Dynamic time-sharing user electricity stealing behavior detection method and system, terminal and medium

Abstract

The invention relates to the field of power monitoring, and particularly provides a dynamic time-sharing user electricity stealing behavior detection method and system, a terminal and a medium, and the method comprises the steps: obtaining and preprocessing measurement data, user file data and dynamic time-sharing electricity price data; calculating a real-time current transformation ratio based on the metering data, comparing the real-time current transformation ratio with a rated value, and judging a transformation ratio abnormal state; constructing a similar user group based on user archives, metering and electricity price data, detecting sudden change of average electricity price of users, calculating dynamic time warping deviation between a power consumption curve of a target user and a historical normal curve and a typical power consumption curve, and judging an abnormal state of the power consumption curve; calculating a power change rate and a standard deviation based on the power curve data, and judging a power form abnormal state; and finally, carrying out weighted scoring on the three types of anomalies, and judging an electricity stealing risk level according to a comprehensive score. According to the method, the detection rate, the accuracy rate and the inspection efficiency of electricity stealing behaviors of meter running electricity feeding price under the dynamic time-of-use electricity price are improved.

Description

Technical Field

[0001] This invention relates to the field of power monitoring, specifically to a method, system, terminal, and medium for detecting dynamic time-sharing user electricity theft. Background Technology

[0002] Some industrial and commercial users remove their electricity meters and install meters with altered current ratios. They then use electricity normally during peak hours to avoid metering, while generating small amounts of electricity during off-peak hours using specialized low-power devices. They then amplify this small amount by using the altered current ratio to record it, thus profiting from the electricity bill. Since this operation is usually completed within a day, it circumvents traditional detection methods that rely on macro-level indicators such as daily line loss rates. Current anti-theft technologies rely on macro-level indicators like daily line loss rates for transformer areas, which have a coarse time granularity and cannot effectively identify peak-to-off-peak electricity transfers within the same day, resulting in a low detection rate for "meter tampering and bill arbitrage." Furthermore, they focus on monitoring abnormal meter wiring or open meter records, lacking the ability to identify combinations of unauthorized meter replacement and software tampering of built-in current ratio parameters, leading to a high false positive rate. While some solutions use machine learning to analyze load curves, they typically do not closely integrate with the peak-to-off-peak characteristics of dynamic time-of-use pricing and lack targeted quantitative analysis of power curve shapes, resulting in poor accuracy in identifying specific electricity theft patterns. Summary of the Invention

[0003] To address the aforementioned issues, this invention provides a method, system, terminal, and medium for detecting electricity theft by users under dynamic time-of-use pricing, thereby improving the detection rate, accuracy, and investigation efficiency of electricity theft by "running the meter and consuming electricity" under dynamic time-of-use pricing.

[0004] In a first aspect, the technical solution of the present invention provides a method for detecting dynamic time-sharing user electricity theft, comprising the following steps: Acquire metering data, user profile data, and dynamic time-of-use electricity price data, and preprocess the acquired data; Based on the preprocessed metering data, the real-time current ratio is calculated and compared with the rated value. The comparison result is used to determine whether a ratio abnormality is triggered. Based on preprocessed user profile data, metering data, and dynamic time-of-use electricity price data, a similar user group is constructed, information on sudden changes in the average electricity price of users is detected, and when the average electricity price changes, the deviation between the target user's electricity consumption curve and the benchmark curve is calculated. Based on the curve deviation, it is determined whether an abnormal electricity consumption curve is triggered. Based on the power curve data in the preprocessed metering data, calculate the power change rate and power standard deviation, and determine whether power pattern abnormality is triggered based on the calculation results. Abnormal transformer ratio, abnormal electricity consumption curve, and abnormal power pattern are scored separately, and a weighted calculation model is used to obtain a comprehensive score. The user's electricity theft risk level is determined based on the comprehensive score.

[0005] Secondly, the technical solution of the present invention provides a dynamic time-sharing user electricity theft detection system, comprising: The data acquisition and preprocessing module is used to acquire metering data, user profile data, and dynamic time-of-use electricity price data, and to preprocess the acquired data. The transformer ratio anomaly detection module is used to calculate the real-time current ratio based on the pre-processed metering data, compare the real-time current ratio with the rated value, and determine whether a transformer ratio anomaly is triggered based on the comparison result. The electricity consumption curve anomaly detection module is used to construct similar user groups based on preprocessed user profile data, metering data and dynamic time-of-use electricity price data, detect sudden changes in the average electricity price of users, calculate the deviation between the target user's electricity consumption curve and the benchmark curve when the average electricity price changes, and determine whether to trigger an electricity consumption curve anomaly based on the curve deviation. The power shape anomaly detection module is used to calculate the power change rate and power standard deviation based on the power curve data in the preprocessed metering data, and to determine whether a power shape anomaly is triggered based on the calculation results. The electricity theft evaluation module is used to score abnormal transformer ratios, abnormal electricity consumption curves, and abnormal power patterns, and uses a weighted calculation model to obtain a comprehensive score. The comprehensive score is used to determine the user's electricity theft risk level.

[0006] Thirdly, the technical solution of the present invention provides a terminal, comprising: The memory is used to store the dynamic time-sharing user electricity theft detection program; A processor is configured to implement the steps of the dynamic time-sharing user electricity theft detection method as described above when executing the dynamic time-sharing user electricity theft detection program.

[0007] Fourthly, the present invention provides a computer-readable storage medium storing a dynamic time-sharing user electricity theft detection program, wherein the dynamic time-sharing user electricity theft detection program, when executed by a processor, implements the steps of the dynamic time-sharing user electricity theft detection method as described in any of the above claims.

[0008] As can be seen from the above technical solutions, this application has the following advantages: By integrating real-time transformer ratio monitoring, user profiling and curve analysis, and power pattern recognition, a detection system covering the entire chain of electricity theft behavior has been constructed. By fusing and cross-validating data such as line loss, metering, and curves, complete detection data of transformer ratio deviation, curve deviation, and power pattern anomaly can be formed, thereby improving the accuracy and reliability of electricity theft detection results. By using real-time ratio calculation and power curve shape analysis, it is possible to accurately capture sudden changes in power consumption and fixed power meter running characteristics that occur within the same day, especially at the peak-valley switching point. This breaks through the limitations of traditional methods in terms of time granularity, enabling proactive detection of the "same-day operation, same-day cover-up" electricity theft mode and improving the electricity theft detection rate. By constructing user profiles and using clustering and cosine similarity to filter similar user groups, the benchmark for comparing electricity consumption curves becomes more reasonable, reducing misjudgments caused by differences in production patterns across different industries and improving the accuracy of curve anomaly identification. By establishing a weighted scoring model, various types of abnormal evidence are quantitatively integrated to output an intuitive comprehensive risk level, thereby improving the detection efficiency of electricity theft. Attached Figure Description

[0009] To more clearly illustrate the technical solution of this application, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of a dynamic time-sharing user electricity theft detection method provided in an embodiment of the present invention.

[0011] Figure 2 This is a schematic diagram of typical electricity consumption curves for similar users.

[0012] Figure 3 This is a schematic diagram of the normalized results of the electricity consumption curve deviation.

[0013] Figure 4 This is a schematic block diagram of a dynamic time-sharing user electricity theft detection system provided in an embodiment of the present invention.

[0014] Figure 5 This is a schematic diagram of the structure of a terminal provided in an embodiment of the present invention. Detailed Implementation

[0015] To make the purpose, features, and advantages of this application more apparent and understandable, specific embodiments and accompanying drawings will be used to clearly and completely describe the technical solution protected by this application. Obviously, the embodiments described below are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0016] Unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this application and in the specification of this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0017] Figure 1 This is a schematic flowchart illustrating a dynamic time-sharing user electricity theft detection method provided in an embodiment of the present invention. Figure 1 The executing entity can be a dynamic time-sharing user electricity theft detection system. The dynamic time-sharing user electricity theft detection method provided in this embodiment of the invention is executed by a computer device; correspondingly, the dynamic time-sharing user electricity theft detection system runs on the computer device. Depending on different needs, the order of the steps in this flowchart can be changed, and some steps can be omitted.

[0018] like Figure 1 As shown, the method includes the following steps.

[0019] S1 acquires metering data, user profile data, and dynamic time-of-use electricity price data, and preprocesses the acquired data.

[0020] Metering data: Real-time current (primary / secondary value), voltage, power (15 minutes / point, generating a 96-point power curve), and time-of-use electricity consumption (peak / valley / flat period) of the smart meter.

[0021] User profile data: User's region code, industry code (e.g., C31-Agricultural and sideline food processing), average monthly electricity consumption (tiered: 10,000-100,000 kWh / 100,000-500,000 kWh), and electricity meter rated parameters (current ratio K, rated voltage level).

[0022] Electricity price data: Dynamic time-of-use electricity pricing time periods (e.g., peak period 8:00-22:00, off-peak period 22:00-8:00 the next day), and electricity price standards for each time period.

[0023] The data collection frequency is set to current / voltage data every 1 minute, power / energy data every 15 minutes, and user profile data is synchronized once a day.

[0024] Data preprocessing includes at least data cleaning and data standardization. Data cleaning uses the 3σ principle to remove abnormal fluctuations in current / voltage (such as instantaneous spikes), fills in missing power data through linear interpolation, and performs consistency checks on user profile data. Data standardization includes converting time-of-use electricity and power data to "kWh" and "kW" units, and unifying the timestamps to UTC+8 time zone to ensure time alignment across data sources.

[0025] S2, based on the preprocessed metering data, calculate the real-time current ratio and compare it with the rated value. Determine whether a ratio anomaly has been triggered based on the comparison result, including the following steps: S201, Extract the primary side current value and the secondary side current value from the preprocessed metering data; S202, calculate the ratio of the primary current value to the secondary current value, which is the real-time transformation ratio; S203, calculate the relative deviation between the real-time transformation ratio and the rated transformation ratio; S204 If the relative deviation continuously exceeds the preset threshold and reaches a predetermined number of times, a ratio anomaly is triggered.

[0026] Specifically, 1-10 kV users with high-voltage supply and low-voltage metering need to be equipped with current transformers, whose rated transformation ratio K (e.g., 100 / 5=20) is a fixed value. Electricity thieves tamper with the meter's built-in transformation ratio parameters (e.g., changing it to 200 / 5=40), causing small currents during off-peak hours to be amplified and recorded. This embodiment uses a high-precision current sensor built into the meter to simultaneously collect the actual primary current (Iprimary value) and the secondary current measured by the meter (Isecondary value), calculate the real-time transformation ratio K, and compare it with the rated K to identify transformation ratio tampering. A Hall effect current sensor with an error of <0.2% is used, supporting primary current acquisition of 10-500A.

[0027] Real-time ratio calculation is performed every 15 minutes, using the following formula: Ratio If it is satisfied 3 times in a row (Transformer ratio deviation threshold) triggers a transformer ratio anomaly warning.

[0028] S3, based on preprocessed user profile data, metering data and dynamic time-of-use electricity price data, constructs similar user groups, detects sudden changes in user average electricity price, and calculates the deviation between the target user's electricity consumption curve and the baseline curve when the average electricity price changes. It then determines whether to trigger an abnormal electricity consumption curve based on the curve deviation.

[0029] The first step is to construct a similar user group, which uses a two-step approach of clustering and similarity matching. Specifically, this includes the following steps: S301, extract the region code, industry code, and monthly average electricity consumption tier from the preprocessed user profile data as user feature vectors; S302, based on user feature vectors, uses a clustering algorithm to divide users into multiple homogeneous groups; S303, within the homogeneous group to which the target user belongs, calculate the cosine similarity between the daily time-of-use electricity consumption curves of the target user and other users in the group, and select a predetermined number of users based on the cosine similarity. These users constitute a similar user group, and the average time-of-use electricity consumption curve of these users is defined as the typical electricity consumption curve of the similar user group.

[0030] Specifically, user characteristics are input, including region code, industry code, and monthly average electricity consumption tiers. K-means clustering is used to further subdivide the number of clusters by industry, such as agricultural and sideline food processing k=5 and machinery manufacturing k=6, to divide users into several homogeneous groups.

[0031] The formula for calculating the 24-hour time-of-use similarity of battery consumption between the target user and candidate users within the same group is as follows:

[0032] In the formula, For the target user's electricity consumption in the i-th hour, For the electricity consumption of candidate users in the i-th hour, the top 10 similar users are selected, and their average time-of-use curve is the "typical electricity consumption curve (T)". Figure 2 This is a schematic diagram of typical electricity consumption curves for similar users.

[0033] The second step is to detect sudden changes in the average electricity price for users. This involves the following steps: S304, calculate the average electricity price to the target user within a billing cycle based on the preprocessed dynamic time-of-use electricity price data and time-of-use electricity consumption data; S305 If the average electricity price delivered to households in the current period decreases by more than the first preset threshold compared to the average electricity price in its historical periods, and there is no record of electricity price policy changes in the user's file, then the period is marked as an abnormal period.

[0034] Specifically, the average electricity price is calculated by taking into account the average monthly electricity price paid to each user. The formula is as follows:

[0035] If the average price in a certain month decreases by more than 30% compared to the average of the previous three months, and there are no records of "work orders for changing categories" (such as changing to preferential electricity prices for agricultural production or school use), that month is marked as a "month suspected of electricity theft".

[0036] The third step involves calculating the deviation between the target user's electricity consumption curve and the baseline curve when the average electricity price suddenly changes. Based on the curve deviation, it is determined whether an electricity consumption curve anomaly is triggered, i.e., the electricity consumption curve deviation is calculated. This specifically includes the following steps: S306, define the average electricity consumption curve of a predetermined number of historical normal cycles before the abnormal cycle as the first reference curve; S307, and defines the typical electricity consumption curve of similar user groups as the second reference curve; S308, calculate the dynamic time normalization distance between the target user's electricity consumption curve and the first reference curve and the second reference curve during the abnormal period, and normalize the dynamic time normalization distance to the preset range to obtain the normalized deviation value. S309, if any normalized deviation value exceeds the second preset threshold, the power consumption curve is determined to be abnormal.

[0037] Specifically, the comparison curve is defined, including: Historical Normal Curve (H): 24-hour time-of-use average electricity consumption over the three months preceding the suspected month. ; Typical curve (T): 24-hour time-of-use average battery consumption for similar user groups. ; Target curve (X): 24-hour time-of-use electricity consumption for the suspected month. .

[0038] Dynamic Time Warping (DTW) deviation calculation: This addresses the issue of time-period offset in time-sharing curves (such as slight fluctuations in user production time). The formula is as follows:

[0039] Where Y is H or T. For dynamic programming of path weights; The DTW value is mapped to the [0,1] interval using min-max normalization, as shown in the formula:

[0040] If any normalization deviation is greater than 0.6 (threshold), an abnormal curve warning is triggered. Figure 3 This is a schematic diagram of the normalized results of the electricity consumption curve deviation.

[0041] S4. Based on the power curve data in the preprocessed metering data, calculate the power change rate and power standard deviation, and determine whether power anomalies are triggered based on the calculation results. This includes the following steps: S401 calculates the rate of power change between adjacent time points based on power curve data; S402, the proportion of time points in which the statistical power change rate exceeds the high fluctuation threshold to the total number of time points; S403, calculate the standard deviation of the power curve data, and calculate the ratio of the standard deviation to the power mean; S404, if the ratio is lower than the low frequency threshold and the ratio is lower than the low fluctuation threshold, then the trigger power pattern is determined to be abnormal.

[0042] Specifically, in the "meter-skimming" electricity theft, the special equipment used during off-peak hours (such as a fixed power resistor box) will cause the 96-point power curve to show the characteristics of "straight up and down with minimal fluctuations", which is significantly different from the dynamic power curve of normal production (such as the fluctuations caused by machine tool start-up and shutdown). This can be identified through quantitative morphological indicators.

[0043] The rate of change of power is obtained by calculating the percentage of the power difference between adjacent 15-minute intervals, using the following formula:

[0044] statistics" The percentage of times ">50%": Normal users (electricity used for production) usually >20%, while electricity theft users <5%.

[0045] The power standard deviation is used to measure the degree of power fluctuation at 96 points. The formula is:

[0046] in The average power value of 96 points indicates electricity theft by users. (Normal users typically account for >30%).

[0047] If both indicators meet the characteristics of electricity theft, an abnormal power pattern warning is triggered, and a power curve pattern analysis report containing the change rate distribution and standard deviation calculation results is generated.

[0048] S5 scores abnormal transformer ratios, abnormal electricity consumption curves, and abnormal power patterns separately, and uses a weighted calculation model to obtain a comprehensive score. Based on the comprehensive score, the user's electricity theft risk level is determined, specifically including the following steps: S501, according to the suspected scoring rules, assigns each of the triggered abnormal turns ratio, abnormal power consumption curve or abnormal power mode to its respective abnormal score. S502, multiply each abnormal score by its corresponding preset weight and sum them to obtain the comprehensive score; S503 compares the overall score with at least two preset risk thresholds and determines the user's electricity theft risk level based on the comparison results.

[0049] Specifically, this embodiment employs a configurable suspected anomaly scoring rule, dynamically assigning a score to each anomaly within its 0-10 range. This rule, based on calculated quantitative indicators, maps these indicators to corresponding anomaly scores using a linear or piecewise linear function. The parameters of the mapping function can be configured and adjusted based on historical data.

[0050] a) Rules for calculating abnormal ratio scores.

[0051] Variable ratio anomaly score The maximum sustained ratio deviation rate detected Decision. Deviation rate. The calculation formula is .

[0052] Set deviation rate threshold and The score is calculated as follows: like ,but ; like ,but ; like ,but .

[0053] b) Rules for calculating abnormal scores on electricity consumption curves.

[0054] Abnormal score of electricity consumption curve The maximum normalized deviation value obtained from the calculation The value is determined by the maximum normalized DTW deviation between the target curve and the two baseline curves (historical curve H and typical curve T).

[0055] Set deviation threshold and The score is calculated as follows: like ,but ; like ,but ; like ,but .

[0056] c) Rules for calculating power form anomaly scores.

[0057] Power morphology anomaly score It is determined by the scores of the two sub-indicators.

[0058] Rate of change score ( ): Percentage of points with a power change rate > 50% .like ,but ;like ,but If it falls between the two, then .

[0059] Volatility score ( ): Calculate volatility .like ,but ;like ,but If it falls between the two, then .

[0060] Finally, the power shape anomaly score .

[0061] in, This is the lower limit of the normal threshold. This is the threshold for identifying electricity theft characteristics. This is the normal fluctuation threshold. This is the threshold for fluctuations in electricity theft.

[0062] Weights are assigned based on the strength of evidence for each anomaly, using the following formula:

[0063] Users with a comprehensive score of ≥7 are classified as "highly suspected users," triggering an investigation work order; those with a score of 5-7 are classified as "users under observation," requiring continuous monitoring for one month; and those with a score <5 are classified as "normal users." The weights for these scores are 40%, 35%, and 25%, respectively.

[0064] The following specific embodiment further illustrates this solution in detail.

[0065] This specific embodiment uses a bearing, gear, and transmission component manufacturing plant (user profile information: region code 37415, industry code 3550 (national industry classification: C3550-bearing, gear, and transmission component manufacturing), with an average monthly electricity consumption of 35,000 kWh and a rated transformation ratio of the electricity meter. (Taking a current transformer specification of 300 / 5, i.e., a primary current of 300A corresponds to a secondary current of 5A) as an example.

[0066] Step 1: System deployment and data source configuration.

[0067] 1.1 Data source acquisition.

[0068] Metering data (obtained from the "New Generation Electricity Information Collection System"): Real-time current of smart meters (primary current) Secondary current Data collection frequency: 1 minute / time), voltage (line voltage 380V, 1 minute / time), 96-point power curve (15 minutes / point, 96 power data points generated daily), and time-of-use electricity (electricity consumption divided into peak / peak / flat / valley / deep valley periods according to dynamic time-of-use policy, statistical granularity 1 hour / time).

[0069] User profile data (obtained from the marketing system): This user's basic information includes area code 37415, industry code 3550, average electricity consumption over the past 12 months of 35,000 kWh (belonging to the "30,000-50,000 kWh / month" category), electricity meter number, and rated transformer ratio. .

[0070] Dynamic Time-of-Use Policy Data: Based on the 2025 dynamic time-of-use electricity pricing policy for industrial and commercial users, the time period division and electricity price standards for 1-10 kV industrial and commercial dynamic time-of-use users are as follows:

[0071] 1.2 Data Preprocessing.

[0072] Outlier Removal: The 3σ principle is used to remove instantaneous spikes in current and voltage. For example, a current spike of 450A that occurs at 13:30 on April 12, 2025, exceeds the 3σ range (average 300A + 3 × 20A = 360A), and the system automatically marks it as an anomaly and removes it.

[0073] Missing value imputation: For occasional missing points in the 96-point power curve, the system uses linear interpolation to imput them. For example, if the power data at 10:45 on May 8, 2025 is missing, the imputed value is the average of the power values ​​at 10:30 (8kW) and 11:00 (8.2kW), which is 8.1kW.

[0074] Step 2: Implementation process for each anomaly detection.

[0075] 2.1 Monitoring of electricity meter transformation ratio.

[0076] Call the system output every 15 minutes and According to the formula Calculate the real-time ratio. Using data from April 15, 2025 as an example:

[0077] The above three real-time ratio deviation rates all far exceeded the 5% threshold, triggering a ratio anomaly warning and generating a "Ratio Deviation Report" (which recorded 128 instances of three consecutive deviations exceeding the standard during April-May of 2025, concentrated in the trough period of 11:00-14:00). According to the suspected scoring rules, it received 8 points.

[0078] 2.2 User profiling and curve analysis.

[0079] 1) Building a similar user group.

[0080] Based on the characteristics of "region code 37415 + industry code 3550 + average monthly electricity consumption of 30,000-50,000 kWh", 20 homogeneous users were selected from the system's user database.

[0081] Based on industry segmentation k-value (k=4 for 3550 industries), the 20 users were grouped into 4 groups, and the target users were classified into the "monthly average of 32,000-38,000 kWh" group (5 users in total).

[0082] Calculate the cosine similarity of the 24-hour time-of-use electricity consumption between the target user and 5 users in the same group, and select the TOP10 similar users (actually 5 users in the same group + 5 users in the neighboring group with similarity > 0.85). Their 24-hour time-of-use average electricity consumption is the typical electricity consumption curve (T), which is characterized by: electricity consumption of 1.2-1.5 kWh / hour during peak hours (17-20) and electricity consumption of 0.3-0.5 kWh / hour during valley hours (11-14) (which conforms to the pattern of "low load debugging during the day and high load production in the evening" in the bearing manufacturing industry).

[0083] 2) Detection of sudden changes in average electricity price for users Average electricity price calculation: according to the formula

[0084] Calculate the monthly average price: Average price in April 2024: 0.77 yuan / kWh; Average price in May 2024: 0.81 yuan / kWh; Average price in April 2025: 0.42 yuan / kWh; Average price in May 2025: 0.43 yuan / kWh.

[0085] The average price in April 2025 decreased by 45.5% compared to the same period in 2024, and the decrease in May 2025 reached 46.9%, both exceeding the 30% threshold. Moreover, the system had no record of "change of work order" for this user (such as changing to agricultural electricity price or residential electricity price). Therefore, April and May 2025 were marked as "suspected electricity theft months".

[0086] 3) Calculation of deviation in electricity consumption curve Historical normal curve (H): 24-hour time-of-use average electricity consumption in April and May of 2024, such as 0.4 kWh / hour during the valley period (11-14) and 1.35 kWh / hour during the peak period (17-20).

[0087] Typical electricity consumption curve (T): The 24-hour time-of-use average electricity consumption of similar user groups, such as 0.3-0.5 kWh / hour during off-peak hours (11-14) and 1.2-1.5 kWh / hour during peak hours (17-20).

[0088] Target curve (X): 24-hour time-of-use electricity consumption in April 2025, with an average of 1.8 kWh / hour during off-peak hours (11-14) and an average of 0.3 kWh / hour during peak hours (17-20).

[0089] Using the system's built-in DTW algorithm tool, the DTW value of X and H is calculated to be 18.2, and the DTW value of X and T is calculated to be 17.8.

[0090] The system statistics show that the DTW value range for users in this industry is 2.3-22.5, according to the formula...

[0091] The calculated normalized deviations of X and H are 0.78 and 0.76, respectively. Both exceed the 0.6 threshold, triggering an abnormal curve warning. Based on the suspected error scoring rule, a score of 7 is awarded.

[0092] 2.3 Point power shape recognition.

[0093] Power data extraction: Retrieve the power curve of 96 points from April 15, 2025 (15 minutes / point), focusing on the data during the trough period (11:00-14:00):

[0094] Calculate characteristic indicators, including power change rate and power standard deviation.

[0095] Power change rate: Among the 96 power points, the number of times the change rate was >50% was 0, accounting for 0% < 5% (characteristic of electricity theft).

[0096] Power standard deviation: =7.8kW, =0.32kW, =4.1 < 10% (Characteristics of electricity theft).

[0097] Both indicators meet the characteristics of "fixed power meter running", triggering an abnormal power pattern warning, and are awarded 5 points according to the suspected scoring rules.

[0098] Step 3: Multi-dimensional comprehensive evaluation.

[0099] The calculation is obtained by weighting using the following formula:

[0100] The calculation yielded: Weighted sum = 8 * 40% + 7 * 35% + 5 * 25% = 3.2 + 2.45 + 1.25 = 6.9 ≈ 7 If the overall score is ≥7 points, the user is determined to be a "highly suspected user of meter tampering and overcharging for electricity". The system will automatically generate an inspection work order and push it to the relevant inspection department.

[0101] Based on the work orders pushed by the system, the inspectors conducted an on-site inspection of the factory, and the verification results are as follows: 1. Evidence of tampering with metering devices: It was discovered that users removed smart meters and installed uncertified modified meters; or that official smart meters were modified by using jammers to alter the current transformers and amplify the current ratio.

[0102] 2. Verification of electricity usage: During peak hours, production electricity is not connected to the official metering circuit, but is directly diverted from before the meter to circumvent metering; during off-peak hours, only the resistance box is running, and "amplified electricity consumption" is recorded by privately installed meters. 4. Closed-loop evidence chain: The transformer ratio deviation data, curve anomaly reports, and power pattern characteristics monitored by the system, together with the tampered electricity meter and shunt circuit found on site, form a complete evidence chain, confirming that the user has engaged in electricity theft by "running the meter and eating electricity price".

[0103] The foregoing has described in detail an embodiment of a dynamic time-sharing user electricity theft detection method. Based on the dynamic time-sharing user electricity theft detection method described in the above embodiment, this invention also provides a dynamic time-sharing user electricity theft detection system corresponding to the method.

[0104] Figure 4 This is a schematic block diagram of a dynamic time-sharing user electricity theft detection system provided in an embodiment of the present invention. In this embodiment, the dynamic time-sharing user electricity theft detection system 400 can be divided into multiple functional modules according to its functions, such as... Figure 4 As shown. The module referred to in this invention is a series of computer program segments that can be executed by at least one processor and perform a fixed function, and which are stored in memory.

[0105] The data acquisition and preprocessing module 410 is used to acquire metering data, user profile data and dynamic time-of-use electricity price data, and to preprocess the acquired data.

[0106] The transformer ratio anomaly detection module 420 is used to calculate the real-time current ratio based on the pre-processed metering data, compare the real-time current ratio with the rated value, and determine whether a transformer ratio anomaly is triggered based on the comparison result.

[0107] The electricity consumption curve anomaly detection module 430 is used to construct similar user groups based on preprocessed user profile data, metering data and dynamic time-of-use electricity price data, detect sudden changes in the average electricity price of users, calculate the deviation between the target user's electricity consumption curve and the benchmark curve when the average electricity price changes, and determine whether to trigger an electricity consumption curve anomaly based on the curve deviation.

[0108] The power shape anomaly detection module 440 is used to calculate the power change rate and power standard deviation based on the power curve data in the preprocessed metering data, and to determine whether a power shape anomaly is triggered based on the calculation results.

[0109] The electricity theft evaluation module 450 is used to score abnormal transformer ratio, abnormal electricity consumption curve and abnormal power pattern, and to obtain a comprehensive score using a weighted calculation model. The comprehensive score is used to determine the user's electricity theft risk level.

[0110] The dynamic time-sharing user electricity theft detection system of this embodiment is used to implement the aforementioned dynamic time-sharing user electricity theft detection method. Therefore, the specific implementation of this system can be found in the embodiment section of the dynamic time-sharing user electricity theft detection method above. Thus, the specific implementation can be referred to the description of the corresponding embodiments, and will not be elaborated here.

[0111] Furthermore, since the dynamic time-sharing user electricity theft detection system in this embodiment is used to implement the aforementioned dynamic time-sharing user electricity theft detection method, its function corresponds to the function of the above method, and will not be repeated here.

[0112] Figure 5 This is a schematic diagram of a terminal 500 provided in an embodiment of the present invention, including: a processor 510, a memory 520, and a communication unit 530. The processor 510 is used to implement the flow steps of the above-described embodiment of the dynamic time-sharing user electricity theft detection method when implementing the dynamic time-sharing user electricity theft detection program stored in the memory 520.

[0113] This invention also provides a computer storage medium, which may be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc. The computer storage medium stores a dynamic time-sharing user electricity theft detection program. When the dynamic time-sharing user electricity theft detection program is executed by a processor, it implements the process steps of the above-described dynamic time-sharing user electricity theft detection method embodiment.

[0114] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting dynamic time-sharing user electricity theft, characterized in that, Includes the following steps: Acquire metering data, user profile data, and dynamic time-of-use electricity price data, and preprocess the acquired data; Based on the preprocessed metering data, the real-time current ratio is calculated and compared with the rated value. The comparison result is used to determine whether a ratio abnormality is triggered. Based on preprocessed user profile data, metering data, and dynamic time-of-use electricity price data, a similar user group is constructed, information on sudden changes in the average electricity price of users is detected, and when the average electricity price changes, the deviation between the target user's electricity consumption curve and the benchmark curve is calculated. Based on the curve deviation, it is determined whether an abnormal electricity consumption curve is triggered. Based on the power curve data in the preprocessed metering data, calculate the power change rate and power standard deviation, and determine whether power pattern abnormality is triggered based on the calculation results. Abnormal transformer ratio, abnormal electricity consumption curve, and abnormal power pattern are scored separately, and a weighted calculation model is used to obtain a comprehensive score. The user's electricity theft risk level is determined based on the comprehensive score.

2. The method for detecting dynamic time-sharing user electricity theft according to claim 1, characterized in that, Based on the preprocessed metering data, the real-time current ratio is calculated and compared with the rated value. The comparison result is used to determine whether a current ratio anomaly has been triggered. Specifically, this includes: Extract the primary and secondary current values ​​from the preprocessed metering data; Calculate the ratio of the primary current value to the secondary current value; this ratio is the real-time transformation ratio. Calculate the relative deviation between the real-time transformation ratio and the rated transformation ratio; If the relative deviation exceeds the preset threshold continuously and reaches a predetermined number of times, a ratio anomaly will be triggered.

3. The method for detecting dynamic time-sharing user electricity theft according to claim 1, characterized in that, Building similar user groups specifically includes: Extract the region code, industry code, and monthly average electricity consumption tier from the preprocessed user profile data as user feature vectors; Based on user feature vectors, a clustering algorithm is used to divide users into multiple homogeneous groups; Within the homogeneous group to which the target user belongs, the cosine similarity between the daily time-of-use electricity consumption curves of the target user and other users in the group is calculated. Based on the cosine similarity, a predetermined number of users are selected. These users constitute a similar user group, and the average time-of-use electricity consumption curve of these users is defined as the typical electricity consumption curve of the similar user group.

4. The method for detecting dynamic time-sharing user electricity theft according to claim 3, characterized in that, Detecting sudden changes in average electricity prices for users, specifically including: Based on the preprocessed dynamic time-of-use electricity price data and time-of-use electricity consumption data, calculate the average electricity price delivered to the target user within a billing cycle; If the average electricity price delivered to households in the current period decreases by more than a first preset threshold compared to the average electricity price in its historical periods, and there is no record of electricity price policy changes in the user's file, then the period is marked as an abnormal period.

5. The method for detecting dynamic time-sharing user electricity theft according to claim 4, characterized in that, When the average electricity price suddenly changes, the deviation between the target user's electricity consumption curve and the baseline curve is calculated. Based on the curve deviation, it is determined whether an electricity consumption curve anomaly is triggered. Specifically, this includes: The average electricity consumption curve of a predetermined number of historical normal cycles prior to the abnormal cycle is defined as the first reference curve. The typical electricity consumption curve of similar user groups is defined as the second reference curve; Calculate the dynamic time warping distance between the target user's electricity consumption curve during the abnormal period and the first and second reference curves, and normalize the dynamic time warping distance to a preset range to obtain the normalized deviation value. If any normalized deviation value exceeds the second preset threshold, the power consumption curve is determined to be abnormal.

6. The method for detecting dynamic time-sharing user electricity theft according to claim 1, characterized in that, Based on the power curve data in the preprocessed metering data, the power change rate and power standard deviation are calculated. The calculation results are used to determine whether power pattern anomalies have been triggered. Specifically, this includes: Based on power curve data, calculate the rate of power change between adjacent time points; The proportion of time points in the total number of time points where the rate of change of statistical power exceeds the high fluctuation threshold; Calculate the standard deviation of the power curve data, and calculate the ratio of the standard deviation to the power mean; If the ratio is lower than the low-frequency threshold and the ratio is lower than the low-fluctuation threshold, then the trigger power pattern is determined to be abnormal.

7. The method for detecting dynamic time-sharing user electricity theft according to claim 1, characterized in that, Abnormal transformer ratio, abnormal electricity consumption curve, and abnormal power pattern are scored separately, and a weighted calculation model is used to obtain a comprehensive score. The user's electricity theft risk level is determined based on the comprehensive score, specifically including: According to the suspected scoring rules, each of the triggered abnormal transformer ratio, abnormal power consumption curve, or abnormal power mode is assigned an abnormal score. The scores of each abnormality are multiplied by their corresponding preset weights and summed to obtain the comprehensive score. The overall score is compared with at least two preset risk thresholds, and the user's electricity theft risk level is determined based on the comparison results.

8. A dynamic time-sharing user electricity theft detection system, characterized in that, include: The data acquisition and preprocessing module is used to acquire metering data, user profile data, and dynamic time-of-use electricity price data, and to preprocess the acquired data. The transformer ratio anomaly detection module is used to calculate the real-time current ratio based on the pre-processed metering data, compare the real-time current ratio with the rated value, and determine whether a transformer ratio anomaly is triggered based on the comparison result. The electricity consumption curve anomaly detection module is used to construct similar user groups based on preprocessed user profile data, metering data and dynamic time-of-use electricity price data, detect sudden changes in the average electricity price of users, calculate the deviation between the target user's electricity consumption curve and the benchmark curve when the average electricity price changes, and determine whether to trigger an electricity consumption curve anomaly based on the curve deviation. The power shape anomaly detection module is used to calculate the power change rate and power standard deviation based on the power curve data in the preprocessed metering data, and to determine whether a power shape anomaly is triggered based on the calculation results. The electricity theft evaluation module is used to score abnormal transformer ratios, abnormal electricity consumption curves, and abnormal power patterns, and uses a weighted calculation model to obtain a comprehensive score. The comprehensive score is used to determine the user's electricity theft risk level.

9. A terminal, characterized in that, include: The memory is used to store the dynamic time-sharing user electricity theft detection program; A processor is configured to implement the steps of the dynamic time-sharing user electricity theft detection method as described in any one of claims 1 to 7 when executing the dynamic time-sharing user electricity theft detection program.

10. A computer-readable storage medium, characterized in that, The readable storage medium stores a dynamic time-sharing user electricity theft detection program, which, when executed by a processor, implements the steps of the dynamic time-sharing user electricity theft detection method as described in any one of claims 1 to 7.

Images