An abnormal power consumption identification method and device of an electric energy metering box

Abstract

The application provides an abnormal power consumption identification method and device of an electric energy metering box, and belongs to the technical field of power systems, which comprises the following steps: acquiring real-time monitoring time series data of total incoming line current of an electric energy metering box and each outgoing line sub-meter current; calculating a loop current missing rate based on the ratio of each outgoing line sub-meter current to the total incoming line current; calculating a missing fluctuation dispersion based on the fluctuation amplitude of the loop current missing rate; calculating a missing fluctuation correlation degree among multiple loops based on the correlation of the fluctuation of the loop current missing rate, and taking the missing fluctuation correlation degree among the first n loops as a relevant loop; screening out a high-risk loop, and performing risk accumulation tracking to calculate a risk accumulation index; and determining that the loop is an abnormal power consumption loop and triggering an alarm when the risk accumulation index exceeds a preset alarm threshold. The application solves the problems that the prior art cannot accurately locate a specific abnormal power consumption loop in an electric energy metering box, and that the identification of complex hidden electricity stealing behavior is lagging behind and the false positive and false negative rates are high.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and specifically to a method and device for identifying abnormal electricity consumption in an electricity metering box. Background Technology

[0002] With the deepening of smart grid construction, electricity metering boxes, as key nodes at the end of the power system, undertake the core functions of electricity distribution, accurate metering, and electricity safety protection. Currently, electricity metering boxes are widely used in the power distribution links of residential communities, commercial buildings, and industrial plants. They introduce municipal power through the main incoming line and then distribute it to each user's circuit through multiple outgoing sub-meters, realizing independent metering and control of each electricity branch.

[0003] However, in actual operation, electricity metering boxes face various abnormal electricity consumption behaviors, with electricity theft being the most prominent. Existing methods for identifying abnormal electricity consumption mainly rely on periodic manual inspections or simple comparisons of the electricity consumption data between the main meter and individual user meters in the distribution area through the main station system to detect abnormal line losses. Manual inspections are inefficient and have blind spots, making it difficult to cover the real-time monitoring needs of a large number of metering boxes. While the distribution area line loss analysis method can detect anomalies from a macroscopic perspective, the data aggregation granularity is coarse and has a strong lag. It usually requires accumulating electricity consumption differences over several days or even months to make a judgment, making it impossible to accurately locate the specific metering box and its subordinate outgoing circuit.

[0004] In addition, some solutions have attempted to install independent sensors inside the metering box, such as door opening detection and temperature detection. However, these methods rely on a single criterion and can only reflect signs of physical damage. They cannot effectively identify advanced electricity theft methods that do not involve breaking the box, such as strong magnetic interference, high-frequency interference, neutral wire bypass, or alteration of the internal wiring of the sub-meter. This results in high false alarm and false alarm rates.

[0005] In summary, the existing technology lacks a method for identifying abnormal electricity consumption in electricity metering boxes that can be located locally, in multiple dimensions, and precisely pinpointed to specific outgoing circuits. This results in a severe deficiency in the ability to identify complex and covert electricity theft and a significant delay in response. Summary of the Invention

[0006] This invention provides a method and device for identifying abnormal electricity consumption in an electricity metering box, which solves the problems of existing technologies that make it difficult to accurately locate specific abnormal electricity consumption circuits in the electricity metering box, and that the identification of complex and concealed electricity theft is lagging, with a high rate of missed and false alarms.

[0007] In a first aspect, the present invention provides a method for identifying abnormal electricity consumption in an electricity metering box, comprising: Acquire real-time monitoring timing data of the total incoming current of the power metering box and the current of each outgoing sub-meter; Based on the ratio of the current of each outgoing line submeter to the total current of the incoming line, the loop current missing rate of each outgoing line circuit in multiple monitoring cycles is calculated. Based on the fluctuation range of the circuit current loss rate of each outgoing circuit in multiple consecutive monitoring cycles, the loss fluctuation dispersion of each outgoing circuit is calculated. Based on the correlation between the current missing rate fluctuations of the outgoing circuit and other outgoing circuits, the missing fluctuation correlation degree between multiple circuits of each circuit is calculated to form a missing correlation degree sequence, and the missing fluctuation correlation degree between the first n circuits in the missing correlation degree sequence is taken as the associated circuit of the outgoing circuit. Based on the missing fluctuation dispersion and the missing fluctuation correlation between multiple loops, high-risk loops are screened out, and the risk accumulation of the high-risk loops is tracked over multiple consecutive monitoring periods to calculate the risk accumulation index. When the risk accumulation index exceeds the preset alarm threshold, the outgoing circuit is identified as an abnormal power consumption circuit and an alarm is triggered.

[0008] Secondly, the present invention provides an abnormal electricity consumption identification device for an electricity metering box, comprising: The data acquisition module is used to acquire real-time monitoring time-series data of the total incoming current of the power metering box and the current of each outgoing sub-meter. The loop current missing rate calculation module is used to calculate the loop current missing rate of each outgoing loop in multiple monitoring cycles based on the ratio of the current of each outgoing sub-meter to the total current of the incoming line. The dispersion calculation module is used to calculate the dispersion of the missing current of each outgoing circuit based on the fluctuation amplitude of the circuit current missing rate of each outgoing circuit in multiple consecutive monitoring periods. The correlation calculation module is used to calculate the correlation degree of the current missing rate fluctuation between the outgoing circuit and other outgoing circuits, based on the correlation of the current missing rate fluctuation between the outgoing circuit and other outgoing circuits, to form a missing correlation degree sequence, and to take the first n missing fluctuation correlation degrees between the circuits in the missing correlation degree sequence as the associated circuits of the outgoing circuit. The risk tracking module is used to screen out high-risk loops based on the missing volatility dispersion and the missing volatility correlation between multiple loops, and to perform risk accumulation tracking on the high-risk loops over multiple consecutive monitoring periods to calculate the risk accumulation index. An abnormal alarm module is used to determine the outgoing circuit as an abnormal power consumption circuit and trigger an alarm when the risk accumulation index exceeds a preset alarm threshold.

[0009] One or more technical solutions provided in this invention have at least the following technical effects or advantages: The technical solution of this invention first synchronously collects the effective current values ​​of the main incoming circuit and each outgoing sub-meter circuit of the power metering box at a preset acquisition cycle, forming time-aligned time sequence data. This provides a complete electrical quantity basis between the main incoming line and each outgoing circuit under the same time section for all subsequent analyses, so that any abnormal current change in any outgoing line can be completely recorded and compared with the main incoming line.

[0010] Furthermore, the difference between the total incoming current and the sum of the currents of each outgoing sub-meter is proportionally allocated to each outgoing circuit and aggregated into a circuit current loss rate within the monitoring period. The unknown current loss in the total circuit is decomposed into quantifiable and comparable independent indicators for each circuit, so that minor anomalies in a single circuit can be separated from the macro data of line loss in the transformer area, providing numerical basis for subsequent fluctuation analysis and correlation analysis.

[0011] Furthermore, a time series of loop current missing rates for each outgoing circuit in multiple consecutive monitoring cycles is constructed, and its standard deviation is calculated as the missing fluctuation dispersion. This transforms the intensity of the fluctuation of the loop current missing rate in the time dimension into a single indicator, enabling effective differentiation between the intermittent loop current missing rate jumps caused by electricity theft and the stable random fluctuations caused by normal metering errors.

[0012] Furthermore, the Pearson correlation coefficient between the time series of missing circuit current rates of each outgoing circuit and other circuits was calculated to form a missing correlation degree sequence. The associated circuits of each circuit were determined by two mechanisms: n-value selection or the average correlation degree correction of the whole box. The circuits that were originally analyzed in isolation were organized into a circuit group with correlation as the link, revealing the coordinated abnormal power consumption behavior pattern that could not be found by independent analysis of a single circuit.

[0013] Furthermore, based on the dual thresholds of missing fluctuation dispersion and loop current missing rate, source risk loops are screened, associated risk loops are marked simultaneously, and risk accumulation tracking is performed on all high-risk loops over multiple consecutive monitoring cycles. Instantaneous anomaly judgments are transformed into cumulative risk indices with time memory effects, so that occasional disturbances gradually decline due to the attenuation mechanism and do not trigger false alarms, while persistent anomalies rapidly approach alarm levels due to cycle-by-cycle accumulation. At the same time, independent correlation alarms are triggered for collaborative anomaly loops that are sources of each other, realizing synchronous monitoring and differentiated risk measurement of independent electricity theft and gang electricity theft.

[0014] Finally, the risk accumulation index of each high-risk circuit is compared with the preset alarm threshold. When the threshold is exceeded, it is identified as an abnormal power consumption circuit and an alarm message containing the circuit identifier, risk accumulation index, current current loss rate of the current circuit and related circuits is generated. This provides the final decision output for the entire identification method, ensuring that the alarm is triggered only when the abnormal risk of the circuit accumulates continuously over multiple monitoring cycles and reaches the predetermined threshold. This effectively filters out instantaneous false alarms caused by occasional disturbances and provides maintenance personnel with complete clues for collaborative investigation.

[0015] In summary, the technical solution of the present invention can achieve localized and accurate identification and hierarchical alarm for independent and gang-related electricity theft, effectively solving the problems of rough positioning, delayed response, and high false alarm rate of existing technologies. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a method for identifying abnormal electricity consumption in an electricity metering box provided by the present invention.

[0017] Figure 2 This is a schematic diagram illustrating the calculation process of the risk accumulation index in an abnormal electricity consumption identification method for an electricity metering box provided by the present invention.

[0018] Figure 3 This is a schematic diagram of the abnormal power consumption identification device for an electricity metering box provided by the present invention.

[0019] In the attached diagram, the labels representing each component are as follows: Data acquisition module 11, loop current missing rate calculation module 12, dispersion calculation module 13, correlation calculation module 14, risk tracking module 15, and anomaly alarm module 16. Detailed Implementation

[0020] This invention provides a method and device for identifying abnormal electricity consumption in an electricity metering box, which solves the problems of existing technologies that make it difficult to accurately locate specific abnormal electricity consumption circuits in the electricity metering box, and that the identification of complex and concealed electricity theft is lagging, with a high rate of missed and false alarms.

[0021] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0022] Example 1, as Figure 1 As shown, the present invention provides a method for identifying abnormal electricity consumption in an electricity metering box, the method comprising: S100: Acquire real-time monitoring timing data of the total incoming current and the current of each outgoing sub-meter of the power metering box.

[0023] This step obtains real-time monitoring time-series data of the current in the main incoming circuit and each outgoing sub-meter circuit of the power metering box. This data serves as the basis for all subsequent analyses, ensuring that the current data of the main incoming line and all outgoing sub-meters are completely aligned at the same time point.

[0024] Step S100 in the method provided by the present invention includes: The effective value of the total incoming current is obtained by preset acquisition period and arranged in chronological order of acquisition time to form the time sequence data of the total incoming current. The effective values ​​of the current of each outgoing line submeter are obtained using the preset acquisition period and arranged in chronological order of acquisition time to form the time sequence data of the current of each outgoing line submeter.

[0025] In this step, the effective value of the current in the main circuit of the power metering box is first synchronously collected at a preset collection period, and then arranged in the order of collection time to form the time sequence data of the main current in the incoming line.

[0026] The specific value of the preset acquisition period is determined comprehensively based on the hardware sampling capability, data storage capacity, and real-time requirements of abnormal electricity consumption detection of the concentrator or intelligent metering terminal used in the electricity metering box, and is usually set between 0.5 seconds and 5 seconds. If the acquisition period is set too long, the resolution of the time-series data will be insufficient, which may lead to missed detection of short-duration covert electricity theft; if the acquisition period is set too short, the data volume will surge, putting excessive pressure on the terminal's processing power, communication bandwidth, and storage space. Preferably, the preset acquisition period can be set to 1 second, that is, synchronously acquiring the effective value of the total incoming current and the effective value of the current of each outgoing sub-meter at a frequency of 1 Hz.

[0027] For example, if the acquisition period is set to 1 second, the effective value of the total incoming current is 45.2A at 10:00:00, 45.5A at 10:00:01, 45.1A at 10:00:02, and so on, forming a time sequence of the total incoming current.

[0028] Furthermore, using the same preset acquisition cycle, the effective current values ​​of each outgoing sub-meter are synchronously acquired and arranged in the order of acquisition time to form the current timing data of each outgoing sub-meter.

[0029] For example, using the same 1-second period, at 10:00:00, the effective value of the current of the first outgoing line submeter is synchronously collected as 12.1A, the second as 10.3A, and the third as 8.6A; at 10:00:01, the first as 12.0A, the second as 10.5A, and the third as 8.4A; and so on, forming the current timing data sequence of the three outgoing line submeters respectively.

[0030] It should be noted that the above values ​​are for illustrative purposes only and do not constitute a limitation on the present invention.

[0031] In summary, this step, by synchronously collecting time-series data of the total incoming current and the current of each outgoing sub-meter in real time, provides a complete electrical quantity basis for subsequent analysis between the total incoming line and each outgoing circuit at the same time section, so that any abnormal change in the current of any outgoing line can be completely recorded and compared with the total incoming line.

[0032] S200: Based on the ratio of the current of each outgoing line submeter to the total current of the incoming line, calculate the loop current missing rate of each outgoing line circuit in multiple monitoring cycles.

[0033] This step calculates the loop current loss rate of each outgoing circuit over multiple consecutive monitoring cycles based on the real-time ratio of the current of each outgoing submeter to the total incoming current, thereby transforming abstract line loss or suspected electricity theft into quantifiable indicators.

[0034] Step S200 in the method provided by the present invention includes: Obtain the RMS value of the total incoming current and the RMS value of the current of each outgoing sub-meter at multiple acquisition times; For each outgoing circuit, calculate the circuit current missing rate at each acquisition time. When the historical baseline current of the outgoing circuit is zero, the circuit current missing rate at the acquisition time is zero. Calculate the arithmetic mean of the loop current loss rate of the outgoing circuit at all times within each monitoring period, and use it as the loop current loss rate of the outgoing circuit in that monitoring period.

[0035] In this step, the effective values ​​of the total incoming current and the effective values ​​of the current of each outgoing line sub-meter, which are recorded synchronously at multiple acquisition times, are first extracted from the constructed time-series data and used as input data.

[0036] For example, continuing the scenario above, the preset acquisition period is 1 second. Data is extracted from 10 acquisition times, from 10:00:00 to 10:00:09. At 10:00:00, the effective value of the total incoming current is 45.2A, the effective value of the current of the first outgoing line submeter is 12.1A, the second is 10.3A, and the third is 8.6A; at 10:00:01, the effective value of the total incoming current is 45.5A, the first is 12.0A, the second is 10.5A, and the third is 8.4A; and so on, obtaining all 10 sets of synchronous data.

[0037] Next, for each outgoing circuit, the time-loop current missing rate at each acquisition moment is calculated. The time-loop current missing rate is calculated as follows: Time-loop current missing rate = (Historical baseline current - Effective value of sub-meter current at the current acquisition moment) / Historical baseline current. Where the historical baseline current of a circuit is zero, the time-loop current missing rate of that outgoing circuit is directly set to zero to avoid division by zero errors.

[0038] The historical baseline current refers to the benchmark value obtained before abnormal electricity consumption identification, based on the effective values ​​of the sub-meter current during the historical normal operation period of the outgoing circuit. The historical baseline current is obtained by collecting the effective values ​​of the sub-meter current at all sampling times within a preset time period, such as the past 30 days, where no abnormal electricity consumption behavior has been confirmed. The arithmetic mean of these values ​​is then calculated as the historical baseline current for that circuit. The historical baseline current is pre-calculated and stored in the identification device during system deployment or each periodic update, for use in calculating the current loss rate at subsequent sampling times. If the circuit is a newly connected circuit and there is insufficient historical data, the average sub-meter current of other normal circuits in the same metering box or the circuit's operating data for the past 7 days is used as a temporary historical baseline current, which is updated after sufficient data has been accumulated.

[0039] For example, the historical baseline current of the first outgoing circuit is 12.0A, and the effective value of the sub-meter current at 10:00:00 is 12.1A. The current missing rate of the circuit at that time is approximately (12.0A - 12.1A) / 12.0A ≈ -0.0083, which is set to 0. A negative missing rate indicates that the current is higher than the historical baseline, which is a normal fluctuation and is not considered abnormal. The historical baseline current of the second outgoing circuit is 10.2A, and the effective value of the sub-meter current at 10:00:00 is 10.3A. The current missing rate of the circuit at that time is approximately (10.2A - 10.3A) / 10.2A ≈ -0.0098, which is set to 0. The historical baseline current of the third outgoing circuit is 8.5A. At 10:00:00, the effective value of the sub-meter current is 0.1A, indicating electricity theft causing an abnormally low sub-meter current. The current loss rate for this circuit at that time is calculated as (8.5A - 0.1A) / 8.5A ≈ 0.9882. The loss rates of the three circuits differ significantly, with the theft circuit having the highest loss rate, while the normal circuit has a loss rate close to zero, demonstrating proper differentiation. If the historical baseline current of a circuit is zero, the current loss rate for that circuit at that time is directly set to 0.

[0040] Finally, a monitoring period is set, which includes multiple data acquisition moments. For each outgoing circuit, the arithmetic mean of the circuit current missing rate at all acquisition moments within the monitoring period is calculated, and the result is the circuit current missing rate of that outgoing circuit in that monitoring period.

[0041] The length of the monitoring period is determined comprehensively based on the power consumption characteristics of the load connected to the power metering box, the typical duration of electricity theft, and the requirements for real-time anomaly detection. It is typically set to cover a time span of 10 to 60 data collection points. If the monitoring period is too short, the loop current loss rate is easily affected by instantaneous load fluctuations, causing drastic changes in the indicator and making it difficult to reflect the true trend of abnormal power consumption. If the monitoring period is too long, it will excessively smooth out the sudden changes in the loop current loss rate caused by short-term electricity theft, resulting in missed detections of instantaneous abnormal power consumption. Preferably, the monitoring period is set to cover a time window of 10 consecutive data collection points; that is, when the preset data collection period is 1 second, the monitoring period is 10 seconds.

[0042] For example, the monitoring period is set to the length of 10 data collection points, i.e., 10 seconds. For the first outgoing circuit, the arithmetic mean of the circuit current missing rate at each of the 10 time points from 10:00:00 to 10:00:09 is calculated. Since the current of each meter fluctuates slightly near the baseline at each time point, the missing rate is close to zero, resulting in a circuit current missing rate of approximately 0.002 for the first outgoing circuit within this monitoring period. The second outgoing circuit is calculated similarly, with a circuit current missing rate of approximately 0.003. The third outgoing circuit has a higher missing rate at each time point, resulting in a circuit current missing rate of approximately 0.85.

[0043] It should be noted that the above values ​​are for illustrative purposes only and do not constitute a limitation on the present invention.

[0044] In summary, this step transforms electricity theft or leakage behavior into independent numerical sequences on each circuit that can be compared horizontally and tracked vertically. This allows minor anomalies in a single circuit to be separated from the macro data of line loss in the transformer area, providing numerical evidence for subsequently identifying suspicious circuits by analyzing the fluctuation dispersion of this indicator and the correlation between circuits.

[0045] S300: Based on the fluctuation range of the circuit current loss rate of each outgoing circuit within multiple consecutive monitoring cycles, calculate the loss fluctuation dispersion of each outgoing circuit.

[0046] This step calculates the dispersion of the current loss rate of each outgoing circuit based on the numerical change of the circuit current loss rate over multiple consecutive monitoring periods, and initially screens out suspected objects with abnormal behavior patterns from a large number of circuits.

[0047] Step S300 in the method provided by the present invention includes: The current loss rate of each outgoing circuit is obtained in multiple consecutive monitoring cycles to form a time series sequence of the current loss rate of each outgoing circuit. The standard deviation of the time series of the missing current rate of the circuit is calculated as the missing fluctuation dispersion of the outgoing circuit.

[0048] In this step, the loop current loss rate of each outgoing circuit is first obtained in multiple consecutive monitoring cycles, and arranged in chronological order of the monitoring cycles to form a time sequence of the loop current loss rate of each outgoing circuit.

[0049] The determination of consecutive monitoring periods is as follows: the value should ensure that the sequence duration covers at least one complete typical electricity consumption cycle, while also meeting the statistical stability requirements of standard deviation calculation. Specifically, the lower limit of consecutive monitoring periods is the monitoring period duration in minutes (30 divided by the monitoring period duration) rounded up to ensure that the sequence duration is not less than 30 minutes; the upper limit of consecutive monitoring periods is the monitoring period duration in minutes (120 divided by the monitoring period duration) to ensure that the sequence duration does not exceed 120 minutes, thus guaranteeing the real-time nature of anomaly detection. In real-time calculation, a sliding window method is used, sliding one monitoring period at a time, while always keeping the missing rate data of the most recent consecutive monitoring periods within the window.

[0050] For example, with a monitoring cycle of 10 seconds, the loop current loss rate over the most recent 180 monitoring cycles (30 minutes) is used to construct a time series. The loop current loss rate of the first outgoing loop is 0.002, 0.005, 0.000, 0.008, 0.003, 0.001, 0.006, 0.004, 0.000, and 0.007, totaling 180 values. For this example, the first 10 values ​​are used to construct the loop current loss rate time series for the first loop. Similarly, the loop current loss rate time series for the second and third outgoing loops are obtained.

[0051] Furthermore, for each outgoing loop, the standard deviation of its loop current missing rate time series is calculated to characterize the degree to which the values ​​in the series deviate from their central trend. The standard deviation is used as the missing fluctuation dispersion of the outgoing loop.

[0052] For example, for the first outgoing circuit, the standard deviation of the time series sequence of its circuit current missing rate over 180 monitoring cycles was calculated, yielding a missing rate fluctuation dispersion of 0.0028 for the first circuit. Similarly, for the second outgoing circuit, the standard deviation was calculated, resulting in a missing rate fluctuation dispersion of 0.0019. For the third outgoing circuit, the standard deviation was calculated, yielding a missing rate fluctuation dispersion of 0.033. The third circuit exhibits the largest missing rate fluctuation dispersion, indicating that the fluctuation amplitude of the current missing rate in this circuit is the most severe.

[0053] It should be noted that the above values ​​are for illustrative purposes only and do not constitute a limitation on the present invention.

[0054] In summary, this step transforms the drastic fluctuations in the circuit current loss rate over time into a quantifiable and comparable single indicator. This effectively distinguishes between the intermittent circuit current loss rate jumps caused by electricity theft and the stable random fluctuations caused by normal metering errors. Thus, it provides a first-dimensional independent criterion for initially screening out suspected high-risk circuits with abnormal behavior patterns from numerous outgoing circuits.

[0055] S400: Based on the correlation between the current loss rate fluctuations of the outgoing circuit and other outgoing circuits, calculate the inter-circuit loss fluctuation correlation degree of each circuit to form a loss correlation degree sequence, and take the first n inter-circuit loss fluctuation correlation degrees in the loss correlation degree sequence as the associated circuits of the outgoing circuit.

[0056] Step S400 in the method provided by the present invention includes: Randomly select one outgoing loop as the first loop; Obtain the time series sequence of the loop current missing rate of the first loop in multiple consecutive monitoring cycles and the time series sequence of the loop current missing rate of each of the other outgoing loops in the same metering box in the same time period. Calculate the Pearson correlation coefficient between the time series sequence of the loop current missing rate of the first loop and the time series sequence of the loop current missing rate of each of the other outgoing loops, and use it as the inter-loop missing fluctuation correlation degree between the first loop and the corresponding other outgoing loops; The missing correlation degree between the loops is sorted from largest to smallest to form the missing correlation degree sequence of the first loop; The other outgoing circuits corresponding to the missing fluctuation correlation of the first circuit are selected from the first circuit, and n is obtained based on the historical failure rate in the power metering box.

[0057] In this step, one circuit is randomly selected from all outgoing circuits of the power metering box as the current analysis object, called the first circuit. Subsequently, the correlation between the first circuit and other outgoing circuits in terms of circuit current loss rate fluctuation is analyzed one by one, with the first circuit as the center.

[0058] For example, in the aforementioned scenario, the electricity metering box has three outgoing circuits. The first outgoing circuit is randomly selected as the first circuit, and correlation analysis is then performed on it.

[0059] Secondly, the time series sequence of the loop current missing rate of the first loop in multiple consecutive monitoring cycles is obtained, and the time series sequence of the loop current missing rate of other outgoing loops in the same metering box in the same time period is also obtained to ensure that the data of each loop corresponds one-to-one in time.

[0060] Next, the Pearson correlation coefficients between the time series of the loop current missing rate of the first loop and the time series of the loop current missing rate of each other outgoing loop are calculated, and the absolute value of the obtained coefficients is taken as the correlation degree of loop current missing rate fluctuations between the first loop and the corresponding other outgoing loops. The purpose of taking the absolute value is to simultaneously capture two cooperative anomaly patterns, namely, unidirectional fluctuations and reverse fluctuations. The correlation degree ranges from [0,1], and the closer the value is to 1, the higher the linear correlation of the loop current missing rate fluctuations of the two loops.

[0061] For example, the Pearson correlation coefficient between the time series of current missing rates of loops 1 and 2, calculated based on 180 monitoring periods, is approximately 0.12. Taking the absolute value, the correlation degree of missing rate fluctuations between loops is 0.12. The Pearson correlation coefficient between the time series of current missing rates of loops 1 and 3 is approximately -0.15. Taking the absolute value, the correlation degree of missing rate fluctuations between loops is 0.15.

[0062] Next, the calculated missing correlation degrees between all loops are sorted in descending order to form the missing correlation degree sequence of the first loop.

[0063] For example, sorting the correlation between path 1 and other loops, the correlation between path 1 and path 3 is 0.15, and the correlation between path 1 and path 2 is 0.12. Therefore, the missing correlation sequence for path 1 is [0.15, 0.12], corresponding to path 3 and path 2 respectively.

[0064] Finally, from the missing correlation degree sequence, the other outgoing circuits corresponding to the missing fluctuation correlation degrees of the top n circuits are selected as the associated circuits of the first circuit. The value of n is determined based on the historical failure rate of the electricity metering box; the higher the historical failure rate, the larger the value of n, in order to cover more potentially associated abnormal circuits.

[0065] The historical failure rate can be obtained by querying the historical work order records of the metering box in the power marketing system or the electricity consumption information collection system. Specifically, the historical failure rate can be calculated as the ratio of the number of abnormal electricity consumption verification work orders issued to the metering box within a certain period of time, such as the past year or the past three years, and which were confirmed on-site, to the total number of monitoring cycles or total number of verifications within that period. If the metering box is a newly commissioned device and does not yet have its own historical records, the average historical failure rate of similar metering boxes in the same distribution area, the same model batch, or the same power supply area is used as the initial reference value, and updated after the metering box has accumulated no less than three months of operating data.

[0066] Specifically, the value of n is automatically determined through a preset mapping table. The mapping table is segmented as follows: when the historical failure rate is below 5%, n is 1; when the historical failure rate is between 5% and 10%, n is 2; when the historical failure rate is between 10% and 20%, n is 3; when the historical failure rate reaches 20% or above, n is 20% of the total number of outgoing circuits of the metering box, rounded up, and not less than 4. When the value of n determined by the mapping table is greater than the total number of other outgoing circuits besides the current first circuit, n automatically takes the actual maximum selectable value, i.e., the total number of other outgoing circuits. The mapping table is pre-installed with an abnormal power consumption identification device during system deployment, and the value of n is automatically determined by looking up the table during operation.

[0067] For example, the historical failure rate of the electricity metering box is 3%, and according to the preset mapping rule, n is 1. The loop corresponding to the first position in the missing correlation sequence [0.15, 0.12] of the first loop is selected, which is the third outgoing loop, as the associated loop of the first loop.

[0068] Step S400 in the method provided by the present invention further includes: Obtain the arithmetic mean of the missing correlation degree between all loops in the missing correlation degree sequence of the first loop, and use it as the average correlation degree of the whole box; When the average correlation degree of the entire container is greater than the preset high correlation threshold, the average correlation degree of the entire container is used to correct the missing fluctuation correlation degree to obtain the corrected missing fluctuation correlation degree, and all the outgoing loops whose corrected missing fluctuation correlation degree is greater than the preset corrected correlation threshold are regarded as the associated loops of the first loop.

[0069] Specifically, the arithmetic mean of the missing current fluctuation correlations among all circuits in the missing correlation sequence of the first circuit is first obtained, which is then used as the overall average correlation of the entire box. The overall average correlation reflects the overall level of correlation between the missing current rate fluctuations of the first circuit and all other outgoing circuits in the metering box.

[0070] For example, the missing correlation sequence of the first outgoing loop is [0.15, 0.12]. Calculate the arithmetic mean, and the average correlation of the whole box is (0.15+0.12) / 2=0.135.

[0071] Secondly, the average correlation degree of the entire box is compared with the preset high correlation threshold. The preset high correlation threshold is an empirical threshold for judging whether there is an overall high correlation anomaly in the metering box. When the average correlation degree of the entire box is greater than the high correlation threshold, it indicates that there is a common correlation among the fluctuations in the current loss rate of most circuits in the metering box, which may indicate that there is a group of people colluding to steal electricity. At this time, the correlation degree correction is initiated. When the average correlation degree of the entire box is less than or equal to the high correlation threshold, it indicates that the correlation degree between circuits in the metering box is within the normal range, and the original method of selecting the n value is directly used to determine the associated circuits.

[0072] Specifically, the method for determining the preset high correlation threshold is as follows: collect the correlation data of the missing fluctuation between circuits of the same area or the same model batch of metering boxes during the historical normal operation period, calculate the average correlation of each metering box, and perform probability distribution statistics on these average correlations of the whole box, and take the 95th or 99th percentile of the distribution as the high correlation threshold.

[0073] For example, if the 95th percentile of the average correlation degree of the entire box under normal operating conditions is 0.10, then the high correlation threshold is preset to 0.10. When the real-time calculated average correlation degree of the entire box exceeds 0.10, it indicates that the overall correlation degree between the circuits in the current metering box has significantly deviated from the normal operating condition, and there may be coordinated electricity theft. At this time, the correlation degree correction process is initiated to further screen out abnormal circuits that still stand out under the background of overall high correlation.

[0074] For example, the preset high correlation threshold is 0.10. The average correlation of the entire box for the first channel is 0.135, which is greater than 0.10, indicating that the fluctuation of the loop current missing rate between loops in this metering box has an overall high correlation, and the correlation correction process is activated.

[0075] It should be noted that when the correlation correction process is enabled, the corrected screening results are used as the final basis for determining the correlation loops, and the selection based on the n value is no longer performed; when the correlation correction process is not enabled, the selection method based on the n value is still used to determine the correlation loops.

[0076] Next, the missing fluctuation correlation between each loop is corrected using the overall average correlation of the entire container, resulting in the corrected missing fluctuation correlation. The purpose of this correction is to use the overall average correlation as a baseline to compare the correlation between each loop with the overall level, thereby identifying anomalous correlation loops that still stand out despite the overall high correlation. The correction method is: Corrected missing fluctuation correlation = Original missing fluctuation correlation × (1 - Overall average correlation of the entire container).

[0077] For example, the average correlation coefficient of the first channel is 0.135. After correction, the corrected missing fluctuation correlation coefficient between the first channel and the third channel is 0.15×(1-0.135)=0.1298; the corrected missing fluctuation correlation coefficient between the first channel and the second channel is 0.12×(1-0.135)=0.1038.

[0078] Finally, all outgoing loops with a corrected correlation degree greater than a preset correction correlation threshold are designated as the first loop's associated loops. The preset correction correlation threshold is determined based on the corrected correlation degree distribution characteristics. Preferably, the correction correlation threshold is set as the average correlation degree of the entire container after multiplicative correction, i.e., the average correlation degree of the entire container × (1 - the average correlation degree of the entire container).

[0079] For example, the corrected correlation threshold is set to 0.135 × (1 - 0.135) = 0.1168. The corrected missing fluctuation correlation between path 1 and path 3 is 0.1298 > 0.1168, so it is included in the correlation loop; the corrected missing fluctuation correlation between path 1 and path 2 is 0.1038 < 0.1168, so it is excluded. Finally, the correlation loop of path 1 is the outgoing loop of path 3.

[0080] When the average correlation of the entire container is greater than 0.9, the correlation correction threshold is set to a preset lower limit of 0.01 to avoid the threshold and correction value being too close due to numerical compression.

[0081] Finally, using the above method, each outgoing circuit in the electricity metering box is treated as the first circuit, and the above steps are repeated to determine its associated circuit. When determining the associated circuit, the sign of the correlation degree of the missing data fluctuation between the original circuits corresponding to that associated circuit is recorded simultaneously as the correlation direction. A positive correlation direction indicates that the missing data rates of the two circuits fluctuate in the same direction, and a negative correlation direction indicates that the missing data rates of the two circuits fluctuate in opposite directions.

[0082] For example, when the third path is the first loop, the path ranked first in its missing correlation sequence is the first path, so the associated loop of the third path is the first path. At the same time, the missing fluctuation correlation between the original loops of the third path and the first path is recorded as -0.15, with a negative sign and a negative correlation direction, indicating that the missing rate fluctuations of the third path and the first path have an inverse relationship of mutual inversion.

[0083] It should be noted that the above values ​​are for illustrative purposes only and do not constitute a limitation on the present invention.

[0084] In summary, this step organizes the previously isolated circuits into a circuit group linked by correlation, thereby revealing collaborative abnormal electricity consumption patterns that cannot be detected by independent analysis of a single circuit. This provides an independent criterion for subsequently including related circuits in risk tracking and achieving the joint identification of gang electricity theft.

[0085] S500: Based on the missing fluctuation dispersion and the missing fluctuation correlation between multiple loops, high-risk loops are screened out, and the risk accumulation tracking of the high-risk loops is carried out over multiple consecutive monitoring periods to calculate the risk accumulation index.

[0086] like Figure 2 As shown, this step selects high-risk circuits from all outgoing circuits based on the dispersion of missing fluctuations in each outgoing circuit and the correlation of missing fluctuations between circuits. The selected high-risk circuits are then tracked for risk accumulation over multiple consecutive monitoring periods, and the risk accumulation index of each high-risk circuit is calculated.

[0087] Step S500 in the method provided by the present invention includes: When the missing fluctuation dispersion of the first loop is greater than a preset fluctuation dispersion threshold, and the loop current missing rate of the first loop in the current monitoring period is greater than a preset loop current missing rate threshold, the first loop is marked as a source risk loop, and the associated loop of the first loop is marked as an associated risk loop. Both the source risk loop and the associated risk loop are marked as high-risk loops; Based on the historical fault records of the power metering box, the preset initial cumulative index, attenuation coefficient, and upper limit of the cumulative index are obtained. For each of the high-risk loops, in the monitoring period when it is first marked as a high-risk loop, the initial cumulative index is used as the risk cumulative index for the monitoring period when it is first marked as a high-risk loop. If the first circuit still meets the screening criteria for high-risk circuits in subsequent monitoring cycles, the risk accumulation index of the first circuit is updated using the circuit current missing rate. When the updated risk accumulation index is greater than the upper limit of the accumulation index, the upper limit of the accumulation index is taken as the risk accumulation index. If the first loop no longer meets the screening criteria for high-risk loops in subsequent monitoring periods, the risk accumulation index of the first loop is updated using the attenuation coefficient. When the updated risk accumulation index is less than the initial accumulation index, the initial accumulation index is taken as the risk accumulation index of the first loop.

[0088] In this step, for each outgoing circuit, it is first determined whether it simultaneously meets two conditions: the missing current fluctuation dispersion is greater than a preset fluctuation dispersion threshold, and the circuit current missing rate in the current monitoring period is greater than a preset circuit current missing rate threshold. Circuits that simultaneously meet these two conditions indicate that their circuit current missing rate fluctuates abnormally sharply and the current missing level is significantly high, and they are marked as source risk circuits.

[0089] The specific value of the preset fluctuation dispersion threshold can be obtained by collecting the missing fluctuation dispersion data of each outgoing circuit of the same transformer area or the same model batch of metering boxes during the historical normal operation period, performing probability distribution statistics on the missing fluctuation dispersion of all samples, and taking the 95th or 99th percentile of the distribution as the fluctuation dispersion threshold.

[0090] For example, if the 95th percentile of the dispersion of missing current under normal operating conditions is 0.020, then the dispersion threshold is preset to 0.020. When the dispersion of missing current in a certain outgoing circuit exceeds this threshold, it indicates that the fluctuation range of its circuit current missing rate has significantly deviated from the normal range, and there may be intermittent electricity theft.

[0091] The specific value of the preset circuit current missing rate threshold can be obtained by collecting circuit current missing rate data of each outgoing circuit in the same distribution area or the same model and batch of metering boxes during historical normal operation periods. A probability distribution of the circuit current missing rate for all samples is then performed, and the 95th or 99th percentile of this distribution is taken as the circuit current missing rate threshold. Simultaneously, data from periods with known electricity theft incidents must be excluded to ensure that the statistical sample only reflects normal operating conditions.

[0092] For example, if the 95th percentile of the circuit current loss rate under normal operating conditions is 0.10, then the circuit current loss rate threshold is preset to 0.10. When the circuit current loss rate of a certain outgoing circuit exceeds this threshold, it indicates that there is currently a significant unexplained current loss in that circuit, which is consistent with the characteristics of electricity theft.

[0093] For example, the preset fluctuation dispersion threshold is 0.020, and the preset loop current missing rate threshold is 0.10. The missing fluctuation dispersion of the first channel is 0.0028, which is less than 0.020, and therefore does not meet the condition. The missing fluctuation dispersion of the second channel is 0.0019, which is less than 0.020, and therefore does not meet the condition. The missing fluctuation dispersion of the third channel is 0.033, which is greater than 0.020; the loop current missing rate for the current monitoring period is 0.83, which is greater than 0.10. The third channel meets both conditions and is marked as a source risk loop.

[0094] Secondly, for each outgoing circuit marked as a source risk circuit, its associated circuits are marked as associated risk circuits. Although associated risk circuits may not simultaneously meet the dual conditions of fluctuation dispersion and circuit current missing rate, they are included in the monitoring because they have a significant correlation with the source risk circuit in terms of circuit current missing rate fluctuation, and are suspected of co-managing abnormal power consumption.

[0095] For example, if path 3 is a source risk loop, the associated loop identified in the previous steps is path 1, and the association direction is negative, then path 1 is marked as an associated risk loop. Path 1 itself is not a source risk loop, but because it has a negative missing rate fluctuation association with path 3, it exhibits a reverse anomaly characteristic of one increasing while the other decreases, and is therefore included in the monitoring.

[0096] Secondly, both the source risk loop and related risk loops are included in the high-risk loop set, and risk accumulation tracking is then performed on each loop in the set separately. For example, routes 1 and 3 are both marked as high-risk loops, while route 2 is not included. Risk accumulation tracking is then performed independently on routes 1 and 3 respectively.

[0097] Next, based on the historical fault records of the electricity metering box, the preset initial cumulative index, attenuation coefficient, and cumulative index upper limit are obtained. Among them, the initial cumulative index is the starting risk value when a high-risk circuit first enters the tracking; the attenuation coefficient is the attenuation rate of the risk index when the circuit returns to normal; and the cumulative index upper limit is the capped value of the risk index to prevent numerical overflow.

[0098] Specifically, the initial cumulative index is the average number of monitoring cycles from the first time a circuit identified as having abnormal power consumption by a metering box or similar metering boxes in the same area is marked during a historical period, to the final triggering of an alarm. The principle for setting the initial cumulative index is to ensure that normal circuits, after being mistakenly marked due to occasional fluctuations, can return to near their initial value within a reasonable timeframe due to the attenuation coefficient, while truly abnormal circuits, when continuously meeting high-risk conditions, can quickly accumulate to the alarm threshold. The initial cumulative index is typically set to a small positive value, such as 1.0.

[0099] The attenuation coefficient ranges from (0,1). A value closer to 1 indicates slower attenuation and higher tolerance for intermittent abnormal power consumption; a value closer to 0 indicates faster attenuation and stronger filtering ability for occasional disturbances. The specific value of the attenuation coefficient must simultaneously meet the following two conditions: First, based on the average interval between two adjacent high-risk markers in historical abnormal power consumption events, the attenuation coefficient should cause the risk accumulation index to decay to near the initial accumulation index within that interval. Second, the attenuation coefficient must be greater than 1 minus the ratio of the typical circuit current loss rate to the preset alarm threshold, ensuring that under typical abnormal scenarios, the accumulation rate of the risk index is greater than the attenuation rate, preventing power thieves from escaping alarms by intermittently manipulating the risk index to oscillate below the alarm threshold. Considering both conditions, the preferred range for the attenuation coefficient is 0.85 to 0.95.

[0100] For example, if we statistically analyze historical abnormal power consumption events in the same distribution area and find that the average circuit current loss rate of the abnormal circuit is 0.80, and the preset alarm threshold is 5.0, then the attenuation coefficient must be greater than 1 - 0.80 / 5.0 = 0.84. Simultaneously, if we statistically analyze the average interval between two adjacent high-risk markers for the abnormal circuit, which is 3 monitoring cycles, then the risk accumulation index must attenuate from the alarm threshold of 5.0 to near the initial accumulation index of 1.0 within 3 cycles, resulting in an attenuation coefficient ≈ 0.58. Of the two conditions above, the first condition is more stringent, requiring an attenuation coefficient greater than 0.84, which can be set to 0.90.

[0101] The cumulative index upper limit is set as follows: Based on the longest continuous monitoring period of abnormal electricity consumption events confirmed on-site in the historical fault records of the electricity metering box or similar metering boxes in the same area. The cumulative index upper limit is set as: initial cumulative index + longest continuous period × average loop current loss rate of historical abnormal events. This method ensures that the cumulative index upper limit covers the most severe abnormal scenarios observed historically, while avoiding excessively large values.

[0102] For example, if the longest continuous anomaly in history is 20 monitoring cycles, the average loop current loss rate is 0.80, and the initial cumulative index is 1.0, then the upper limit of the cumulative index = 1.0 + 20 × 0.80 = 17.0. Considering the need to reserve a safety margin in practical applications, the upper limit of the cumulative index is taken as 20.0.

[0103] In the early stages of deployment, when there is a lack of sufficient historical data, the initial cumulative index is set to 1.0, the attenuation coefficient is set to 0.85, the upper limit of the cumulative index is set to 10.0, and the preset alarm threshold is set to 5.0. After accumulating no less than three months of operating data, the above parameters are automatically adjusted to the optimal values ​​that meet the constraints based on the statistical results, such as adjusting the attenuation coefficient to 0.90 and adjusting the upper limit of the cumulative index to 20.0.

[0104] Then, for each high-risk loop, the initial cumulative index is used as the risk cumulative index for the monitoring period in which it is first marked as a high-risk loop. For example, in the current monitoring period, the first loop is marked as a high-risk loop for the first time, and its risk cumulative index is assigned the initial cumulative index of 1.0. The third loop is also marked for the first time, and its risk cumulative index is also assigned the value of 1.0.

[0105] Furthermore, when entering a subsequent monitoring cycle, if the circuit still meets the screening criteria for high-risk circuits, i.e., it remains a source risk circuit or a related risk circuit, then the risk accumulation index is updated by accumulating the circuit current loss rate of the current monitoring cycle. If the updated value exceeds the upper limit of the accumulation index, then the upper limit of the accumulation index is taken as the risk accumulation index for the current cycle to prevent the risk index from growing indefinitely.

[0106] In this step, updating the risk accumulation index of the first circuit using the circuit current loss rate includes: When the high-risk loop is the source risk loop, the risk accumulation index is updated using the loop current missing rate of the subsequent monitoring cycle, which is used as the updated risk accumulation index. When the high-risk loop is a related risk loop, the updated risk accumulation index is calculated and obtained, and the updated risk accumulation index is updated by using the inter-loop missing fluctuation correlation degree of the source risk loop corresponding to the related risk loop, and is used as the updated risk accumulation index. When there are two high-risk loops that are both source and associated risk loops, the risk accumulation index is updated using the loop current loss rate of the subsequent monitoring period. This updated risk accumulation index is then used for the two high-risk loops, and a correlation alarm is triggered.

[0107] Specifically, when the high-risk circuit is the source of the risk, the risk accumulation index is updated by directly using the circuit current missing rate of the current subsequent monitoring period. The update method is: updated risk accumulation index = risk accumulation index of the previous period + circuit current missing rate of the current period.

[0108] For example, the initial cumulative index is 1.0, and the upper limit of the cumulative index is 20.0. In the 11th monitoring period, the third path is still marked as a source risk loop, and the loop current missing rate for the current period is 0.86. The updated risk cumulative index = 1.0 + 0.86 = 1.86, which does not exceed the upper limit of 20.0. The risk cumulative index of the third path for this period is 1.86.

[0109] Furthermore, when a high-risk loop is a related risk loop (i.e., it is only marked as a related risk loop but is not itself a source risk loop), a preliminary update value is first calculated using the current loop current loss rate in the same way as for source risk loops. Then, the inter-loop loss fluctuation correlation between the related risk loop and its corresponding source risk loop is used to weight and correct this preliminary update value. The weighted result is used as the updated risk accumulation index. The anomaly identification of a related risk loop depends on its synergistic relationship with the source loop; therefore, its risk accumulation is regulated by the correlation coefficient. The higher the correlation, the closer the risk accumulation is to full accumulation; the lower the correlation, the more suppressed the risk accumulation.

[0110] For example, in this scenario, the first loop is a related risk loop, not the source risk loop itself. The current loop current missing rate in the current cycle is 0.005. The correlation between the missing fluctuations of the third loop and the first loop is 0.15. Therefore, the risk accumulation index of the first loop after the update = the risk accumulation index of the previous cycle + 0.005 × 0.15 = the risk accumulation index of the previous cycle + 0.00075. Since the missing rate of normal loops is close to zero and the correlation is low, their risk accumulation rate is much lower than that of the source loop.

[0111] Furthermore, when two high-risk circuits are both source and associated risk circuits, it indicates that these two circuits not only exhibit abnormalities in their respective independent circuit current loss rates and fluctuation dispersions, but also show a high degree of correlation in their circuit current loss rate fluctuations, suggesting a strong suspicion of coordinated electricity theft. In this case, the risk accumulation index for each circuit is updated by fully summing their respective current cycle circuit current loss rates, without applying correlation-weighted discounts. Simultaneously, a correlation alarm is triggered to alert maintenance personnel to the coordinated abnormal behavior of this pair of circuits.

[0112] When two high-risk loops are mutually source risk loops and related risk loops, a correlation alarm will only be triggered once in the monitoring cycle when the relationship is first identified. If the relationship continues to exist in subsequent monitoring cycles, it will not be triggered again, and the respective risk accumulation index will only be updated in the alarm information.

[0113] For example, in this scenario, the third loop is the source risk loop, while the first loop is only a related risk loop and not the source risk loop. The two loops are neither source risk loops nor related risk loops to each other, so this step is not triggered. In another scenario, if both the first and third loops simultaneously satisfy the condition that the missing fluctuation dispersion is greater than a preset fluctuation dispersion threshold and the loop current missing rate is greater than a preset loop current missing rate threshold, both are marked as source risk loops. Simultaneously, the related loop for the first loop is the third loop, and the related loop for the third loop is the first loop. Therefore, the two loops are both source risk loops and related risk loops to each other. In this scenario, when this relationship is first identified in the 11th monitoring cycle, both the first and third loops update their risk accumulation index by fully accumulating the loop current missing rate for their respective current cycles, without applying correlation weighting discounts, and trigger a correlation alarm. The alarm information includes the direction of the loop association to assist maintenance personnel in determining whether it is collaborative electricity theft or cross-household electricity theft. If the relationship persists in subsequent monitoring periods, correlation alarms will no longer be triggered repeatedly; only the risk accumulation indices of the two loops will continue to be updated.

[0114] Finally, in subsequent monitoring periods, if the loop no longer meets the screening criteria for a high-risk loop, the risk accumulation index is updated using a decay coefficient. The update method is as follows: the risk accumulation index for the current period equals the risk accumulation index for the previous period multiplied by the decay coefficient. If the decayed value is less than the initial accumulation index, the initial accumulation index is taken as the risk accumulation index for the current period, so that the risk index falls back to the initial baseline after the loop returns to normal, rather than being reset to zero, thus preserving historical risk traces.

[0115] Setting the lower limit of attenuation to the initial cumulative exponent instead of zero is to preserve historical risk traces of the loop, ensuring that loops that have exhibited abnormal behavior remain appropriately sensitive in subsequent monitoring. If the application scenario requires a complete reset after a long period of normal operation, the lower limit of attenuation can be adjusted to zero.

[0116] For example, in the 15th monitoring cycle, the third circuit no longer meets the screening criteria for high-risk circuits, and the circuit current loss rate falls below the threshold. The risk accumulation index of the previous cycle was 4.56, with a decay coefficient of 0.90. After decay, it becomes 4.56 × 0.90 = 4.104, which is greater than the initial accumulation index of 1.0. The risk accumulation index for this cycle is 4.104. If the criteria are not met for several consecutive cycles, the risk accumulation index will continue to decay: 4.104 × 0.90 ≈ 3.694 in the 16th cycle, 3.694 × 0.90 ≈ 3.324 in the 17th cycle, and 3.324 × 0.90 ≈ 2.992 in the 18th cycle, until it decays to below the initial accumulation index of 1.0, at which point it is set to 1.0 and remains unchanged.

[0117] It should be noted that the above values ​​are for illustrative purposes only and do not constitute a limitation on the present invention.

[0118] In summary, this step ensures that occasional disturbances gradually decrease due to the attenuation mechanism and do not trigger false alarms, while persistent anomalies rapidly approach the alarm level due to cycle-by-cycle accumulation. At the same time, it triggers independent correlation alarms for collaborative anomaly loops that are mutually sourced, thus achieving synchronous monitoring and differentiated risk measurement of independent and organized electricity theft.

[0119] S600: When the risk accumulation index exceeds the preset alarm threshold, the outgoing circuit is determined to be an abnormal power consumption circuit and an alarm is triggered.

[0120] This step compares the calculated risk accumulation index with the preset alarm threshold. When the risk accumulation index of any high-risk circuit exceeds the threshold, the outgoing circuit is identified as an abnormal power consumption circuit, and an alarm is triggered to the main station system or maintenance personnel.

[0121] Step S600 in the method provided by the present invention includes: For each of the high-risk loops, obtain the risk accumulation index and preset alarm threshold for the current monitoring period; When the risk accumulation index is greater than or equal to the preset alarm threshold, the high-risk circuit is determined to be an abnormal power consumption circuit, and an abnormal power consumption alarm message is generated. The abnormal power consumption alarm message includes the circuit identifier of the high-risk circuit, the current risk accumulation index, the current circuit current loss rate, and the associated circuits of the high-risk circuit. When the risk accumulation index is less than the preset alarm threshold, the tracking status of the high-risk loop is maintained and no alarm is triggered.

[0122] In this step, for each outgoing circuit marked as a high-risk circuit, the risk accumulation index after accumulation or attenuation update in the current monitoring period is obtained, and the preset alarm threshold is retrieved for comparison.

[0123] The preset alarm threshold refers to a risk judgment threshold that is pre-set and stored in the identification device based on statistical analysis of historical normal operation data of the electricity metering box and known abnormal electricity consumption events before abnormal electricity consumption identification is performed. The specific value can be set by collecting the maximum value distribution of the risk accumulation index of each circuit of the same area or the same model and batch of metering boxes during the historical normal operation period, and taking the 99th percentile or 99.5th percentile of the distribution as the alarm threshold. For example, the preset alarm threshold value is 5.0.

[0124] For example, in the 11th monitoring period, the risk accumulation index of channel 3 is 1.86, and the risk accumulation index of channel 1 is 1.00075. The preset alarm threshold is 5.0, and the risk accumulation indices of both channels are at a low level and have not yet reached the alarm threshold. Assuming that after continuous accumulation and attenuation over several consecutive monitoring periods, in the 17th monitoring period, the risk accumulation index of channel 3 has reached 5.72, and the risk accumulation index of channel 1 is 1.02.

[0125] Furthermore, when the risk accumulation index of a high-risk circuit is greater than or equal to a preset alarm threshold, it indicates that the abnormal risk of the circuit has accumulated to the point where intervention is required. The high-risk circuit is then identified as an abnormal power consumption circuit, and an abnormal power consumption alarm is generated. This alarm information includes at least: the circuit identifier of the high-risk circuit, the current risk accumulation index, the current current loss rate of the circuit, and all associated circuits identified by the high-risk circuit, so that maintenance personnel can jointly verify and coordinate abnormal situations.

[0126] For example, in the 17th monitoring cycle, the risk accumulation index of circuit 3 is 5.72, which is greater than or equal to the preset alarm threshold of 5.0, thus circuit 3 is determined to be an abnormal power consumption circuit. The generated alarm information includes: the circuit identifier is circuit 3 outgoing circuit, the current risk accumulation index is 5.72, the current circuit current missing rate is 0.86, the associated circuit is circuit 1, and the association direction is negative. If the association direction is positive, it prompts an investigation into whether there is any coordinated power theft between the two circuits; if the association direction is negative, it prompts an investigation into whether the source risk circuit is stealing current from the associated circuit across households.

[0127] Furthermore, when the risk accumulation index of a high-risk loop is less than the preset alarm threshold, it indicates that the abnormal risk of the loop has not yet reached the level requiring intervention. The tracking status of the high-risk loop is maintained, no alarm is triggered, and the accumulation or decay update continues in subsequent monitoring cycles based on whether it still meets the high-risk conditions.

[0128] For example, in the 17th monitoring cycle, the risk accumulation index of channel 1 is 1.02, which is less than the preset alarm threshold of 5.0, so no alarm is triggered, and the tracking status of channel 1 is maintained. In the 18th monitoring cycle, if channel 1 is still marked as a high-risk loop, its risk accumulation index will continue to be updated.

[0129] It should be noted that the above values ​​are for illustrative purposes only and do not constitute a limitation on the present invention.

[0130] In summary, this step ensures that an alarm is triggered only when the abnormal risk of a loop accumulates continuously over multiple monitoring cycles and reaches a predetermined threshold. This effectively filters out momentary false alarms caused by occasional disturbances, ensuring the accuracy and traceability of alarm information. At the same time, the alarm information includes the associated loop identifier, providing maintenance personnel with complete clues for collaborative troubleshooting.

[0131] In summary, this invention can achieve localized and accurate identification and hierarchical alarm for independent and group electricity theft, effectively solving the problems of rough positioning, delayed response, and high false alarm rate of existing technologies.

[0132] Example 2, as Figure 3 As shown, the present invention provides an abnormal electricity consumption identification device for an electricity metering box, the device comprising: The data acquisition module 11 is used to acquire real-time monitoring time sequence data of the total incoming current of the power metering box and the current of each outgoing sub-meter.

[0133] The real-time monitoring timing data for acquiring the total incoming current of the power metering box and the current of each outgoing sub-meter includes: The effective value of the total incoming current is obtained by preset acquisition period and arranged in chronological order of acquisition time to form the time sequence data of the total incoming current. The effective values ​​of the current of each outgoing line submeter are obtained using the preset acquisition period and arranged in chronological order of acquisition time to form the time sequence data of the current of each outgoing line submeter.

[0134] The loop current missing rate calculation module 12 is used to calculate the loop current missing rate of each outgoing loop in multiple monitoring cycles based on the ratio of the current of each outgoing submeter to the total current of the incoming line.

[0135] The calculation of the loop current loss rate of each outgoing circuit in multiple monitoring cycles, based on the ratio of the current of each outgoing sub-meter to the total incoming current, includes: Obtain the RMS value of the total incoming current and the RMS value of the current of each outgoing sub-meter at multiple acquisition times; For each outgoing circuit, calculate the circuit current missing rate at each acquisition time. When the historical baseline current of the outgoing circuit is zero, the circuit current missing rate at the acquisition time is zero. Calculate the arithmetic mean of the loop current loss rate of the outgoing circuit at all times within each monitoring period, and use it as the loop current loss rate of the outgoing circuit in that monitoring period.

[0136] The discreteness calculation module 13 is used to calculate the discreteness of the missing current of each outgoing circuit based on the fluctuation amplitude of the circuit current missing rate of each outgoing circuit in multiple consecutive monitoring cycles.

[0137] The calculation of the current loss dispersion of each outgoing circuit based on the fluctuation amplitude of the circuit current loss rate within multiple consecutive monitoring periods includes: The current loss rate of each outgoing circuit is obtained in multiple consecutive monitoring cycles to form a time series sequence of the current loss rate of each outgoing circuit. The standard deviation of the time series of the missing current rate of the circuit is calculated as the missing fluctuation dispersion of the outgoing circuit.

[0138] The correlation calculation module 14 is used to calculate the correlation degree of the missing fluctuation between multiple circuits of each circuit based on the correlation of the current missing rate fluctuation between the outgoing circuit and other outgoing circuits, to form a missing correlation degree sequence, and to take the first n missing fluctuation correlation degrees between the circuits in the missing correlation degree sequence as the associated circuits of the outgoing circuit.

[0139] The step of calculating the correlation degree of current loss rate fluctuations between multiple circuits of each circuit based on the correlation between the current loss rate fluctuations of the outgoing circuit and other outgoing circuits, forming a loss correlation degree sequence, and taking the first n circuits of the loss correlation degree sequence as the associated circuits of the outgoing circuit, includes: Randomly select one outgoing loop as the first loop; Obtain the time series sequence of the loop current missing rate of the first loop in multiple consecutive monitoring cycles and the time series sequence of the loop current missing rate of each of the other outgoing loops in the same metering box in the same time period. Calculate the Pearson correlation coefficient between the time series sequence of the loop current missing rate of the first loop and the time series sequence of the loop current missing rate of each of the other outgoing loops, and use it as the inter-loop missing fluctuation correlation degree between the first loop and the corresponding other outgoing loops; The missing correlation degree between the loops is sorted from largest to smallest to form the missing correlation degree sequence of the first loop; The other outgoing circuits corresponding to the missing fluctuation correlation of the first circuit are selected from the first circuit, and n is obtained based on the historical failure rate in the power metering box.

[0140] This also includes: Obtain the arithmetic mean of the missing correlation degree between all loops in the missing correlation degree sequence of the first loop, and use it as the average correlation degree of the whole box; When the average correlation degree of the entire container is greater than the preset high correlation threshold, the average correlation degree of the entire container is used to correct the missing fluctuation correlation degree to obtain the corrected missing fluctuation correlation degree, and all the outgoing loops whose corrected missing fluctuation correlation degree is greater than the preset corrected correlation threshold are regarded as the associated loops of the first loop.

[0141] The risk tracking module 15 is used to screen out high-risk loops based on the missing fluctuation dispersion and the missing fluctuation correlation between multiple loops, and to perform risk accumulation tracking on the high-risk loops over multiple consecutive monitoring periods to calculate the risk accumulation index.

[0142] Specifically, based on the missing volatility dispersion and the missing volatility correlation among multiple loops, high-risk loops are screened out, and the risk accumulation of these high-risk loops is tracked over multiple consecutive monitoring periods to calculate a risk accumulation index, including: When the missing fluctuation dispersion of the first loop is greater than a preset fluctuation dispersion threshold, and the loop current missing rate of the first loop in the current monitoring period is greater than a preset loop current missing rate threshold, the first loop is marked as a source risk loop, and the associated loop of the first loop is marked as an associated risk loop. Both the source risk loop and the associated risk loop are marked as high-risk loops; Based on the historical fault records of the power metering box, the preset initial cumulative index, attenuation coefficient, and upper limit of the cumulative index are obtained. For each of the high-risk loops, in the monitoring period when it is first marked as a high-risk loop, the initial cumulative index is used as the risk cumulative index for the monitoring period when it is first marked as a high-risk loop. If the first circuit still meets the screening criteria for high-risk circuits in subsequent monitoring cycles, the risk accumulation index of the first circuit is updated using the circuit current missing rate. When the updated risk accumulation index is greater than the upper limit of the accumulation index, the upper limit of the accumulation index is taken as the risk accumulation index. If the first loop no longer meets the screening criteria for high-risk loops in subsequent monitoring periods, the risk accumulation index of the first loop is updated using the attenuation coefficient. When the updated risk accumulation index is less than the initial accumulation index, the initial accumulation index is taken as the risk accumulation index of the first loop.

[0143] The step of updating the risk accumulation index of the first circuit using the circuit current loss rate includes: When the high-risk loop is the source risk loop, the risk accumulation index is updated using the loop current missing rate of the subsequent monitoring cycle, which is used as the updated risk accumulation index. When the high-risk loop is a related risk loop, the updated risk accumulation index is calculated and obtained, and the updated risk accumulation index is updated by using the inter-loop missing fluctuation correlation degree of the source risk loop corresponding to the related risk loop, and is used as the updated risk accumulation index. When there are two high-risk loops that are both source and associated risk loops, the risk accumulation index is updated using the loop current loss rate of the subsequent monitoring period. This updated risk accumulation index is then used for the two high-risk loops, and a correlation alarm is triggered.

[0144] The abnormal alarm module 16 is used to determine the outgoing circuit as an abnormal power consumption circuit and trigger an alarm when the risk accumulation index exceeds the preset alarm threshold.

[0145] Wherein, when the risk accumulation index exceeds the preset alarm threshold, the outgoing circuit is determined to be an abnormal power consumption circuit and an alarm is triggered, including: For each of the high-risk loops, obtain the risk accumulation index and preset alarm threshold for the current monitoring period; When the risk accumulation index is greater than or equal to the preset alarm threshold, the high-risk circuit is determined to be an abnormal power consumption circuit, and an abnormal power consumption alarm message is generated. The abnormal power consumption alarm message includes the circuit identifier of the high-risk circuit, the current risk accumulation index, the current circuit current loss rate, and the associated circuits of the high-risk circuit. When the risk accumulation index is less than the preset alarm threshold, the tracking status of the high-risk loop is maintained and no alarm is triggered.

[0146] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0147] This specification and accompanying drawings are merely illustrative examples of the invention and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its scope. Therefore, if such modifications and modifications fall within the scope of the invention and its equivalents, the invention is intended to include these modifications and modifications.

Claims

1. A method for identifying abnormal electricity consumption in an electricity metering box, characterized in that, include: Acquire real-time monitoring timing data of the total incoming current and the current of each outgoing sub-meter of the power metering box; Based on the ratio of the current of each outgoing line submeter to the total current of the incoming line, the loop current missing rate of each outgoing line circuit in multiple monitoring cycles is calculated. Based on the fluctuation range of the circuit current loss rate of each outgoing circuit in multiple consecutive monitoring cycles, the loss fluctuation dispersion of each outgoing circuit is calculated. Based on the correlation between the current missing rate fluctuations of the outgoing circuit and other outgoing circuits, the missing fluctuation correlation degree between multiple circuits of each circuit is calculated to form a missing correlation degree sequence, and the missing fluctuation correlation degree between the first n circuits in the missing correlation degree sequence is taken as the associated circuit of the outgoing circuit. Based on the missing fluctuation dispersion and the missing fluctuation correlation between multiple loops, high-risk loops are screened out, and the risk accumulation of the high-risk loops is tracked over multiple consecutive monitoring periods to calculate the risk accumulation index. When the risk accumulation index exceeds the preset alarm threshold, the outgoing circuit is identified as an abnormal power consumption circuit and an alarm is triggered.

2. The method for identifying abnormal electricity consumption in an electricity metering box according to claim 1, characterized in that, The real-time monitoring time-series data for acquiring the total incoming current of the power metering box and the current of each outgoing sub-meter includes: The effective value of the total incoming current is obtained by preset acquisition period and arranged in chronological order of acquisition time to form the time sequence data of the total incoming current. The effective values ​​of the current of each outgoing line submeter are obtained using the preset acquisition period and arranged in chronological order of acquisition time to form the time sequence data of the current of each outgoing line submeter.

3. The method for identifying abnormal electricity consumption in an electricity metering box according to claim 1, characterized in that, The calculation of the loop current loss rate for each outgoing circuit across multiple monitoring cycles, based on the ratio of the current of each outgoing sub-meter to the total incoming current, includes: Obtain the RMS value of the total incoming current and the RMS value of the current of each outgoing sub-meter at multiple acquisition times; For each outgoing circuit, calculate the circuit current missing rate at each acquisition time. When the historical baseline current of the outgoing circuit is zero, the circuit current missing rate at the acquisition time is zero. Calculate the arithmetic mean of the loop current loss rate of the outgoing circuit at all times within each monitoring period, and use it as the loop current loss rate of the outgoing circuit in that monitoring period.

4. The method for identifying abnormal electricity consumption in an electricity metering box according to claim 1, characterized in that, The calculation of the current loss dispersion of each outgoing circuit based on the fluctuation amplitude of the circuit current loss rate over multiple consecutive monitoring periods includes: The current loss rate of each outgoing circuit is obtained in multiple consecutive monitoring cycles to form a time series sequence of the current loss rate of each outgoing circuit. The standard deviation of the time series of the missing current rate of the circuit is calculated as the missing fluctuation dispersion of the outgoing circuit.

5. The method for identifying abnormal electricity consumption in an electricity metering box according to claim 1, characterized in that, The step of calculating the correlation degree of current loss rate fluctuations between multiple circuits of each circuit based on the correlation between the current loss rate fluctuations of the outgoing circuit and other outgoing circuits, forming a loss correlation degree sequence, and taking the first n circuits of the loss correlation degree sequence as the associated circuits of the outgoing circuit, includes: Randomly select one outgoing loop as the first loop; Obtain the time series sequence of the loop current missing rate of the first loop in multiple consecutive monitoring cycles and the time series sequence of the loop current missing rate of each of the other outgoing loops in the same metering box in the same time period. Calculate the Pearson correlation coefficient between the time series sequence of the loop current missing rate of the first loop and the time series sequence of the loop current missing rate of each of the other outgoing loops, and use it as the inter-loop missing fluctuation correlation degree between the first loop and the corresponding other outgoing loops; The missing correlation degree between the loops is sorted from largest to smallest to form the missing correlation degree sequence of the first loop; The other outgoing circuits corresponding to the missing fluctuation correlation of the first circuit are selected from the first circuit, and n is obtained based on the historical failure rate in the power metering box.

6. The method for identifying abnormal electricity consumption in an electricity metering box according to claim 5, characterized in that, Also includes: Obtain the arithmetic mean of the missing correlation degree between all loops in the missing correlation degree sequence of the first loop, and use it as the average correlation degree of the whole box; When the average correlation degree of the entire container is greater than the preset high correlation threshold, the average correlation degree of the entire container is used to correct the missing fluctuation correlation degree to obtain the corrected missing fluctuation correlation degree, and all the outgoing loops whose corrected missing fluctuation correlation degree is greater than the preset corrected correlation threshold are regarded as the associated loops of the first loop.

7. The method for identifying abnormal electricity consumption in an electricity metering box according to claim 1, characterized in that, Based on the missing volatility dispersion and the missing volatility correlation among multiple loops, high-risk loops are screened out, and the risk accumulation of these high-risk loops is tracked over multiple consecutive monitoring periods to calculate a risk accumulation index, including: When the missing fluctuation dispersion of the first loop is greater than a preset fluctuation dispersion threshold, and the loop current missing rate of the first loop in the current monitoring period is greater than a preset loop current missing rate threshold, the first loop is marked as a source risk loop, and the associated loop of the first loop is marked as an associated risk loop. Both the source risk loop and the associated risk loop are marked as high-risk loops; Based on the historical fault records of the power metering box, the preset initial cumulative index, attenuation coefficient, and upper limit of the cumulative index are obtained. For each of the high-risk loops, in the monitoring period when it is first marked as a high-risk loop, the initial cumulative index is used as the risk cumulative index for the monitoring period when it is first marked as a high-risk loop. If the first circuit still meets the screening criteria for high-risk circuits in subsequent monitoring cycles, the risk accumulation index of the first circuit is updated using the circuit current missing rate. When the updated risk accumulation index is greater than the upper limit of the accumulation index, the upper limit of the accumulation index is taken as the risk accumulation index. If the first loop no longer meets the screening criteria for high-risk loops in subsequent monitoring periods, the risk accumulation index of the first loop is updated using the attenuation coefficient. When the updated risk accumulation index is less than the initial accumulation index, the initial accumulation index is taken as the risk accumulation index of the first loop.

8. The method for identifying abnormal electricity consumption in an electricity metering box according to claim 7, characterized in that, The step of updating the risk accumulation index of the first circuit using the circuit current loss rate includes: When the high-risk loop is the source risk loop, the risk accumulation index is updated using the loop current missing rate of the subsequent monitoring cycle, which is used as the updated risk accumulation index. When the high-risk loop is a related risk loop, the updated risk accumulation index is calculated and obtained, and the updated risk accumulation index is updated by using the inter-loop missing fluctuation correlation degree of the source risk loop corresponding to the related risk loop, and is used as the updated risk accumulation index. When there are two high-risk loops that are both source and associated risk loops, the risk accumulation index is updated using the loop current loss rate of the subsequent monitoring period. This updated risk accumulation index is then used for the two high-risk loops, and a correlation alarm is triggered.

9. The method for identifying abnormal electricity consumption in an electricity metering box according to claim 1, characterized in that, When the risk accumulation index exceeds a preset alarm threshold, the outgoing circuit is identified as an abnormal power consumption circuit and an alarm is triggered, including: For each of the high-risk loops, obtain the risk accumulation index and preset alarm threshold for the current monitoring period; When the risk accumulation index is greater than or equal to the preset alarm threshold, the high-risk circuit is determined to be an abnormal power consumption circuit, and an abnormal power consumption alarm message is generated. The abnormal power consumption alarm message includes the circuit identifier of the high-risk circuit, the current risk accumulation index, the current circuit current loss rate, and the associated circuits of the high-risk circuit. When the risk accumulation index is less than the preset alarm threshold, the tracking status of the high-risk loop is maintained and no alarm is triggered.

10. An abnormal power consumption identification device for an electricity metering box, characterized in that, A method for identifying abnormal electricity consumption in an electricity metering box according to any one of claims 1 to 9 includes: The data acquisition module is used to acquire real-time monitoring time-series data of the total incoming current of the power metering box and the current of each outgoing sub-meter. The loop current missing rate calculation module is used to calculate the loop current missing rate of each outgoing loop in multiple monitoring cycles based on the ratio of the current of each outgoing sub-meter to the total current of the incoming line. The dispersion calculation module is used to calculate the dispersion of the missing current of each outgoing circuit based on the fluctuation amplitude of the circuit current missing rate of each outgoing circuit in multiple consecutive monitoring periods. The correlation calculation module is used to calculate the correlation degree of the current missing rate fluctuation between the outgoing circuit and other outgoing circuits, based on the correlation of the current missing rate fluctuation between the outgoing circuit and other outgoing circuits, to form a missing correlation degree sequence, and to take the first n missing fluctuation correlation degrees between the circuits in the missing correlation degree sequence as the associated circuits of the outgoing circuit. The risk tracking module is used to screen out high-risk loops based on the missing volatility dispersion and the missing volatility correlation between multiple loops, and to perform risk accumulation tracking on the high-risk loops over multiple consecutive monitoring periods to calculate the risk accumulation index. An abnormal alarm module is used to determine the outgoing circuit as an abnormal power consumption circuit and trigger an alarm when the risk accumulation index exceeds a preset alarm threshold.

Images