Abnormity assessment-based system and method for assessing life extension of electric energy meter in time

Abstract

The invention belongs to the technical field of power system and equipment management, and provides a system and a method for evaluating the life extension of an immediate electric energy meter based on anomaly evaluation, and the system comprises a data collection module which is used for collecting and obtaining the power utilization data of a transformer area group and the work operation data of the immediate electric energy meter; the batch division management module is used for carrying out batch division and management on the immediate electric energy meters based on the work operation data in combination with the immediate electric energy meter equipment parameter data; the anomaly analysis module is used for carrying out anomaly analysis on the electric energy meters which are divided in batches and are in time on the basis of an anomaly detection algorithm in combination with the working operation data to obtain an anomaly analysis result; and the life extension evaluation module is used for calculating a comprehensive life extension score by using a polynomial regression function according to the abnormal analysis result and the marketing data of the electric energy meter, and obtaining life extension evaluation results of the plurality of batches of immediate electric energy meters according to the comprehensive life extension score. According to the invention, online intelligent diagnosis, batch intelligent division, batch abnormity monitoring and batch quality evaluation can be realized.

Description

Technical Field

[0001] This invention relates to the field of power system and equipment management technology, and in particular to a system and method for assessing the life extension of near-dead electricity meters based on anomaly assessment. Background Technology

[0002] With the rapid development of smart grids, electricity meters, as core devices for electricity metering and user electricity consumption data collection, have become crucial for ensuring grid operational efficiency and service quality through accurate monitoring and scientific management of their operating status. Currently, electricity meter management generally adopts a periodic rotation model, that is, replacing meters in batches according to a fixed service life (e.g., 8-10 years). However, this model has significant problems: on the one hand, some electricity meters still have good metering performance after reaching the fixed service life, and forced replacement leads to resource waste; on the other hand, a few electricity meters may fail prematurely due to early quality problems or environmental factors. If this is not identified in time, it will lead to the accumulation of metering errors, abnormally high line losses, and even user complaints and disputes.

[0003] Traditional energy meter condition assessment mainly relies on manual sampling and basic data analysis. For example, a portion of energy meters are periodically sampled for laboratory error testing, and the overall condition of the batch of meters is judged by combining historical verification records. However, this method has the following shortcomings: manual sampling has low coverage, making it difficult to fully reflect the true condition of the batch of meters, and it is time-consuming and labor-intensive, unable to meet the real-time requirements of large-scale power grid management; it is mostly based on a single error value or simple statistical indicators (such as average error) for evaluation, lacking integrated analysis of multi-dimensional information such as return records, negative review data, and abnormal correlations in transformer areas, leading to one-sided evaluation results; the condition assessment model is mostly based on static threshold judgment, without introducing dynamic weight adjustment mechanisms (such as the influence of sales records and user feedback), making it difficult to adapt to the life extension decision-making needs in complex scenarios.

[0004] In recent years, although some studies have attempted to introduce machine learning algorithms to optimize the condition assessment of electricity meters, these studies are mostly limited to a single technical level (such as error prediction) and have not formed a complete solution covering data acquisition, anomaly assessment, and dynamic decision-making. For example, the published patent CN202311436119.9 proposes a wire breakage detection method based on a detection chip, but it depends on hardware costs and cannot be extended to multi-dimensional condition assessment.

[0005] Therefore, it is necessary to provide a system and method for assessing the life extension of near-expiration electricity meters based on anomaly assessment. Summary of the Invention

[0006] This invention provides a system and method for evaluating the lifespan extension of near-expiration energy meters based on anomaly assessment. It enables accurate lifespan assessment of near-expiration energy meters, effectively extending their service life and reducing replacement costs. By acquiring electricity consumption data from distribution areas and operational data from near-expiration energy meters, the timeliness and accuracy of the data are ensured. Multiple anomaly detection algorithms are employed for multi-dimensional data analysis, improving the accuracy and reliability of anomaly detection. Combining marketing data with energy meter data and using a multinomial regression function to calculate a comprehensive lifespan extension score makes the assessment results more objective and scientific. Quality monitoring is conducted on near-expiration energy meters that meet the lifespan extension criteria, further ensuring the reliability and safety of the energy meters.

[0007] This invention provides a near-life extension assessment system for energy meters based on anomaly assessment, comprising:

[0008] The data acquisition module is used to collect real-time electricity consumption data of the distribution area and the operating data of the near-expiration electricity meters using automated metering equipment. The operating data includes error values, voltage, current, power factor, clock deviation, and data acquisition completeness.

[0009] The batch division management module is used to divide and manage near-expiration energy meters in batches based on the K-means++ clustering algorithm, the group electricity consumption data of the transformer area, the working operation data of the near-expiration energy meters, and the equipment parameter data of the near-expiration energy meters.

[0010] The anomaly analysis module is used to perform anomaly analysis on batch-classified near-expiration energy meters based on the set anomaly detection algorithm, combined with the group electricity consumption data of the transformer area and the working operation data of the near-expiration energy meters, and obtain the anomaly analysis results; the anomaly detection algorithm includes a box plot algorithm, DBSCAN clustering algorithm and LSTM-attention mechanism combined model;

[0011] The life extension assessment module is used to calculate a comprehensive life extension score based on the anomaly analysis results and the marketing data of the electricity meters, using a multinomial regression function. Based on the comprehensive life extension score, the life extension assessment results of multiple batches of near-expiration electricity meters are obtained.

[0012] Furthermore, the batch partitioning management module includes a batch partitioning unit and a dynamic management unit;

[0013] The batch division unit is used to intelligently divide the near-expiry energy meters using the K-means++ clustering algorithm, based on the electricity consumption data of the distribution area and the operation data of the near-expiry energy meters, combined with the production batch, service life and operating environment data in the equipment parameter data of the near-expiry energy meters, to obtain the near-expiry energy meters marked with batch labels.

[0014] The dynamic management unit is used to dynamically adjust the batch labels of near-expiration energy meters based on the group electricity consumption data of the transformer area and the working operation data of the near-expiration energy meters, according to the set batch adjustment rules. The batch adjustment rules are determined based on the equipment health status and usage risk level of the near-expiration energy meters. Among them, the equipment health status is used to quantitatively characterize the overall working operation status of the near-expiration energy meters, and the usage risk level is used to quantitatively characterize the potential usage risks of the near-expiration energy meters.

[0015] Furthermore, the anomaly analysis module includes an electricity consumption behavior data acquisition unit, an anomaly detection algorithm selection unit, and an anomaly analysis implementation unit;

[0016] The electricity consumption behavior data acquisition unit is used to acquire electricity consumption behavior data based on the electricity consumption data of the transformer area group;

[0017] The anomaly detection algorithm selection unit is used to screen and determine the anomaly detection algorithm; the anomaly detection algorithm includes the box plot algorithm, the DBSCAN clustering algorithm, and the LSTM-attention mechanism combined model;

[0018] The anomaly analysis implementation unit is used to identify local outliers in error values ​​based on the box plot algorithm, obtaining the first anomaly analysis result; based on the DBSCAN clustering algorithm, it calculates the spatial density of user electricity consumption patterns in the electricity consumption behavior data, and clusters users with spatial density below a preset spatial density threshold into anomaly consumption groups, obtaining the second anomaly analysis result; based on the LSTM-attention mechanism combined model, it inputs historical error value data of the electricity meter and outputs the predicted error value for the future time window. If the predicted error value exceeds the allowable range, it is considered an anomaly, obtaining the third anomaly analysis result; the LSTM-attention mechanism combined model has 3 hidden layers and 2 attention heads; in the LSTM-attention mechanism combined model, the attention weight is calculated using an additive attention model; combining the first, second, and third anomaly analysis results, the anomaly analysis results for multiple batches of near-expiration electricity meters are obtained.

[0019] Furthermore, based on the box plot algorithm, the drawn box plots are used to identify local outliers in the error values, obtaining the first anomaly analysis results, including:

[0020] Statistical analysis of the error values ​​was performed to calculate the interquartile range of the error values.

[0021] Based on the quartile range, set upper and lower threshold values;

[0022] Error values ​​exceeding the upper or lower threshold are considered local outliers.

[0023] Local outliers are used as the first anomaly analysis result.

[0024] Furthermore, the life extension assessment module includes an anomaly assessment unit, a life extension score calculation unit, and a life extension assessment result generation unit;

[0025] The anomaly assessment unit is used to self-test the anomaly analysis results based on the set confidence level evaluation model, and generate anomaly cause data with different confidence levels.

[0026] The lifespan extension score calculation unit is used to calculate the comprehensive lifespan extension score based on the marketing data of the electricity meter using a multinomial regression function, and obtain the comprehensive lifespan extension score result.

[0027] The life extension assessment result generation unit is used to obtain the life extension assessment result based on the abnormal cause data and the comprehensive life extension score result. The life extension assessment result is sent to the energy meter through the configured edge computing gateway to trigger the device self-test program and realize the closed-loop control of assessment and execution.

[0028] Furthermore, based on the established confidence level evaluation model, the anomaly analysis results are self-tested, generating anomaly cause data at different confidence levels, including:

[0029] The abnormal data in the anomaly analysis results are compared with the real-time data collected in the metering automation system to see if the voltage and current in the abnormal data and the real-time data are synchronously abnormal, thus obtaining the first comparison result; the abnormal data in the anomaly analysis results are compared with the user complaint records in the electricity meter marketing data to see if the abnormal data is associated with negative reviews or return data, thus obtaining the second comparison result.

[0030] Based on the first and second comparison results, the confidence levels are defined as follows: if the comparison results show that all abnormal data are confirmed, the abnormal data is defined as high confidence; if the comparison results show that some abnormal data are confirmed, the abnormal data is defined as low confidence; if the comparison results show that some abnormal data are not confirmed or there are isolated cases, the abnormal data is defined as low confidence.

[0031] Based on the confidence level, corresponding anomaly cause data is generated.

[0032] Furthermore, based on the marketing data of the electricity meter, a comprehensive lifespan extension score is calculated using a multinomial regression function to obtain the comprehensive lifespan extension score result, including:

[0033] Based on the marketing data of electricity meters, extract the number of return records, the number of negative reviews, the sales record ranking value, and the positive review ranking value of electricity meters;

[0034] The negative feedback value is calculated based on the number of return records and their weight, and the number of negative reviews and their weight.

[0035] Using sales record ranking value and positive review ranking value as variables, and negative feedback value as a constant, a comprehensive life extension score is calculated using a multinomial regression function. The comprehensive life extension score calculation formula is as follows:

[0036] G=α*S xs +β*S hp -γ*(δ*N th +ε*N cp )

[0037] In the above formula, G represents the comprehensive life extension score, α represents the weight of the sales record ranking value, and S xs S represents the sales record ranking value, β represents the weight of the positive review ranking value, and S hp Represents the ranking value for positive reviews, (δ*N) th +ε*N cp ) represents the negative feedback value, γ represents the global weight coefficient of the negative feedback value, δ represents the weight of the number of return records, and N th ε represents the number of return records, ε represents the weight of the number of negative reviews, and N represents the number of negative reviews. cp This represents the number of negative reviews; the weights α, β, γ, δ, and ε all range from 0.5 to 1.2.

[0038] Furthermore, based on the anomaly cause data and comprehensive life extension score results, life extension assessment results for multiple batches of near-expiration energy meters were obtained, including:

[0039] Set a first comprehensive life extension score threshold and a second comprehensive life extension score threshold;

[0040] If the comprehensive life extension score is greater than the first comprehensive life extension score threshold, the near-expiration electricity meter is determined to meet the life extension conditions.

[0041] If the comprehensive life extension score is less than the first comprehensive life extension score threshold but greater than the second comprehensive life extension score threshold, the near-expiration electricity meter is determined to need to be re-inspected.

[0042] If the comprehensive life extension score is less than the second comprehensive life extension score threshold, the near-expiration electricity meter is determined to need to be replaced.

[0043] Based on the determination that the near-expiry electricity meters meet the life extension conditions and the determination that the near-expiry electricity meters need to be replaced, the batch of the near-expiry electricity meters that meet the life extension conditions is set as the life extension batch, and the batch of the near-expiry electricity meters that need to be replaced is set as the replacement batch.

[0044] Obtain all near-expiration energy meters in the batch to be extended in service life; and obtain all near-expiration energy meters in the batch to be replaced in service life.

[0045] Based on the set batch quality consistency index, the quality consistency index of the batch to be extended is calculated to obtain a first index value. If the first index value is greater than the set first index threshold, then the entire batch of near-expiration energy meters in the batch to be extended is evaluated as having its entire batch extended. The quality consistency index of the batch to be replaced is calculated to obtain a second index value. If the second index value is greater than the set second index threshold, then the entire batch of near-expiration energy meters in the batch to be replaced is evaluated as having its entire batch replaced. The batch quality consistency index is used to quantitatively evaluate whether the near-expiration energy meters in the same batch are consistent in terms of equipment performance, working status, location distribution characteristics, and the degree of synchronization of performance degradation. The higher the value, the better the batch quality consistency.

[0046] Furthermore, it also includes a life extension monitoring module, which is used to monitor the quality of near-expiration energy meters that meet the life extension conditions. If a decline in the quality of the life extension energy meter is detected, a re-verification process is triggered. The life extension monitoring module includes a remaining life prediction model construction unit, a dynamic monitoring mechanism generation unit, and a judgment and response unit.

[0047] The remaining lifetime prediction model construction unit is used to construct a remaining lifetime prediction model using the gradient boosting decision tree algorithm to predict the remaining lifetime of near-expiration energy meters that meet the life extension conditions. The input of the remaining lifetime prediction model includes the energy meter's error trend, running time, average ambient temperature and humidity, and historical failure count, and the output is the remaining lifetime prediction value.

[0048] The dynamic monitoring mechanism generation unit is used to calculate the actual monitoring frequency based on the defined monitoring frequency benchmark, the calculability rate of the distribution area, and the work order hit rate, and to implement dynamic monitoring based on the actual monitoring frequency; the formula for calculating the actual monitoring frequency is:

[0049]

[0050] In the above formula, f represents the actual monitoring frequency, ρ represents the dynamic adjustment coefficient, the calculability rate of the transformer area and the work order hit rate are set; f0 represents the baseline value of the monitoring frequency, and M represents the predicted value of the remaining lifespan; the value of the dynamic adjustment coefficient ρ is determined by fitting historical data: when the calculability rate of the transformer area is greater than 95% and the work order hit rate is greater than 90%, ρ is 1.2; otherwise, ρ is 0.8.

[0051] The judgment and response unit is used to make judgments and responses based on the monitoring results obtained from the actual monitoring frequency. Specifically, if the increase in error value in two consecutive monitoring sessions is greater than the set amplitude threshold, or the remaining life prediction value is less than the set time threshold, a re-inspection strategy is generated; if the error value in a single monitoring session is greater than the set error value threshold, an emergency replacement strategy is generated.

[0052] Methods for extending the lifespan of near-expiration electricity meters based on anomaly assessment include:

[0053] Automated metering equipment is used to collect real-time electricity consumption data of the distribution area and the operating data of the near-expiration electricity meters; the operating data includes error value, voltage, current, power factor, clock deviation and data collection completeness rate;

[0054] The K-means++ clustering algorithm is used to classify and manage near-expiration energy meters in batches based on the electricity consumption data of the distribution area and the operation data of the near-expiration energy meters, combined with the equipment parameter data of the near-expiration energy meters.

[0055] Based on the established anomaly detection algorithm, combined with the electricity consumption data of the distribution area and the operating data of the near-expiration electricity meters, anomaly analysis is performed on the near-expiration electricity meters that have been divided into batches to obtain the anomaly analysis results; the anomaly detection algorithm includes a box plot algorithm, DBSCAN clustering algorithm and LSTM-attention mechanism combined model.

[0056] Based on the anomaly analysis results and combined with the marketing data of the electricity meters, a comprehensive life extension score was calculated using a multinomial regression function. Based on the comprehensive life extension score, the life extension assessment results of multiple batches of near-expiration electricity meters were obtained.

[0057] Compared with existing technologies, this invention has the following advantages and beneficial effects: it enables accurate lifespan assessment of near-expiration electricity meters, effectively extending their service life and reducing replacement costs; the data acquisition module obtains real-time electricity consumption data of the distribution area and the operating data of near-expiration electricity meters, ensuring the timeliness and accuracy of the data; the anomaly analysis module employs multiple anomaly detection algorithms to perform multi-dimensional data analysis, improving the accuracy and reliability of anomaly detection; the lifespan assessment module combines electricity meter marketing data and uses a multinomial regression function to calculate a comprehensive lifespan score, making the assessment results more objective and scientific. Simultaneously, the system also includes a lifespan monitoring module to monitor the quality of near-expiration electricity meters that meet the lifespan extension conditions, further ensuring the reliability and safety of the electricity meters.

[0058] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0059] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0060] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0061] Figure 1 This is a schematic diagram of the life extension assessment system for near-expiration energy meters based on anomaly assessment.

[0062] Figure 2 This is a schematic diagram of the anomaly analysis module structure;

[0063] Figure 3 This is a schematic diagram illustrating the steps of a method for assessing the lifespan extension of near-expiration energy meters based on anomaly evaluation. Detailed Implementation

[0064] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0065] This invention provides a near-life extension assessment system for energy meters based on anomaly assessment, such as... Figure 1 As shown, it includes:

[0066] The data acquisition module is used to collect real-time electricity consumption data of the distribution area and the operating data of the near-expiration electricity meters using automated metering equipment. The operating data includes error values, voltage, current, power factor, clock deviation, and data acquisition completeness.

[0067] The batch division management module is used to divide and manage near-expiration energy meters in batches based on the K-means++ clustering algorithm, the group electricity consumption data of the transformer area, the working operation data of the near-expiration energy meters, and the equipment parameter data of the near-expiration energy meters.

[0068] The anomaly analysis module is used to perform anomaly analysis on batch-classified near-expiration energy meters based on the set anomaly detection algorithm, combined with the group electricity consumption data of the transformer area and the working operation data of the near-expiration energy meters, and obtain the anomaly analysis results; the anomaly detection algorithm includes a box plot algorithm, DBSCAN clustering algorithm and LSTM-attention mechanism combined model;

[0069] The life extension assessment module is used to calculate a comprehensive life extension score based on the anomaly analysis results and the marketing data of the electricity meters, using a multinomial regression function. Based on the comprehensive life extension score, the life extension assessment results of multiple batches of near-expiration electricity meters are obtained.

[0070] The working principle of this technical solution is as follows: In order to realize the life extension assessment system for near-expiration energy meters based on anomaly assessment, this invention uses a data acquisition module to collect real-time data on the electricity consumption of the distribution area and the operating parameters of the near-expiration energy meters through metering automation equipment. These data serve as the basic input of the system, including but not limited to key operating indicators such as energy meter error value, voltage and current parameters, power factor, clock deviation, and data acquisition completeness rate. Taking a residential community power supply area as an example, the metering automation system can monitor the electricity load curve of each user in real time and record various operating parameters of the near-expiration energy meters. Among them, the error value reflects the metering accuracy, the voltage and current parameters characterize the working load status, the power factor reflects the electricity consumption characteristics, the clock deviation affects the timing accuracy, and the data acquisition completeness rate assesses the reliability of the data.

[0071] After data collection is completed, the system uses the K-means++ clustering algorithm through the batch division management module to scientifically classify near-expiration electricity meters based on the electricity consumption characteristics of the distribution area and the operating parameters of the electricity meters, combined with the inherent attributes of the equipment. For example, in view of the differences in electricity load characteristics of different building types, the system can automatically identify and classify batches with similar electricity consumption patterns and equipment parameters. This classification method significantly improves the pertinence and evaluation efficiency of subsequent analysis.

[0072] The anomaly analysis module employs a multi-algorithm fusion strategy, including a combined model of box plot algorithm, DBSCAN clustering algorithm, and LSTM-attention mechanism, to detect anomalies in the classified energy meters. The box plot algorithm determines the normal range of parameters through interquartile range and identifies outliers exceeding the threshold; the DBSCAN clustering algorithm detects outliers based on density features; and the LSTM-attention mechanism predicts normal states through time series modeling and identifies deviation samples. This multi-dimensional analysis method can accurately identify various anomalies.

[0073] Ultimately, the life extension assessment module integrates anomaly analysis results with marketing data (including equipment service life, maintenance records, etc.) and uses a multinomial regression model to calculate a comprehensive life extension score. This scoring system provides a quantitative basis for power companies to formulate equipment replacement strategies and achieve optimal resource allocation. For example, for batches that meet the score, maintenance measures can be taken to extend their service life, thereby reducing equipment replacement costs.

[0074] The beneficial effects of the above technical solution are as follows: The solution provided in this embodiment enables accurate lifespan assessment of near-expiration electricity meters, effectively extending their service life and reducing replacement costs; the data acquisition module obtains real-time electricity consumption data from the distribution area and operational data from near-expiration electricity meters, ensuring data timeliness and accuracy; the anomaly analysis module employs multiple anomaly detection algorithms to perform multi-dimensional data analysis, improving the accuracy and reliability of anomaly detection; and the lifespan assessment module combines electricity meter marketing data with a multinomial regression function to calculate a comprehensive lifespan score, making the assessment results more objective and scientific.

[0075] In one embodiment, the batch partitioning management module includes a batch partitioning unit and a dynamic management unit;

[0076] The batch division unit is used to intelligently divide the near-expiry energy meters using the K-means++ clustering algorithm, based on the electricity consumption data of the distribution area and the operation data of the near-expiry energy meters, combined with the production batch, service life and operating environment data in the equipment parameter data of the near-expiry energy meters, to obtain the near-expiry energy meters marked with batch labels.

[0077] The dynamic management unit is used to dynamically adjust the batch labels of near-expiration energy meters based on the group electricity consumption data of the transformer area and the working operation data of the near-expiration energy meters, according to the set batch adjustment rules. The batch adjustment rules are determined based on the equipment health status and usage risk level of the near-expiration energy meters. Among them, the equipment health status is used to quantitatively characterize the overall working operation status of the near-expiration energy meters, and the usage risk level is used to quantitatively characterize the potential usage risks of the near-expiration energy meters.

[0078] The working principle of this technical solution is as follows: The K-means++ clustering algorithm utilizes the electricity consumption data of the distribution area, such as the peak and off-peak electricity consumption in the area during different time periods, the fluctuation range of electricity consumption, and the operating data of near-expiration electricity meters, such as changes in metering accuracy, measurement errors of current and voltage, and communication stability. It also combines this data with the production batch, service life, and operating environment data from the near-expiration electricity meter equipment parameter data. The operating environment data can include the temperature and humidity of the installation location, and whether there is strong electromagnetic interference. Through this multi-dimensional data, the algorithm can find similar features in the data and classify near-expiration electricity meters with similar characteristics into the same batch. For example, near-expiration electricity meters with the same production batch, similar service life, and installed in the same temperature and humidity environment, and with similar trends in metering accuracy changes during peak and off-peak electricity consumption periods, will be classified into the same batch and labeled with the corresponding batch label. This allows for preliminary classification and management of near-expiration electricity meters, providing a basis for subsequent evaluation and processing.

[0079] The dynamic management unit dynamically adjusts the batch labels of near-expiration energy meters based on the electricity consumption data of the distribution area and the operating data of the near-expiration energy meters, according to the set batch adjustment rules. The batch adjustment rules are determined based on the equipment health status and usage risk level of the near-expiration energy meters. The equipment health status is a quantitative representation of the overall operating status of the near-expiration energy meters, such as by measuring the stability of the metering accuracy and the aging degree of the internal circuit components. The usage risk level is a quantitative representation of the potential usage risks of the near-expiration energy meters, such as the likelihood of metering errors due to energy meter failure and the impact on the safe operation of the power grid. When the operating data of the near-expiration energy meters changes, causing a change in their equipment health status or usage risk level, the dynamic management unit will adjust their batch labels according to the rules.

[0080] In practical applications, suppose there is a power supply area comprising multiple residential communities, containing a batch of near-expiration electricity meters. These meters are from the same production batch, have a service life of 8-9 years, and are all installed in residential buildings, where the operating environment is relatively stable. Initially, the batch division unit uses the K-means++ clustering algorithm to group these meters into a batch, labeled "Batch A," based on their metering accuracy and communication stability during daily electricity use. Over time, some of the residential buildings housing these meters undergo circuit renovations, causing changes in the power environment and resulting in temporary voltage fluctuations. These meters are affected, their metering accuracy becomes unstable, their equipment health declines, and their usage risk level increases. After detecting these changes, the dynamic management unit, according to the batch adjustment rules, moves these meters from "Batch A" to "Batch B." "Batch B" is specifically designed for near-expiration electricity meters with lower equipment health and higher usage risk, allowing for closer monitoring and evaluation of these meters to determine if early replacement or repair is necessary, thereby extending their effective service life and ensuring the stable operation of the power grid.

[0081] The beneficial effects of the above technical solution are as follows: The solution provided in this embodiment can realize refined management of near-expiration electricity meters, intelligent batch division based on multi-dimensional data, and fully consider the characteristics of electricity meters under different operating environments and the power consumption of the distribution area, making the division results more scientific and reasonable; the dynamic management unit dynamically adjusts the batch labels according to the health status of the equipment and the level of use risk, and timely reflects the status changes of near-expiration electricity meters, allowing power operation and maintenance personnel to take differentiated maintenance strategies for different batches of electricity meters.

[0082] In one embodiment, such as Figure 2 As shown, the anomaly analysis module includes an electricity consumption behavior data acquisition unit, an anomaly detection algorithm selection unit, and an anomaly analysis implementation unit;

[0083] The electricity consumption behavior data acquisition unit is used to acquire electricity consumption behavior data based on the electricity consumption data of the transformer area group;

[0084] The anomaly detection algorithm selection unit is used to screen and determine the anomaly detection algorithm; the anomaly detection algorithm includes the box plot algorithm, the DBSCAN clustering algorithm, and the LSTM-attention mechanism combined model;

[0085] The anomaly analysis implementation unit is used to identify local outliers in error values ​​based on the box plot algorithm, obtaining the first anomaly analysis result; based on the DBSCAN clustering algorithm, it calculates the spatial density of user electricity consumption patterns in the electricity consumption behavior data, and clusters users with spatial density below a preset spatial density threshold into anomaly consumption groups, obtaining the second anomaly analysis result; based on the LSTM-attention mechanism combined model, it inputs historical error value data of the electricity meter and outputs the predicted error value for the future time window. If the predicted error value exceeds the allowable range, it is considered an anomaly, obtaining the third anomaly analysis result; the LSTM-attention mechanism combined model has 3 hidden layers and 2 attention heads; in the LSTM-attention mechanism combined model, the attention weight is calculated using an additive attention model; combining the first, second, and third anomaly analysis results, the anomaly analysis results for multiple batches of near-expiration electricity meters are obtained.

[0086] The working principle of this technical solution is as follows:

[0087] First, the electricity consumption behavior data acquisition unit acquires electricity consumption behavior data based on the electricity consumption data of the transformer area. In real-world scenarios, a transformer area contains the electricity consumption information of many users, which records data such as electricity consumption and electricity consumption time at different times. For example, in a residential area, the electricity consumption of different families varies at different times such as daytime, nighttime, weekdays, and weekends. The electricity consumption behavior data acquisition unit will collect this diverse data.

[0088] Next, the anomaly detection algorithm selection process was used to determine the appropriate algorithm. Three algorithms were chosen: box plot, DBSCAN clustering, and a combined LSTM-attention mechanism model. The choice of algorithm depends on the specific analytical needs and data characteristics. For example, when focusing on local outliers in the data, the box plot algorithm is preferred. If the goal is to identify anomalous electricity consumption groups with similar patterns and low spatial density, the DBSCAN clustering algorithm is more suitable. And for situations requiring prediction of future error values ​​from electricity meters, the combined LSTM-attention mechanism model comes into play.

[0089] Then, the anomaly analysis implementation unit performs anomaly analysis based on different algorithms: Based on the box plot algorithm, using the drawn box plots, it identifies local outliers in the error values ​​to obtain the first anomaly analysis result; taking a factory transformer area as an example, the factory has multiple workshops, each with different electrical equipment and varying electricity consumption. Using the box plot algorithm, box plots of the error values ​​of the electricity meters in each workshop are drawn. If the error value of the electricity meter in a certain workshop far exceeds the upper and lower boundaries of the box plot, then the electricity meter in that workshop is identified as a local outlier, possibly indicating a fault or metering anomaly; for example, under normal circumstances, the error values ​​of the electricity meters in each workshop fluctuate within a reasonable range, but on one day, the error value of the electricity meter in one workshop suddenly deviates significantly, and this anomaly can be clearly identified through the box plot; based on the DBSCAN clustering algorithm, the spatial density of user electricity consumption patterns in the electricity behavior data is calculated, and... Users with spatial density below a preset spatial density threshold are clustered into an abnormal electricity consumption group, yielding a second anomaly analysis result. Assuming a commercial area's electricity consumption patterns can be broadly categorized into several types, such as restaurants and retail stores, the DBSCAN clustering algorithm calculates spatial density based on factors like usage time and electricity consumption. If the electricity consumption patterns of certain stores differ from the majority of businesses, and their spatial density is below the preset threshold, they will be clustered into an abnormal electricity consumption group. For example, some stores operate at night, with their electricity consumption primarily concentrated in the evening, unlike most... Retail stores operating during the daytime exhibit significant differences in electricity consumption patterns. After spatial density calculations, they may be identified as abnormal electricity users, requiring further investigation into potential electricity theft or equipment malfunctions. Based on an LSTM-attention mechanism combined model, historical error values ​​from electricity meters are input, and predicted error values ​​for future time windows are output. If the predicted error value exceeds the allowable range, it is considered an anomaly, yielding a third anomaly analysis result. Taking a large office building area as an example, this building contains multiple elevators, air conditioners, and other equipment. Electricity meter error values ​​are affected by equipment operating status, seasons, and other factors. Inputting historical error value data from the electricity meters over the past few months or even years into the LSTM-attention mechanism combined model allows the model to consider the variation patterns of error values ​​over different time periods and various influencing factors. For example, during the high temperatures of summer, frequent air conditioner use may cause fluctuations in electricity meter error values. If the model predicts that the error value on a future day exceeds the allowable range, it indicates a potential anomaly in the electricity meter. For instance, if the model predicts a significant increase in the error value of an elevator's electricity meter in the coming days, it is necessary to inspect and maintain the elevator's electrical equipment in advance.

[0090] Finally, by combining the results of the first, second, and third anomaly analyses, anomaly analysis results for multiple batches of near-expiration energy meters were obtained. This comprehensive analysis method allows for the detection of anomalies in near-expiration energy meters from multiple perspectives, improving the accuracy and reliability of the detection.

[0091] The beneficial effects of the above technical solution are as follows: the solution provided in this embodiment can comprehensively and accurately assess the operating status of near-expiration electricity meters and effectively identify potential fault hazards; by combining a variety of advanced anomaly detection algorithms, not only is the accuracy and efficiency of anomaly detection improved, but the robustness and adaptability of the system are also greatly enhanced; by cross-validating the results of electricity consumption behavior spatial density analysis (DBSCAN) and error prediction (LSTM), the correlation between voltage sag events and error mutations can be unexpectedly discovered, reducing the false alarm rate.

[0092] In one embodiment, based on the box plot algorithm, local outliers of error values ​​are identified using the plotted box plots to obtain the first anomaly analysis result, including:

[0093] Statistical analysis of the error values ​​was performed to calculate the interquartile range of the error values.

[0094] Based on the quartile range, set upper and lower threshold values;

[0095] Error values ​​exceeding the upper or lower threshold are considered local outliers.

[0096] Local outliers are used as the first anomaly analysis result.

[0097] The working principle of this technical solution is as follows: The box plot algorithm visually displays the distribution of error values. When calculating error values, the median of all error values ​​is first calculated and used as the median of the box plot. Next, the first quartile (Q1) and the third quartile (Q3) are determined, representing the 25% and 75% concentrations of the error value dataset, respectively. The distance between Q1 and Q3 is called the interquartile range (IQR). Based on the interquartile range, upper and lower thresholds are set. Typically, the upper threshold is set to Q3 plus 1.5 times the IQR, while the lower threshold is... Setting the IQR to Q1 minus 1.5 times can effectively identify outliers in error values ​​in most cases, i.e., those error values ​​that significantly deviate from the normal range. When the error value exceeds the set upper or lower threshold, it is considered a local outlier. These outliers represent abnormal states of the electricity meter during operation, such as a sudden increase or decrease in measurement error, which is caused by factors such as aging or damage of internal components of the electricity meter or interference from the external environment. Using the identified local outliers as the first anomaly analysis result can provide important clues for subsequent anomaly detection and analysis.

[0098] The beneficial effects of the above technical solution are as follows: by adopting the solution provided in this embodiment, through in-depth analysis of outliers, the operating status of the electricity meter can be further understood, potential fault hazards can be discovered and dealt with in a timely manner, thereby ensuring the accuracy and reliability of the electricity meter.

[0099] In one embodiment, the life extension assessment module includes an anomaly assessment unit, a life extension score calculation unit, and a life extension assessment result generation unit.

[0100] The anomaly assessment unit is used to self-test the anomaly analysis results based on the set confidence level evaluation model, and generate anomaly cause data with different confidence levels.

[0101] The lifespan extension score calculation unit is used to calculate the comprehensive lifespan extension score based on the marketing data of the electricity meter using a multinomial regression function, and obtain the comprehensive lifespan extension score result.

[0102] The life extension assessment result generation unit is used to obtain the life extension assessment result based on the abnormal cause data and the comprehensive life extension score result. The life extension assessment result is sent to the energy meter through the configured edge computing gateway to trigger the device self-test program and realize the closed-loop control of assessment and execution.

[0103] The working principle of this technical solution is as follows: First, the anomaly assessment unit receives the anomaly analysis results from the previous step. These results include local outliers in the meter's error values. Based on a pre-set confidence evaluation model, this unit conducts in-depth analysis of these outliers and attempts to assign them different levels of confidence. This process aims to filter out the outliers that are most likely to represent real faults or abnormal states, while excluding false alarms that may be caused by accidental factors or noise interference. Next, the life extension score calculation unit utilizes the meter's marketing data, which may include the meter's historical usage records, maintenance records, user feedback, etc. This unit uses a multinomial regression function to comprehensively analyze this data to calculate a comprehensive life extension score. This score reflects the meter's overall health status and potential life extension potential in its current state. Finally, the life extension assessment result generation unit integrates the anomaly cause data and the comprehensive life extension score result to generate a final life extension assessment report. This report will detail the potential fault hazards of the meter, the causes of abnormal states, and corresponding life extension measures and recommendations.

[0104] The beneficial effects of the above technical solution are as follows: By adopting the solution provided in this embodiment, accurate life extension assessment of near-expiration electricity meters can be achieved, effectively avoiding premature replacement of healthy electricity meters. At the same time, it can also identify electricity meters with potential fault risks in a timely manner and take corresponding repair or replacement measures, thereby greatly improving the efficiency and safety of electricity meter use. In addition, this solution also makes full use of electricity meter marketing data and conducts a comprehensive and objective evaluation of the health status of electricity meters through data analysis methods using multinomial regression functions, providing strong data support for the life extension management of electricity meters.

[0105] In one embodiment, based on a defined confidence level evaluation model, the anomaly analysis results are self-tested to generate anomaly cause data with different confidence levels, including:

[0106] The abnormal data in the anomaly analysis results are compared with the real-time data collected in the metering automation system to see if the voltage and current in the abnormal data and the real-time data are synchronously abnormal, thus obtaining the first comparison result; the abnormal data in the anomaly analysis results are compared with the user complaint records in the electricity meter marketing data to see if the abnormal data is associated with negative reviews or return data, thus obtaining the second comparison result.

[0107] Based on the first and second comparison results, the confidence levels are defined as follows: if the comparison results show that all abnormal data are confirmed, the abnormal data is defined as high confidence; if the comparison results show that some abnormal data are confirmed, the abnormal data is defined as low confidence; if the comparison results show that some abnormal data are not confirmed or there are isolated cases, the abnormal data is defined as low confidence.

[0108] Based on the confidence level, corresponding anomaly cause data is generated.

[0109] The working principle of this technical solution is as follows: First, the invention uses a built-in confidence evaluation model to self-verify the initially detected abnormal data. This process involves comparing the detected abnormal data with real-time data in the metering automation system, paying particular attention to whether voltage and current data synchronously show abnormalities, in order to verify the authenticity of the abnormal data. If the abnormal data matches the abnormal situation in the real-time data, this will serve as the first verification, i.e., the first comparison result. Then, the abnormal data is compared with user complaint records in the electricity meter marketing system to check whether these abnormal data are related to negative reviews or returns from users. This comparison helps to further verify the authenticity of abnormal data from the perspective of actual user experience, forming a second comparison result. Then, based on these two comparison results, a confidence level is defined. If all comparison results confirm the existence of abnormal data, then this abnormal data will be considered high confidence. If only some comparison results confirm the abnormal data, or if some abnormal data is not confirmed, then this abnormal data will be considered low confidence. In particular, if the comparison results show that the abnormal data is not confirmed, or the abnormal data is isolated, it will also be defined as low confidence. Finally, based on the defined confidence level, corresponding anomaly cause data is generated. This data not only reflects the potential problems of the electricity meter, but also includes a confidence assessment of the authenticity of these problems, providing strong support for subsequent processing and decision-making.

[0110] The beneficial effects of the above technical solution are as follows: By adopting the solution provided in this embodiment, the accuracy and efficiency of evaluating abnormal data of near-expiration electricity meters can be significantly improved. Through the built-in confidence evaluation model, the system can automatically perform multi-level verification on the initially detected abnormal data to ensure the authenticity of the data. At the same time, the abnormal data is compared with real-time data and user complaint records, which not only technically verifies the existence of abnormal data, but also verifies it from the perspective of user feedback, making the evaluation results more comprehensive and reliable.

[0111] In one embodiment, based on the marketing data of the electricity meter, a comprehensive lifespan extension score is calculated using a multinomial regression function to obtain the comprehensive lifespan extension score result, including:

[0112] Based on the marketing data of electricity meters, extract the number of return records, the number of negative reviews, the sales record ranking value, and the positive review ranking value of electricity meters;

[0113] The negative feedback value is calculated based on the number of return records and their weight, and the number of negative reviews and their weight.

[0114] Using sales record ranking value and positive review ranking value as variables, and negative feedback value as a constant, a comprehensive life extension score is calculated using a multinomial regression function. The comprehensive life extension score calculation formula is as follows:

[0115] G=α*S xs +β*S hp -γ*(δ*N th +ε*N cp )

[0116] In the above formula, G represents the comprehensive life extension score, α represents the weight of the sales record ranking value, and S xs S represents the sales record ranking value, β represents the weight of the positive review ranking value, and S hp Represents the ranking value for positive reviews, (δ*N) th +ε*N cp ) represents the negative feedback value, γ represents the global weight coefficient of the negative feedback value, δ represents the weight of the number of return records, and N th ε represents the number of return records, ε represents the weight of the number of negative reviews, and N represents the number of negative reviews. cp This represents the number of negative reviews; the weights α, β, γ, δ, and ε all range from 0.5 to 1.2.

[0117] The working principle of this technical solution is as follows: First, the invention collects marketing data on electricity meters, including but not limited to the number of return records, the number of negative reviews, sales ranking values, and positive review ranking values. Then, based on preset weights, the number of return records and the number of negative reviews are weighted to calculate a negative feedback value. The purpose of this step is to quantify the degree of negative feedback received by the electricity meter in the market. Subsequently, the invention uses the sales ranking value and positive review ranking value as independent variables and the negative feedback value as a constant term, inputting them into a multinomial regression function. Based on these input data, the multinomial regression function, through mathematical calculations, derives a comprehensive lifespan extension score. This score represents the electricity meter's potential for lifespan extension. A quantitative assessment is used, with higher scores indicating greater potential for extending the lifespan of the electricity meter. In the comprehensive lifespan extension score calculation formula, the weights of each variable are set by the system based on historical data and expert experience to ensure the accuracy and rationality of the score. These weight values ​​reflect the importance of different factors in assessing the lifespan extension potential of the electricity meter. Finally, based on the comprehensive lifespan extension score results, this invention conducts a lifespan extension assessment on near-expiration electricity meters. If the score is higher than a preset threshold, it recommends extending the lifespan of the electricity meter to prolong its service life and reduce replacement costs. Conversely, if the score is lower than the threshold, replacement is recommended. In this way, the system can help power companies maintain and manage electricity meters more scientifically and rationally.

[0118] The beneficial effects of the above technical solution are as follows: by adopting the solution provided in this embodiment, the multi-dimensional data of the electricity meter can be comprehensively considered, including marketing feedback and performance, and quantitative evaluation can be carried out through a scientific mathematical model, avoiding the limitations of traditional evaluation methods that rely too much on subjective judgment or single indicators.

[0119] In one embodiment, based on anomaly cause data and comprehensive life extension score results, life extension assessment results for multiple batches of near-expiration energy meters are obtained, including:

[0120] Set a first comprehensive life extension score threshold and a second comprehensive life extension score threshold; the first comprehensive life extension score threshold is 85; the second comprehensive life extension score threshold is 60;

[0121] If the comprehensive life extension score is greater than the first comprehensive life extension score threshold, the near-expiration electricity meter is determined to meet the life extension conditions.

[0122] If the comprehensive life extension score is less than the first comprehensive life extension score threshold but greater than the second comprehensive life extension score threshold, the near-expiration electricity meter is determined to need to be re-inspected.

[0123] If the comprehensive life extension score is less than the second comprehensive life extension score threshold, the near-expiration electricity meter is determined to need to be replaced.

[0124] Based on the determination that the near-expiry electricity meters meet the life extension conditions and the determination that the near-expiry electricity meters need to be replaced, the batch of the near-expiry electricity meters that meet the life extension conditions is set as the life extension batch, and the batch of the near-expiry electricity meters that need to be replaced is set as the replacement batch.

[0125] Obtain all near-expiration energy meters in the batch to be extended in service life; and obtain all near-expiration energy meters in the batch to be replaced in service life.

[0126] Based on the set batch quality consistency index, the quality consistency index of the batch to be extended is calculated to obtain a first index value. If the first index value is greater than the set first index threshold, then the entire batch of near-expiration energy meters in the batch to be extended is evaluated as having its entire batch extended. The quality consistency index of the batch to be replaced is calculated to obtain a second index value. If the second index value is greater than the set second index threshold, then the entire batch of near-expiration energy meters in the batch to be replaced is evaluated as having its entire batch replaced. The batch quality consistency index is used to quantitatively evaluate whether the near-expiration energy meters in the same batch are consistent in terms of equipment performance, working status, location distribution characteristics, and the degree of synchronization of performance degradation. The higher the value, the better the batch quality consistency.

[0127] The working principle of this technical solution is as follows: the comprehensive life extension score serves as the basic indicator for measuring whether a single near-expiry energy meter is suitable for life extension, while the batch quality consistency index considers whether the energy meters can be extended or replaced as a whole batch from the perspective of the overall batch. The purpose of this is to ensure the performance evaluation of a single energy meter while taking into account the overall quality of the batch, thereby improving the scientific nature and efficiency of the evaluation.

[0128] The comprehensive life extension score is a quantitative score calculated based on the comprehensive data of abnormal causes. It reflects the overall performance status of the near-expiration energy meter. A first comprehensive life extension score threshold (85 points) and a second comprehensive life extension score threshold (60 points) are set to classify near-expiration energy meters into different categories so that different handling measures can be taken. When the comprehensive life extension score is greater than the first comprehensive life extension score threshold (85 points), it indicates that the near-expiration energy meter is currently performing well and can meet the requirements for continued use, so it is determined that it meets the life extension conditions. If the comprehensive life extension score is between the first comprehensive life extension score threshold (85 points) and the second comprehensive life extension score threshold (60 points), it means that although the energy meter is currently performing well, there may be some potential minor problems, requiring further re-inspection to accurately assess its performance, so it is determined that re-inspection is necessary. When the comprehensive life extension score is less than the second comprehensive life extension score threshold (60 points), it indicates that the near-expiration energy meter is performing poorly, and continued use may lead to malfunctions, affecting the accuracy of electricity metering, so it is determined that replacement is necessary.

[0129] Based on the assessment results of individual near-expiration energy meters, the batches containing near-expiration energy meters that meet the life extension criteria are designated as the batch to be extended, and the batches containing near-expiration energy meters that need to be replaced are designated as the batch to be replaced. Then, the batch quality consistency index is used to further evaluate these two batches. The batch quality consistency index is used to quantitatively assess whether the near-expiration energy meters in the same batch are consistent in terms of equipment performance, working status, location distribution characteristics, and the degree of synchronization of performance degradation. A higher value indicates better batch quality consistency. The first index value is calculated for the batch to be extended. If this value is greater than the set first index threshold, it indicates that the overall performance of the energy meters in this batch is relatively consistent and good, and an assessment result for extending the life of the entire batch can be made. This is because if the performance consistency of the energy meters within a batch is high, then the assessment result of a single meter can represent the situation of the entire batch. Extending the life of the entire batch can improve work efficiency and reduce costs. The second index value is calculated for the batch to be replaced. If this value is greater than the set second index threshold, it indicates that the overall performance of the energy meters in this batch is poor and relatively consistent, so an assessment result for replacing the entire batch is made.

[0130] In practical applications, suppose a power company has three batches (A, B, and C) of near-expiration electricity meters that need to be evaluated. For batch A, a comprehensive lifespan extension score is calculated. Most meters in batch A have a score greater than 85, with only a few scoring between 70 and 80. This indicates that most meters in batch A perform well and meet the lifespan extension criteria, while a few require re-inspection. For batch B, the comprehensive lifespan extension scores are mostly between 60 and 80, indicating that most meters in this batch need re-inspection to determine if their lifespan can be extended. For batch C, the comprehensive lifespan extension scores are generally below 60, meaning that most meters in this batch need replacement. The batch quality consistency index (first index value) for batch A is calculated, and this value... The index value is greater than the set first index threshold, indicating that the electricity meters in batch A have high consistency in terms of equipment performance and working status. Although a few meters need to be re-inspected, batch A can be judged as a batch to be extended in terms of lifespan, and an assessment result for extending the lifespan of the entire batch can be made. The batch quality consistency index (second index value) of batch C is calculated. This value is greater than the set second index threshold, indicating that the electricity meters in batch C have poor performance and high consistency. Therefore, batch C is judged as a batch to be replaced, and an assessment result for replacing the entire batch can be made. For batch B, since the assessment results of individual meters often require re-inspection and the batch quality consistency index is low, it indicates that the performance of electricity meters in the batch varies greatly. It is not possible to simply process the entire batch. Each meter needs to be re-inspected before a decision is made.

[0131] The beneficial effects of the above technical solution are as follows: by adopting the solution provided in this embodiment, it is possible to achieve a scientific and reasonable assessment of electricity meters nearing their expiration date, providing strong support for the equipment maintenance and management of power companies.

[0132] In one embodiment, the system further includes a life extension monitoring module for monitoring the quality of near-expiration energy meters that meet the life extension conditions. If a quality decline is detected in the life extension energy meter, a re-verification process is triggered. The life extension monitoring module includes a remaining life prediction model construction unit, a dynamic monitoring mechanism generation unit, and a judgment and response unit.

[0133] The remaining lifetime prediction model construction unit is used to construct a remaining lifetime prediction model using the gradient boosting decision tree algorithm to predict the remaining lifetime of near-expiration energy meters that meet the life extension conditions. The input of the remaining lifetime prediction model includes the energy meter's error trend, running time, average ambient temperature and humidity, and historical failure count, and the output is the remaining lifetime prediction value.

[0134] The dynamic monitoring mechanism generation unit is used to calculate the actual monitoring frequency based on the defined monitoring frequency benchmark, the calculability rate of the distribution area, and the work order hit rate, and to implement dynamic monitoring based on the actual monitoring frequency; the formula for calculating the actual monitoring frequency is:

[0135]

[0136] In the above formula, f represents the actual monitoring frequency, ρ represents the dynamic adjustment coefficient, the calculability rate of the transformer area and the work order hit rate are set; f0 represents the baseline value of the monitoring frequency, and M represents the predicted value of the remaining lifespan; the value of the dynamic adjustment coefficient ρ is determined by fitting historical data: when the calculability rate of the transformer area is greater than 95% and the work order hit rate is greater than 90%, ρ is 1.2; otherwise, ρ is 0.8.

[0137] The judgment and response unit is used to make judgments and responses based on the monitoring results obtained from the actual monitoring frequency. Specifically, if the increase in error value in two consecutive monitoring sessions is greater than the set amplitude threshold, or the remaining life prediction value is less than the set time threshold, a re-inspection strategy is generated; if the error value in a single monitoring session is greater than the set error value threshold, an emergency replacement strategy is generated.

[0138] The working principle of this technical solution is as follows: The entire near-expiration energy meter life extension assessment system based on anomaly assessment takes the life extension monitoring module as its core, and conducts comprehensive and dynamic quality monitoring on near-expiration energy meters that meet the life extension conditions. Through a series of scientific calculations and judgments, it ensures the normal operation and metering accuracy of the energy meters.

[0139] The remaining life prediction model building unit uses the gradient boosting decision tree algorithm to construct the model. This algorithm can fully consider multiple key factors of the electricity meter, such as the error trend reflecting the change in the meter's accuracy over time; the operating time reflecting the degree of use of the electricity meter; the average ambient temperature and humidity considering the impact of the external environment on the electricity meter's performance; and the number of historical failures illustrating the stability of the electricity meter. By comprehensively analyzing these factors, the model can predict the remaining life of the electricity meter relatively accurately. Suppose there is an electricity meter A nearing its expiration date, whose error trend has been slowly increasing recently, its operating time has reached 10 years, the average temperature and humidity of its environment are high and fluctuate greatly, and it has had 3 historical failures. After inputting these data into the remaining life prediction model, the model outputs a remaining life prediction value of 2 years for the electricity meter.

[0140] The dynamic monitoring mechanism generation unit calculates the actual monitoring frequency based on the monitoring frequency benchmark, the calculability rate of the distribution area, and the work order hit rate. The monitoring frequency benchmark is a basic monitoring frequency setting, while the calculability rate of the distribution area reflects the proportion of electricity meter data in that area that can be used for calculation and analysis. The work order hit rate reflects the effectiveness of maintenance work order processing. By introducing a dynamic adjustment coefficient, the actual monitoring frequency is adjusted according to different situations of the calculability rate of the distribution area and the work order hit rate, thus achieving dynamic monitoring. In practical applications, the monitoring frequency benchmark is set to once a month. For the distribution area where the near-expiration electricity meter A is located, the statistical calculability rate of the distribution area is 96%, and the work order hit rate is 92%. According to the rule, the dynamic adjustment coefficient is 1.2. The remaining life prediction model yields a remaining life prediction value M of 2 years (i.e., 730 days) for electricity meter A. Substituting this into the actual monitoring frequency calculation formula, the actual monitoring frequency is calculated to be approximately 1.2 times per month, meaning that the electricity meter may be monitored approximately every 25 days.

[0141] The judgment and response unit makes judgments and takes action based on the monitoring results obtained from the actual monitoring frequency. It sets different judgment conditions, including the magnitude of error increase, the predicted remaining life, and the error threshold. It generates corresponding strategies based on different monitoring situations to ensure the normal operation of the electricity meter. For example, during dynamic monitoring of electricity meter A, if two consecutive monitoring tests show that its error increase reaches 1.5%, while the set threshold is 1%; or if the predicted remaining life drops to 1.5 years, while the set time threshold is 1.8 years, the judgment and response unit will generate a re-inspection strategy, arranging for professionals to re-inspect the electricity meter to determine whether it can continue to operate safely. If, in a monitoring test, the error value of electricity meter A reaches 5%, while the set error threshold is 3%, the judgment and response unit will immediately generate an emergency replacement strategy to replace the electricity meter in a timely manner to avoid problems caused by inaccurate metering.

[0142] Once the response unit generates a re-inspection strategy or an emergency replacement strategy, the system will trigger a re-verification process to ensure the quality and metering accuracy of near-expiration electricity meters and guarantee the stable operation of the power system.

[0143] The beneficial effects of the above technical solution are as follows: By adopting the solution provided in this embodiment, the intelligent life extension monitoring module realizes the whole-chain management of the near-expiration electricity meter from prediction to monitoring to response, which effectively improves the utilization efficiency of the electricity meter and the reliability of the power grid operation.

[0144] A method for extending the lifespan of near-expiration electricity meters based on anomaly assessment, such as... Figure 3 As shown, it includes:

[0145] Automated metering equipment is used to collect real-time electricity consumption data of the distribution area and the operating data of the near-expiration electricity meters; the operating data includes error value, voltage, current, power factor, clock deviation and data collection completeness rate;

[0146] The K-means++ clustering algorithm is used to classify and manage near-expiration energy meters in batches based on the electricity consumption data of the distribution area and the operation data of the near-expiration energy meters, combined with the equipment parameter data of the near-expiration energy meters.

[0147] Based on the established anomaly detection algorithm, combined with the electricity consumption data of the distribution area and the operating data of the near-expiration electricity meters, anomaly analysis is performed on the near-expiration electricity meters that have been divided into batches to obtain the anomaly analysis results; the anomaly detection algorithm includes a box plot algorithm, DBSCAN clustering algorithm and LSTM-attention mechanism combined model.

[0148] Based on the anomaly analysis results and combined with the marketing data of the electricity meters, a comprehensive life extension score was calculated using a multinomial regression function. Based on the comprehensive life extension score, the life extension assessment results of multiple batches of near-expiration electricity meters were obtained.

[0149] The working principle of this technical solution is as follows: First, this invention uses automated metering equipment to collect real-time electricity consumption data from the distribution area and operational data from near-expiration electricity meters. The operational data includes error values, voltage, current, power factor, clock deviation, and data collection completeness, providing a comprehensive and accurate information foundation for subsequent analysis. Next, a preset anomaly detection algorithm is used to perform multi-dimensional anomaly analysis on the collected data. This step aims to identify abnormal patterns or deviations from normal conditions in the data. The anomaly detection algorithm includes a box plot algorithm, a DBSCAN clustering algorithm, and a combined LSTM-attention mechanism model. The results of the anomaly analysis are then compared with the electricity meter's operational data. By combining sales data, which typically includes information such as the purchase date, historical maintenance records, and usage environment of the electricity meter, and by introducing a multinomial regression function, the comprehensive life extension potential of the electricity meter can be quantitatively assessed. The multinomial regression function is chosen because it can capture the nonlinear relationships in the data. It can output a comprehensive life extension score based on the input multidimensional data. This score directly reflects the feasibility and potential benefits of extending the life of the electricity meter. Finally, based on the calculated comprehensive life extension score, a life extension assessment result can be obtained. This result not only provides decision-makers with a clear basis for whether to extend the life of the electricity meter, but also provides an important reference for subsequent monitoring and response strategy formulation.

[0150] The beneficial effects of the above technical solution are as follows: by adopting the solution provided in this embodiment, it is possible to achieve accurate life extension assessment of electricity meters nearing their expiration date, effectively extend the service life of electricity meters, and reduce replacement costs.

[0151] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A life extension assessment system for near-expiration energy meters based on anomaly assessment, characterized in that, include: The data acquisition module is used to collect real-time electricity consumption data of the distribution area and the operating data of the near-expiration electricity meters using automated metering equipment. The operating data includes error values, voltage, current, power factor, clock deviation, and data acquisition completeness. The batch division management module is used to divide and manage near-expiration energy meters in batches based on the K-means++ clustering algorithm, the group electricity consumption data of the transformer area, the working operation data of the near-expiration energy meters, and the equipment parameter data of the near-expiration energy meters. The anomaly analysis module is used to perform anomaly analysis on batch-classified near-expiration energy meters based on the set anomaly detection algorithm, combined with the group electricity consumption data of the transformer area and the working operation data of the near-expiration energy meters, and obtain the anomaly analysis results; the anomaly detection algorithm includes a box plot algorithm, DBSCAN clustering algorithm and LSTM-attention mechanism combined model; The life extension assessment module is used to calculate a comprehensive life extension score based on the anomaly analysis results and the marketing data of the electricity meters, using a multinomial regression function. Based on the comprehensive life extension score, the life extension assessment results of multiple batches of near-expiration electricity meters are obtained.

2. The near-term energy meter life extension assessment system based on anomaly assessment according to claim 1, characterized in that, The batch partitioning management module includes a batch partitioning unit and a dynamic management unit; The batch division unit is used to intelligently divide the near-expiry energy meters using the K-means++ clustering algorithm, based on the electricity consumption data of the distribution area and the operation data of the near-expiry energy meters, combined with the production batch, service life and operating environment data in the equipment parameter data of the near-expiry energy meters, to obtain the near-expiry energy meters marked with batch labels. The dynamic management unit is used to dynamically adjust the batch labels of near-expiration energy meters based on the group electricity consumption data of the transformer area and the working operation data of the near-expiration energy meters, according to the set batch adjustment rules. The batch adjustment rules are determined based on the equipment health status and usage risk level of the near-expiration electricity meters; among them, the equipment health status is used to quantitatively characterize the overall working status of the near-expiration electricity meters; the usage risk level is used to quantitatively characterize the potential usage risks of the near-expiration electricity meters.

3. The near-term energy meter life extension assessment system based on anomaly assessment according to claim 1, characterized in that, The anomaly analysis module includes an electricity consumption behavior data acquisition unit, an anomaly detection algorithm selection unit, and an anomaly analysis implementation unit; The electricity consumption behavior data acquisition unit is used to acquire electricity consumption behavior data based on the electricity consumption data of the transformer area group; The anomaly detection algorithm selection unit is used to screen and determine the anomaly detection algorithm; the anomaly detection algorithm includes the box plot algorithm, the DBSCAN clustering algorithm, and the LSTM-attention mechanism combined model; The anomaly analysis implementation unit is used to identify local outliers of error values ​​based on the box plot algorithm and the drawn box plot to obtain the first anomaly analysis result; and to calculate the spatial density of users' electricity consumption patterns in the electricity consumption behavior data based on the DBSCAN clustering algorithm, and to cluster users whose spatial density is lower than the preset spatial density threshold into anomaly electricity consumption groups to obtain the second anomaly analysis result. Based on the LSTM-attention mechanism combined model, the historical error data of the electricity meter is input, and the predicted error value of the future time window is output. If the predicted error value exceeds the allowable range, it is considered an anomaly, and the third anomaly analysis result is obtained. The LSTM-attention mechanism combined model has 3 hidden layers and 2 attention heads. In the LSTM-attention mechanism combined model, the attention weight is calculated using an additive attention model. By combining the results of the first, second, and third anomaly analyses, anomaly analysis results were obtained for multiple batches of near-expiration electricity meters.

4. The near-term energy meter life extension assessment system based on anomaly assessment according to claim 3, characterized in that, Based on the box plot algorithm, local outliers of error values ​​are identified using the plotted box plots to obtain the first anomaly analysis results, including: Statistical analysis of the error values ​​was performed to calculate the interquartile range of the error values. Based on the quartile range, set upper and lower threshold values; Error values ​​exceeding the upper or lower threshold are considered local outliers. Local outliers are used as the first anomaly analysis result.

5. The near-term energy meter life extension assessment system based on anomaly assessment according to claim 1, characterized in that, The life extension assessment module includes an anomaly assessment unit, a life extension score calculation unit, and a life extension assessment result generation unit; The anomaly assessment unit is used to self-test the anomaly analysis results based on the set confidence level evaluation model, and generate anomaly cause data with different confidence levels. The lifespan extension score calculation unit is used to calculate the comprehensive lifespan extension score based on the marketing data of the electricity meter using a multinomial regression function, and obtain the comprehensive lifespan extension score result. The life extension assessment result generation unit is used to obtain the life extension assessment result based on the abnormal cause data and the comprehensive life extension score result. The life extension assessment result is sent to the energy meter through the configured edge computing gateway to trigger the device self-test program and realize the closed-loop control of assessment and execution.

6. The near-expiration energy meter life extension assessment system based on anomaly assessment according to claim 5, characterized in that, Based on the established confidence level evaluation model, the anomaly analysis results are self-tested, generating anomaly cause data at different confidence levels, including: The abnormal data in the anomaly analysis results are compared with the real-time data collected in the metering automation system to see if the voltage and current in the abnormal data and the real-time data are synchronously abnormal, thus obtaining the first comparison result; the abnormal data in the anomaly analysis results are compared with the user complaint records in the electricity meter marketing data to see if the abnormal data is associated with negative reviews or return data, thus obtaining the second comparison result. Based on the first and second comparison results, the confidence levels are defined as follows: if the comparison results show that all abnormal data are confirmed, the abnormal data is defined as high confidence; if the comparison results show that some abnormal data are confirmed, the abnormal data is defined as low confidence; if the comparison results show that some abnormal data are not confirmed or there are isolated cases, the abnormal data is defined as low confidence. Based on the confidence level, corresponding anomaly cause data is generated.

7. The near-expiration energy meter life extension assessment system based on anomaly assessment according to claim 5, characterized in that, Based on the marketing data of electricity meters, a comprehensive lifespan extension score is calculated using a multinomial regression function to obtain the comprehensive lifespan extension score results, including: Based on the marketing data of electricity meters, extract the number of return records, the number of negative reviews, the sales record ranking value, and the positive review ranking value of electricity meters; The negative feedback value is calculated based on the number of return records and their weight, and the number of negative reviews and their weight. Using sales record ranking value and positive review ranking value as variables, and negative feedback value as a constant, a comprehensive life extension score is calculated using a multinomial regression function. The comprehensive life extension score calculation formula is as follows: G=α*S xs +β*S hp -γ*(δ*N th +e*N cp ) In the above formula, G represents the comprehensive life extension score, α represents the weight of the sales record ranking value, and S xs S represents the sales record ranking value, β represents the weight of the positive review ranking value, and S hp Represents the ranking value for positive reviews, (δ*N) th +ε*N cp ) represents the negative feedback value, γ represents the global weight coefficient of the negative feedback value, δ represents the weight of the number of return records, and N th ε represents the number of return records, ε represents the weight of the number of negative reviews, and N represents the number of negative reviews. cp This represents the number of negative reviews; the weights α, β, γ, δ, and ε all range from 0.5 to 1.

2.

8. The near-term energy meter life extension assessment system based on anomaly assessment according to claim 5, characterized in that, Based on the anomaly cause data and comprehensive life extension score results, life extension assessment results for multiple batches of near-expiration energy meters were obtained, including: Set a first comprehensive life extension score threshold and a second comprehensive life extension score threshold; If the comprehensive life extension score is greater than the first comprehensive life extension score threshold, the near-expiration electricity meter is determined to meet the life extension conditions. If the comprehensive life extension score is less than the first comprehensive life extension score threshold but greater than the second comprehensive life extension score threshold, the near-expiration electricity meter is determined to need to be re-inspected. If the comprehensive life extension score is less than the second comprehensive life extension score threshold, the near-expiration electricity meter is determined to need to be replaced. Based on the determination that the near-expiry electricity meters meet the life extension conditions and the determination that the near-expiry electricity meters need to be replaced, the batch of the near-expiry electricity meters that meet the life extension conditions is set as the life extension batch, and the batch of the near-expiry electricity meters that need to be replaced is set as the replacement batch. Obtain all near-expiration energy meters in the batch to be extended in service life; and obtain all near-expiration energy meters in the batch to be replaced in service life. Based on the set batch quality consistency index, the quality consistency index of the batch to be extended is calculated to obtain a first index value. If the first index value is greater than the set first index threshold, then the entire batch of near-expiration energy meters in the batch to be extended is evaluated as having its entire batch extended. The quality consistency index of the batch to be replaced is calculated to obtain a second index value. If the second index value is greater than the set second index threshold, then the entire batch of near-expiration energy meters in the batch to be replaced is evaluated as having its entire batch replaced. The batch quality consistency index is used to quantitatively evaluate whether the near-expiration energy meters in the same batch are consistent in terms of equipment performance, working status, location distribution characteristics, and the degree of synchronization of performance degradation. The higher the value, the better the batch quality consistency.

9. The near-term energy meter life extension assessment system based on anomaly assessment according to claim 8, characterized in that, It also includes a life extension monitoring module, which is used to monitor the quality of near-expiration energy meters that meet the life extension conditions. If the quality of the energy meter with extended life is detected to decline, the re-verification process is triggered. The life extension monitoring module includes a remaining life prediction model construction unit, a dynamic monitoring mechanism generation unit, and a response determination unit; The remaining lifetime prediction model construction unit is used to construct a remaining lifetime prediction model using the gradient boosting decision tree algorithm to predict the remaining lifetime of near-expiration energy meters that meet the life extension conditions. The input of the remaining lifetime prediction model includes the energy meter's error trend, running time, average ambient temperature and humidity, and historical failure count, and the output is the remaining lifetime prediction value. The dynamic monitoring mechanism generation unit is used to calculate the actual monitoring frequency based on the defined monitoring frequency benchmark value, the calculability rate of the transformer area, and the work order hit rate, and to implement dynamic monitoring based on the actual monitoring frequency. The formula for calculating the actual monitoring frequency is: In the above formula, f represents the actual monitoring frequency, ρ represents the dynamic adjustment coefficient, the calculability rate of the transformer area and the work order hit rate are set; f0 represents the baseline value of the monitoring frequency, and M represents the predicted value of the remaining lifespan; the value of the dynamic adjustment coefficient ρ is determined by fitting historical data: when the calculability rate of the transformer area is greater than 95% and the work order hit rate is greater than 90%, ρ is 1.2; otherwise, ρ is 0.

8. The judgment and response unit is used to make judgments and responses based on the monitoring results obtained from the actual monitoring frequency. Specifically, if the increase in error value in two consecutive monitoring sessions is greater than the set amplitude threshold, or the remaining life prediction value is less than the set time threshold, a re-inspection strategy is generated; if the error value in a single monitoring session is greater than the set error value threshold, an emergency replacement strategy is generated.

10. A method for extending the lifespan of near-expiration energy meters based on anomaly assessment, characterized in that, include: Automated metering equipment is used to collect real-time electricity consumption data of the distribution area and the operating data of the near-expiration electricity meters; the operating data includes error value, voltage, current, power factor, clock deviation and data collection completeness rate; The K-means++ clustering algorithm is used to classify and manage near-expiration energy meters in batches based on the electricity consumption data of the distribution area and the operation data of the near-expiration energy meters, combined with the equipment parameter data of the near-expiration energy meters. Based on the established anomaly detection algorithm, combined with the electricity consumption data of the distribution area and the operating data of the near-expiration electricity meters, anomaly analysis is performed on the near-expiration electricity meters that have been divided into batches to obtain the anomaly analysis results; the anomaly detection algorithm includes a box plot algorithm, DBSCAN clustering algorithm and LSTM-attention mechanism combined model. Based on the anomaly analysis results and combined with the marketing data of the electricity meters, a comprehensive life extension score was calculated using a multinomial regression function. Based on the comprehensive life extension score, the life extension assessment results of multiple batches of near-expiration electricity meters were obtained.

Images