining characteristic data of an electrical device provided by the present invention.

[0091] In an optional embodiment, the acquiring characteristic data of the electric device based on the power parameter time series of the abnormal electric user includes:

[0092] Extracting a current time subsequence corresponding to the operation period of the electrical equipment from the current time series;

[0093] Based on the current time subsequence, obtaining a current change value when the electrical equipment is started, and a current average value and a current standard deviation during steady-state operation;

[0094] Performing harmonic analysis on the current time subsequence to obtain multiple harmonic components and corresponding harmonic amplitudes;

[0095] Characteristic data of the electrical equipment are obtained based on the current change value, the current average value, the current standard deviation, the harmonic component and the harmonic amplitude.

[0096] It should be noted that the characteristic data of the electrical equipment is formed through the current change value of the electrical equipment at startup, the steady-state operating current characteristics (current average value and current standard deviation), and the harmonic components (covering multiple harmonic components and corresponding harmonic amplitudes). These characteristic data help to reveal the unique electrical behavior pattern of the electrical equipment, thereby realizing the identification of the electrical equipment, as follows:

[0097] (1) Current change value at startup: The current change value of different electrical equipment at startup is different. For example, motor-driven equipment (such as air conditioners and washing machines) usually has a significant current peak at startup, while other equipment (such as lamps and electronic equipment) has a relatively gentle current change at startup. Therefore, the current change value of electrical equipment at startup can be used as one of the key factors of characteristic data.

[0098] (2) Steady-state operating current characteristics: The average current and standard deviation during steady-state operation reflect the current consumption and fluctuation characteristics of the equipment during long-term operation. For example, refrigerators usually have periodic current fluctuations, which are caused by the intermittent operation of their compressors; while computers have small current fluctuations under normal working conditions, but the current will increase significantly under high load.

[0099] (3) Harmonic components: Harmonic analysis can reveal the high-frequency components and periodic changes in the operation of the equipment. This information is closely related to the internal structure and operation mechanism of the equipment. For example, there are often more harmonic components in the operation of the motor, but their frequency is relatively stable; and the distribution of harmonic components in electronic equipment also has a specific pattern due to the presence of switching power supplies inside. When performing harmonic analysis on the current time subseries, methods such as fast Fourier transform (FFT) can be used to decompose the signal to obtain multiple harmonic components and their corresponding harmonic amplitudes.

[0100] The above feature data is input into a pre-trained device recognition model. The device recognition model is a classification model designed based on a neural network, support vector machine or decision tree architecture, which is trained using a supervised learning method. The training data covers the feature data of various types of electrical equipment at work, which can be obtained through laboratory measurements and other methods. The device recognition model constructed in this way can effectively identify a variety of electrical equipment such as air conditioners, refrigerators, washing machines, etc. As the types of electrical equipment in the training data increase, more types of equipment can be identified.

[0101] See also Figure 3 , is a schematic diagram of an embodiment of obtaining operating data of an electrical equipment provided by the present invention.

[0102] In an optional embodiment, the acquiring the operation data of the electric equipment based on the power parameter time series of the abnormal electric user includes:

[0103] Extracting the current time subsequence, voltage time subsequence and power time subsequence corresponding to the electrical equipment in the operation period from the current time series, the voltage time series and the power time series respectively;

[0104] Calculating a phase difference between the current time subsequence and the voltage time subsequence;

[0105] Based on the power time subsequence and the phase difference, calculating the active power and reactive power of the electrical equipment under different operating states;

[0106] Based on the power load time series, obtaining the power load of the power-consuming equipment in different time periods;

[0107] Based on the phase difference, the active power, the reactive power and the power load, the operation data of the power-consuming device is obtained.

[0108] It should be noted that the operating data of the power equipment is formed by phase difference, active power, reactive power and power load; the reference operating data comes from the benchmark operating data of the corresponding equipment type obtained through laboratory measurements and other methods. The operating data of the current power equipment is compared with the reference operating data, that is, the deviation between the two can be evaluated through statistical analysis (such as mean and standard deviation analysis), cluster analysis or machine learning (such as anomaly detection algorithm). The following two methods can be used to determine whether abnormal power users have committed electricity theft:

[0109] (1) After obtaining the deviation of each dimension of the operating data, the overall deviation between the two is calculated by using the weighted average algorithm and the geometric mean algorithm. Subsequently, by judging whether the overall deviation exceeds the preset range, it is determined whether the abnormal electricity user has committed electricity theft;

[0110] (2) When the deviation exceeds the corresponding preset threshold in any dimension, it can be determined that the abnormal electricity user has committed electricity theft. For example, if the current operating data of the power-consuming equipment is compared with the reference operating data and the active power is significantly lower than the reference operating expectation, the reactive power or phase difference is abnormal, and the power load has unreasonable fluctuations compared with the standard equipment, it indicates that the abnormal electricity user has committed electricity theft.

[0111] It is worth noting that when the electrical equipment is operating normally, its reference operating data has a typical value range and change trend. In order to more clearly describe the technical solution provided by the embodiment of the present invention, some specific embodiments are provided below for reference:

[0112] (1) Abnormal phase difference: Power theft can cause abnormal phase difference in equipment, such as human interference that destroys the normal relationship between current and voltage.

[0113] (2) Power changes: Active power and reactive power can reflect the power consumption of equipment under different operating conditions. Usually, power theft will affect the actual power usage of equipment, causing its active power to be lower than it should be, or its reactive power to fluctuate significantly.

[0114] (3) Changes in power load: Power load indicates the power demand of equipment in different time periods. Under normal circumstances, people's power consumption behavior has certain regularities. For example, household power consumption will have a higher load in the morning, evening or on weekends. However, power theft will break these regularities, for example, causing unreasonable power load fluctuations in abnormal time periods or making the total power load significantly lower.

[0115] By comparing the operating data of abnormal electricity users with the reference data of normal equipment, potential abnormal deviations can be discovered, as described in Table 1.

[0116] Table 1. Typical judgments on electricity theft

[0117]

[0118] In an optional embodiment, the method further includes:

[0119] Based on the similarity, determining a probability of abnormal power consumption;

[0120] Obtaining payment information and abnormal power usage frequency of the user to be detected, and determining the violation risk level;

[0121] Obtaining the load fluctuation rate and power consumption fluctuation rate of the user to be detected within a preset time period, and determining the power consumption risk level;

[0122] Based on the abnormal power usage probability, the violation risk level and the power usage risk level, a first profile of the user to be detected is established.

[0123] It should be noted that the abnormal power consumption probability of the user to be detected can be calculated through the similarity between the power consumption data of the user to be detected and the reference power consumption data of the power consumption mode category to which it belongs. For example, the similarity value is first standardized to a range, such as 0 to 1, and then the abnormal power consumption probability is obtained by "1-standardized similarity value"; or, the abnormal power consumption probability is obtained by the Z-Score method, that is,

[0124] ;in, is the similarity value of the user to be detected, is the standard deviation of similarity values ​​of user groups, is the mean of the user groups; is the abnormal power consumption probability of the user to be detected.

[0125] The payment information of the user to be tested includes: payment timeliness rate, accumulated arrears, arrears amount, payment method (online or offline), and duration of arrears. The abnormal frequency of electricity consumption records the number of abnormal electricity consumption events that occurred in the user within a specific period of time in the past. By analyzing the payment information and the abnormal frequency of electricity consumption, it is possible to preliminarily determine whether there are long-term suspicious situations in the user's electricity consumption behavior. If the user frequently has abnormal electricity consumption events or has a record of arrears, the risk level of violation is relatively high. The multi-factor analysis method can be used to quantify and comprehensively calculate the payment information and abnormal frequency data, and finally give a specific risk level. This method is also applicable to the determination of electricity risk level.

[0126] The power load fluctuation rate reflects the change in the user's power demand within a preset time period, while the power consumption fluctuation rate refers to the change in the user's power consumption within a specific time period; the preset time period is at least one of the monthly, quarterly and annual periods. Through the analysis of these two, the stability of the user's power consumption behavior can be evaluated. For example, if the user's load and power consumption are highly volatile, it may mean that their power consumption behavior is unstable and there is a risk of abnormal power consumption.

[0127] Based on the above-mentioned abnormal power consumption probability, violation risk level and power consumption risk level, a first profile of the user to be detected is comprehensively established to provide a basis for the monitoring and management of the user's subsequent power consumption data. The power consumption behavior monitoring strategy can be determined based on the user's first profile, and differentiated real-time monitoring rules can be formulated to obtain the user's real-time power consumption data, determine whether it has power theft behavior, and further improve the accuracy of power theft identification. For example, when the power consumption risk level in the user's first profile is high risk, a real-time power consumption data monitoring rule corresponding to the user is formulated. If the user's daily power consumption exceeds 1000 degrees or is less than 100 degrees, an automatic warning will be issued, and the power theft detection method provided in the above embodiments will be immediately used for detection and identification. This helps to improve the accuracy of power theft detection and further reduce the missed alarm rate and false alarm rate.

[0128] In an optional embodiment, the method further includes:

[0129] Based on the payment information, obtaining corresponding payment characteristics;

[0130] Based on the abnormal power consumption probability, the payment characteristics and the power consumption fluctuation rate, a second portrait of the user to be detected is calculated.

[0131] It should be noted that the embodiment of the present invention combines the probability of abnormal electricity consumption, payment characteristics (such as the payment timeliness rate and the number of arrears) and the fluctuation rate of electricity consumption, and adopts the weighted average method to calculate the second portrait of the user to be detected; wherein, the larger the value corresponding to the second portrait, the lower the probability that the user to be detected has abnormal electricity consumption behavior and electricity theft behavior. The weights of various indicators are determined based on the statistical analysis of historical data. When the value corresponding to the second portrait of the user is lower than the preset threshold, the electricity consumption data of the user needs to be monitored for a long time. In addition, the historical change trend of the second portrait of the user can also be judged. If the change trend is not good, the electricity consumption data of the user is monitored and analyzed for a long time. Therefore, the embodiment of the present invention can use the second portrait of the user to initially anchor the situation of suspected abnormal electricity consumption, and then combine the electricity theft detection method provided in the above embodiments to detect and identify, which can effectively improve the accuracy of electricity theft detection and further reduce the false alarm rate and false alarm rate.

[0132] Exemplarily, the weighted average method is used to assign weights of 3, 3, 2, and 2 to the power consumption fluctuation rate, payment timeliness rate, number of arrears, and probability of abnormal power consumption, respectively, to calculate the second portrait of the user to be detected. For example, if the value corresponding to the second portrait of the user is 88, which is higher than the preset threshold of 80, there is no need to conduct long-term monitoring of the user's power consumption data, and only regular monitoring is required. Analyze the changes in the user's second portrait in the past six months. If it is found that the value corresponding to the second portrait has steadily increased from 85 to 88, it means that the user's power consumption behavior is becoming more regular and normal; if it is found that the value corresponding to the second portrait is gradually decreasing or fluctuating abnormally, it is necessary to conduct long-term monitoring of the user's power consumption data.

[0133] See also Figure 4 , is a structural diagram of an embodiment of the electricity theft detection system provided by the present invention.

[0134] A second aspect of the present invention provides an electricity theft detection system, which is used to implement the electricity theft detection method described in any one of the embodiments of the first aspect, and the system includes:

[0135] The power consumption data acquisition module 11 is used to acquire the power consumption data of the user to be detected on the day; wherein the power consumption data includes: power consumption time series, first power consumption statistical characteristics and power parameter time series;

[0136] A similarity calculation module 12 is used to calculate the similarity between the power consumption data of the user to be detected and the reference power consumption data of the power consumption mode category to which it belongs;

[0137] The abnormal power consumption determination module 13 is used to determine that the user to be detected is an abnormal power consumption user when the similarity exceeds a preset range;

[0138] The device data acquisition module 14 is used to acquire characteristic data and operation data of the power-consuming device based on the power parameter time series of the abnormal power user;

[0139] The electrical equipment identification module 15 is used to input the characteristic data into the equipment identification model to obtain the type of the electrical equipment;

[0140] The electricity stealing behavior judgment module 16 is used to judge whether the abnormal electricity user has committed electricity stealing behavior by comparing the operation data with reference operation data of the type of the electricity-consuming device.

[0141] It should be noted that the electricity theft detection system provided in the second embodiment of the present invention can implement all the processes of the electricity theft detection method described in the first embodiment above. The functions of each module in the system and the technical effects achieved are respectively the same as the functions and technical effects achieved by the electricity theft detection method described in the first embodiment above, and will not be repeated here.

[0142] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for detecting electricity theft, characterized in that: include: Obtaining the power consumption data of the user to be detected on the day; wherein the power consumption data includes: a power consumption time series, a first power consumption statistical feature and a power parameter time series; Calculating the similarity between the power consumption data of the user to be detected and the reference power consumption data of the power consumption mode category to which it belongs; When the similarity exceeds a preset range, determining that the user to be detected is an abnormal electricity user; Based on the power parameter time series of the abnormal power user, characteristic data and operation data of the power equipment are obtained; wherein the characteristic data include: harmonic components, harmonic amplitudes, current change values ​​when the power equipment is started, and current average values ​​and current standard deviations during steady-state operation; Inputting the characteristic data into a device identification model to obtain the type of electrical equipment; By comparing the operating data with reference operating data of the type of electrical equipment, determining whether the abnormal electricity user has committed electricity theft; The power parameter time series includes: power time series, current time series, voltage time series and power load time series; The acquiring characteristic data of the electric equipment based on the electric power parameter time series of the abnormal electric user comprises: Extracting a current time subsequence corresponding to the operation period of the electrical equipment from the current time series; Based on the current time subsequence, obtaining a current change value when the electrical equipment is started, and a current average value and a current standard deviation during steady-state operation; Performing harmonic analysis on the current time subsequence to obtain multiple harmonic components and corresponding harmonic amplitudes; Obtaining characteristic data of the electrical equipment based on the current change value, the current average value, the current standard deviation, the harmonic component and the harmonic amplitude; The acquiring the operation data of the electric equipment based on the time series of the electric power parameter of the abnormal electric user comprises: Extracting the current time subsequence, voltage time subsequence and power time subsequence corresponding to the electrical equipment in the operation period from the current time series, the voltage time series and the power time series respectively; Calculating a phase difference between the current time subsequence and the voltage time subsequence; Based on the power time subsequence and the phase difference, calculating the active power and reactive power of the electrical equipment under different operating states; Based on the power load time series, obtaining the power load of the power-consuming equipment in different time periods; Based on the phase difference, the active power, the reactive power and the power load, the operation data of the power-consuming device is obtained.

2. The electricity theft detection method according to claim 1, characterized in that: All the power consumption mode categories are obtained by the following steps: Acquire a first data set of a user group; wherein the first data set includes: geographic areas, electricity consumption types, second electricity consumption statistical characteristics in multiple time periods, average daily electricity consumption time series and average daily power parameter time series of a number of individual users; the user group includes the user to be detected; Based on a preset clustering algorithm and the first data set, all the individual users are clustered and grouped to obtain a plurality of power consumption pattern categories.

3. The electricity theft detection method according to claim 1, characterized in that: All the power consumption mode categories are obtained by the following steps: Acquire a second data set of a user group; wherein the second data set includes: geographic areas, electricity usage types, electricity voltage levels, third electricity usage statistical characteristics within multiple time periods, and electricity usage time period statistical distribution data of a number of individual users; the user group includes the user to be detected; Based on a preset clustering algorithm and the second data set, all the individual users are clustered and grouped to obtain a plurality of power consumption pattern categories.

4. The electricity theft detection method according to claim 1, characterized in that: The reference power consumption data is obtained by the following steps Based on the power usage mode category, determining a corresponding user group; According to the specified time period, calculate the candidate power consumption data of the user group under different time types; wherein the time type includes: working day, weekend or holiday; the candidate power consumption data includes: the fourth power consumption statistical feature, the average daily power consumption time series and the average daily power parameter time series; Based on the time type of the to-be-detected user on that day, corresponding candidate power usage data is selected as the reference power usage data.

5. The electricity theft detection method according to claim 1, characterized in that: The method further comprises: Based on the similarity, determining a probability of abnormal power consumption; Obtaining payment information and abnormal power usage frequency of the user to be detected, and determining the violation risk level; Obtaining the load fluctuation rate and power consumption fluctuation rate of the user to be detected within a preset time period, and determining the power consumption risk level; Based on the abnormal power usage probability, the violation risk level and the power usage risk level, a first profile of the user to be detected is established.

6. The electricity theft detection method according to claim 5, characterized in that: The method further comprises: Based on the payment information, obtaining corresponding payment characteristics; Based on the abnormal power consumption probability, the payment characteristics and the power consumption fluctuation rate, a second portrait of the user to be detected is calculated.

7. A power theft detection system, characterized in that: The method for detecting electricity theft according to any one of claims 1 to 6, wherein the system comprises: An electricity consumption data acquisition module is used to acquire the electricity consumption data of the user to be detected on the day; wherein the electricity consumption data includes: a time series of electricity consumption, a first electricity consumption statistical feature and a time series of electricity parameters; A similarity calculation module, used to calculate the similarity between the power consumption data of the user to be detected and the reference power consumption data of the power consumption mode category to which it belongs; An abnormal power consumption determination module, configured to determine that the user to be detected is an abnormal power consumption user when the similarity exceeds a preset range; An equipment data acquisition module, used to acquire characteristic data and operation data of electric equipment based on the time series of electric power parameters of the abnormal electric user; An electric device identification module, used to input the characteristic data into an equipment identification model to obtain the type of electric device; The electricity stealing behavior judgment module is used to judge whether the abnormal electricity user has committed electricity stealing behavior by comparing the operation data with reference operation data of the type of electrical equipment.

Images