A smart electric energy meter fault data analysis method and system

Abstract

This invention relates to the field of data analysis technology, and in particular to a method and system for analyzing fault data of smart energy meters. The method includes: determining the data state of target fault data based on the estimated anomaly degree and the influence characterization value, and determining the data correction method based on the data state; in the zonal comparison correction, determining the zoning method based on the fault development coefficient and the period difference degree to obtain several divided regions, determining the region state of each divided region based on the number of abnormal factors and the influence stacking degree, and determining the data processing method based on the region state; under preset correction conditions, determining the adjustment method based on the correction comparison coefficient; in the local correction, determining abnormal sub-data based on the fluctuation comparison degree, and performing data correction based on the instability comparison degree; this invention can improve the accuracy of energy meter metering correction.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to a method and system for analyzing fault data of smart energy meters. Background Technology

[0002] As a key device in the power system for measuring electricity consumption, the accuracy of smart meters directly affects the economic interests of power suppliers and users, as well as the stable operation of the power system. However, in actual use, smart meters are prone to measurement errors due to various factors, leading to inaccurate data. Therefore, how to correct faulty data from smart meters to improve the accuracy of metering correction is a technical problem that urgently needs to be solved by those skilled in the art.

[0003] Chinese Patent Publication No. CN117572330A discloses a method and device for bidirectional compensation of electricity meter metering. The method includes: obtaining the electricity consumption error curve of the product during electricity consumption through batch simulation; dividing the metering range of the electricity meter into multiple electricity consumption compensation intervals according to preset electricity consumption segment points; performing quadratic curve fitting on the electricity consumption error curve to obtain compensation data corresponding to different electricity consumption compensation intervals and distributing it; obtaining the power supply error curve of the product during power supply depletion through batch simulation; dividing the metering range of the electricity meter into multiple power supply compensation intervals according to preset power supply segment points; performing quadratic curve fitting on the power supply error curve to obtain compensation data corresponding to different power supply compensation intervals and distributing it; and selecting the compensation data of the corresponding interval for compensation based on the type of sampling current and the compensation interval to which the sampling current belongs. It is evident that the above technical solution has the following problems: the metering compensation method is singular, relying solely on error curve compensation data, and cannot adaptively adjust the compensation method according to the actual data state, resulting in poor accuracy of electricity meter metering compensation. Summary of the Invention

[0004] To address this issue, the present invention provides a method and system for analyzing fault data of smart energy meters, which overcomes the problem that the existing technology has a single metering compensation method, which only compensates for data through error curves and cannot adaptively adjust the correction method according to the actual data status, resulting in poor accuracy of energy meter metering correction.

[0005] To achieve the above objectives, the present invention provides a method for analyzing fault data of smart energy meters, comprising:

[0006] The data status of the target fault data is determined based on the estimated anomaly degree and the impact characterization value, and the data correction method is determined based on the data status. The data correction method is either partition comparison correction or local correction.

[0007] In the partition comparison and correction, the partitioning method is determined based on the fault development coefficient and the period difference to obtain several partitioned regions. The regional status of each partitioned region is determined according to the number of abnormal factors and the degree of influence stacking, and the data processing method is determined according to the regional status.

[0008] The partitioning method is to perform associative partitioning based on a matching threshold or influence relevance, and the data processing method is to perform multi-factor iterative correction or data correction based on sub-anomaly degree;

[0009] Under the preset correction conditions, the adjustment method is determined based on the correction comparison coefficient, which is to adjust the range of correction factors or the number of sub-data.

[0010] In local correction, abnormal sub-data are identified based on fluctuation comparison degree, and data correction is performed based on instability comparison degree.

[0011] Furthermore, if the data status of the target fault data is that the estimated abnormality is greater than or equal to the preset estimated abnormality or the impact characterization value is greater than or equal to the preset impact characterization value, then the data correction method is partition comparison correction.

[0012] Furthermore, if the data status of the target fault data is that the estimated anomaly is less than the preset estimated anomaly and the impact characterization value is less than the preset impact characterization value, then the data correction method is local correction.

[0013] Furthermore, the zoning method is determined based on the fault development coefficient and the periodicity difference, including:

[0014] If the fault development coefficient is greater than or equal to the preset fault development coefficient or the period difference is greater than or equal to the preset period difference, the partitioning method is to perform associated partitioning based on the matching threshold.

[0015] If the fault development coefficient is less than the preset fault development coefficient and the period difference is less than the preset period difference, then the partitioning method is to partition based on the correlation of influence.

[0016] Furthermore, if the region status is such that the number of abnormal factors is greater than the standard number and the degree of influence stacking is greater than or equal to the preset degree of influence stacking, then the data processing method is multi-factor iterative correction.

[0017] In multi-factor iterative correction, the correction method is determined based on the correction dependency and factor interaction degree, including:

[0018] If the correction dependency is greater than or equal to the preset correction dependency or the factor interaction degree is greater than or equal to the preset factor interaction degree, the correction method is serial correction based on the correction response threshold.

[0019] If the correction dependency is less than the preset correction dependency and the factor interaction degree is less than the preset factor interaction degree, then the correction method is to perform parallel correction based on the influence coverage.

[0020] Furthermore, serial correction based on the correction response threshold includes:

[0021] The priority coefficient for correction of each abnormal factor is determined based on the correction response threshold, and the energy consumption readings at each time point are adjusted according to the instability offset of each characteristic factor.

[0022] For a single point in time,

[0023] If the instability offset is greater than the preset instability offset, the energy consumption reading at that time point will be reduced according to the degree of dependence.

[0024] If the instability offset is less than the preset instability offset, the energy consumption reading at that time point will be increased according to the degree of dependence.

[0025] The priority coefficient for correcting a single feature factor is positively correlated with the correction response threshold corresponding to that anomalous factor.

[0026] Furthermore, if the correction comparison coefficient is greater than or equal to the preset correction comparison coefficient, the adjustment method is to increase the range of correction factors.

[0027] The increase in the range of the correction factor is positively correlated with the predicted deviation coefficient.

[0028] Furthermore, if the correction comparison coefficient is less than the preset correction comparison coefficient, the adjustment method is to increase the number of sub-data.

[0029] The increase in the number of sub-data points is positively correlated with the prediction deviation coefficient.

[0030] Furthermore, if the region status is such that the number of abnormal factors is equal to the standard number or the influence stacking degree is less than the preset influence stacking degree, then the data processing method is to perform data correction based on the sub-abnormality degree.

[0031] This invention also provides a smart energy meter fault data analysis system, comprising:

[0032] The correction analysis module is used to determine the data status of the target fault data based on the estimated anomaly degree and the impact characterization value, and to determine the data correction method based on the data status. The data correction method is either partition comparison correction or local correction.

[0033] The comparison and correction module, which is connected to the correction analysis module, is used to determine the partitioning method based on the fault development coefficient and the period difference degree in the partition comparison and correction to obtain several partitioned regions, determine the regional status of each partitioned region according to the number of abnormal factors and the degree of influence stacking, and determine the data processing method according to the regional status.

[0034] The partitioning method is to perform associative partitioning based on a matching threshold or influence relevance, and the data processing method is to perform multi-factor iterative correction or data correction based on sub-anomaly degree;

[0035] The correction optimization module, which is connected to the comparison correction module, is used to determine the adjustment method based on the correction comparison coefficient under preset correction conditions, whether to adjust the range of correction factors or the number of sub-data.

[0036] A local correction module, which is connected to the correction analysis module, is used to determine abnormal sub-data based on fluctuation comparison degree and perform data correction based on instability comparison degree during local correction.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows: In the technical solution of the present invention, the data status of the target fault data is determined according to the estimated anomaly degree and the influence characterization value. The estimated anomaly degree and the influence characterization value effectively reflect the proportion of the abnormal part of the target fault data and the complexity of the influencing factors. Then, different data correction methods are adaptively selected according to the data status, so that the selection of data correction methods is more in line with the actual application scenario. In addition, the reasonable selection of partition comparison correction or local correction can effectively improve data quality and enhance the reliability and stability of the power system.

[0038] Furthermore, this invention effectively reflects the actual changes in target fault data through fault development coefficient and periodicity. Based on these factors, different partitioning methods are adaptively selected, avoiding errors that may arise from fixed partitioning methods. When the fault development coefficient or periodicity is high, partitioning based on matching thresholds can better analyze the dynamic changes and periodic characteristics of the fault. When the fault development coefficient and periodicity are low, partitioning based on influence correlation can partition according to the inherent characteristics of the data, accurately adapting to fault characteristics to optimize the data correction process. This effectively improves the accuracy and reliability of electricity meter data processing, ensuring the stable operation of the power system.

[0039] Furthermore, in this invention, the regional state of each partitioned region is determined based on the number of abnormal factors and the degree of influence stacking. The number of abnormal factors and the degree of influence stacking effectively reflect the complexity of the fault data in each partitioned region and the degree of interaction between the influencing factors. Then, different data processing methods are adaptively selected according to the regional state, which can effectively handle complex fault scenarios and reduce the errors that may be caused by a single correction method.

[0040] Furthermore, this invention effectively reflects the sensitivity of the correction order of each abnormal factor and the correlation and synergistic effect between factors by using correction dependency and factor interaction. Then, different correction methods are adaptively selected according to correction dependency and factor interaction. Serial correction based on correction response threshold can ensure the accuracy and stability of each correction step. Parallel correction based on sub-abnormality can improve correction efficiency and data correction efficiency. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the fault data analysis method for smart energy meters according to the present invention;

[0042] Figure 2 This is a flowchart illustrating how the present invention determines the data correction method based on the data status;

[0043] Figure 3 This is a flowchart illustrating the process of determining the partitioning method based on the fault development coefficient and the periodicity difference in this invention.

[0044] Figure 4 This is a module connection diagram of the intelligent energy meter fault data analysis system of the present invention. Detailed Implementation

[0045] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0046] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0047] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0048] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0049] Please see Figures 1 to 3 As shown, the present invention provides a method for analyzing fault data of smart energy meters, comprising:

[0050] The data status of the target fault data is determined based on the estimated anomaly degree and the impact characterization value, and the data correction method is determined based on the data status. The data correction method is either partition comparison correction or local correction.

[0051] In the partition comparison and correction, the partitioning method is determined based on the fault development coefficient and the period difference to obtain several partitioned regions. The regional status of each partitioned region is determined according to the number of abnormal factors and the degree of influence stacking, and the data processing method is determined according to the regional status.

[0052] The partitioning method is to perform associative partitioning based on a matching threshold or influence relevance, and the data processing method is to perform multi-factor iterative correction or data correction based on sub-anomaly degree;

[0053] Under the preset correction conditions, the adjustment method is determined based on the correction comparison coefficient, which is to adjust the range of correction factors or the number of sub-data.

[0054] In local correction, abnormal sub-data are identified based on fluctuation comparison degree, and data correction is performed based on instability comparison degree.

[0055] The application scenario of this invention is the correction of metering fault data of smart energy meters. This invention includes target fault data and several standard data. The target fault data is the data that needs to be corrected when a metering fault occurs, and the standard data is the data that does not have metering errors. Both the target fault data and the standard data include the energy consumption readings displayed on the dial of the energy meter, which are monitored at 5-second intervals over a period of time, in kWh. The time of data collection is set as a time point, that is, a time point is set every 5 seconds.

[0056] The target fault data and standard data correspond to several target monitoring data. The target monitoring data include, but are not limited to, the temperature, humidity, electromagnetic intensity and harmonic content monitored every 5 seconds within the time period corresponding to the target fault data or standard data. The temperature, humidity and electromagnetic intensity are obtained by temperature sensor, humidity sensor and electromagnetic sensor respectively. The harmonic content is obtained by connecting the harmonic analyzer to the metering circuit where the smart energy meter is located. This is the content that is easy for those skilled in the art to understand, and will not be elaborated in detail.

[0057] This invention includes several historical records, each of which records at least one instance of smart meter fault data correction, including the estimated deviation coefficient, fluctuation comparison degree, estimated anomaly degree, impact characterization value, and fault development coefficient. Each historical record also has a corresponding pass / fail marker, which indicates whether the smart meter fault data correction process meets user requirements. The pass / fail marker can be manually recorded. It is understood that users can determine whether the smart meter fault data correction process meets their requirements based on self-defined indicators. These self-defined indicators can be, but are not limited to, correction time, which will not be elaborated here. The correction time is the time consumed to complete the correction of the target fault data.

[0058] The preset correction condition is that the target fault data is corrected through data processing and the estimated deviation coefficient is greater than the preset estimated deviation coefficient. The estimated deviation coefficient = |adjustment reference value - preset adjustment reference value|. The adjustment reference value is the standard deviation of the adjustment coefficient corresponding to each time point in the target fault data. The preset adjustment reference value is the standard deviation of the adjustment coefficient corresponding to each time point in the historical record that can meet user needs and perform multi-factor iterative correction or data correction based on sub-anomaly degree. The adjustment coefficient corresponding to a single time point = |energy consumption reading after multi-factor iterative correction or data correction based on sub-anomaly degree at this time point -energy consumption reading at this time point before correction| / average value of the sub-anomaly degree of the target monitoring data corresponding to each feature factor at this time point.

[0059] The user can determine the value of the preset estimated deviation coefficient according to the actual application scenario. The greater the user's need to improve the accuracy of data correction, the smaller the value of the preset estimated deviation coefficient. A preset estimated deviation coefficient value is provided. The system detects the historical records of user adjustments to the range of correction factors or the number of sub-data, and records the average value of the estimated deviation coefficients corresponding to the historical records that meet the user's needs as the preset estimated deviation coefficient.

[0060] Specifically, if the data status of the target fault data is that the estimated abnormality is greater than or equal to the preset estimated abnormality or the impact characterization value is greater than or equal to the preset impact characterization value, then the data correction method is partition comparison correction.

[0061] The data state includes a first data state and a second data state. The first data state is when the estimated anomaly is greater than or equal to the preset estimated anomaly or the influence characterization value is greater than or equal to the preset influence characterization value. The second data state is when the estimated anomaly is less than the preset estimated anomaly and the influence characterization value is less than the preset influence characterization value.

[0062] The target fault data is divided into n equal parts to obtain several sub-data. The value of n is positively correlated with the time length of the target fault data, where the time length is the number of time points included in the target fault data.

[0063] Estimated anomaly rate = Number of anomalous sub-data points in the target fault data / Total number of sub-data points in the target fault data;

[0064] Abnormal sub-data is sub-data with a fluctuation comparison degree greater than the preset fluctuation comparison degree. For a single sub-data, the power consumption corresponding to each sub-interval in the sub-data is detected. The time period between any two adjacent time points is recorded as a sub-interval. A single sub-data contains several sub-intervals. The power consumption corresponding to a single sub-interval is the absolute value of the difference between the power consumption readings corresponding to the two time points of the sub-interval.

[0065] The fluctuation comparison degree corresponding to a single sub-data point = the standard deviation of power consumption corresponding to each sub-interval in the sub-data point / the average power consumption corresponding to each sub-interval in the sub-data point;

[0066] The preset fluctuation comparison degree can be determined by the user based on the actual application scenario. The greater the user's need to improve the accuracy of data correction, the smaller the preset fluctuation comparison degree should be. One preset fluctuation comparison degree is provided, which is the average fluctuation comparison degree of each sub-interval corresponding to each target correction data in the historical record that can meet the user's needs. The target correction data is the data obtained after correcting the target fault data.

[0067] The impact characteristic value is the maximum value among the anomaly coefficients of each target monitoring data corresponding to the target fault data;

[0068] The anomaly coefficient of a single target monitoring data is the maximum value among the sub-anomalies corresponding to each time point in the target monitoring data.

[0069] It is understandable that each target monitoring data corresponds to an influencing factor, which includes, but is not limited to, temperature, humidity, electromagnetic intensity, and harmonic content.

[0070] For a single target monitoring data, the target monitoring data is denoted as the analysis target monitoring data, and the impact factor corresponding to the analysis target monitoring data is denoted as the analysis impact factor. The sub-anomaly degree corresponding to a single time point in the target monitoring data is = |the monitoring data value corresponding to that time point in the target monitoring data - the average value of the monitoring data values ​​corresponding to each time point in the standard monitoring data corresponding to the analysis impact factor| / (the standard deviation of the monitoring data values ​​corresponding to each time point in the standard monitoring data corresponding to the analysis impact factor × the average value of the monitoring data values ​​corresponding to each time point in the standard monitoring data corresponding to the analysis impact factor).

[0071] Users can determine the preset values ​​for the estimated anomaly degree and the preset impact characterization value based on the actual application scenario. The smaller the preset values ​​for the estimated anomaly degree and the preset impact characterization value, the greater the user's need for partition comparison and correction. The system provides a preset value for the estimated anomaly degree and the preset impact characterization value, detects historical records of local correction, and records the average value of the estimated anomaly degree corresponding to the historical records that meet the user's needs as the preset estimated anomaly degree, and records the average value of the impact characterization value corresponding to the historical records that meet the user's needs as the preset impact characterization value.

[0072] Specifically, if the data status of the target fault data is that the estimated abnormality is less than the preset estimated abnormality and the impact characterization value is less than the preset impact characterization value, then the data correction method is local correction.

[0073] Specifically, the zoning method is determined based on the fault development coefficient and the periodicity difference, including:

[0074] If the fault development coefficient is greater than or equal to the preset fault development coefficient or the period difference is greater than or equal to the preset period difference, the partitioning method is to perform associated partitioning based on the matching threshold.

[0075] If the fault development coefficient is less than the preset fault development coefficient and the period difference is less than the preset period difference, then the partitioning method is to partition based on the correlation of influence.

[0076] Wherein, the fault development coefficient is the average of the development coefficients corresponding to each sub-interval;

[0077] The development coefficient corresponding to a single subinterval = |the volatility comparison degree corresponding to the subinterval - the volatility comparison degree corresponding to the subinterval adjacent to and preceding the subinterval|. It should be noted that if there is no subinterval preceding a single subinterval, the development coefficient corresponding to the subinterval is 0.

[0078] The periodicity difference is the standard deviation of the volatility comparison degree corresponding to each sub-interval;

[0079] Users can determine the values ​​of the preset fault development coefficient and the preset period difference degree according to the actual application scenario. The smaller the values ​​of the preset fault development coefficient and the preset period difference degree, the greater the user's need to associate partitions according to the matching threshold. The system provides a preset fault development coefficient and preset period difference degree value, detects the historical records of users associating partitions according to the matching threshold, and records the average fault development coefficient corresponding to the historical records that can meet the user's needs as the preset fault development coefficient, and records the average period difference degree corresponding to the historical records that can meet the user's needs as the preset period difference degree.

[0080] The association partitioning is performed based on the matching threshold, including: performing association analysis on each sub-interval; when performing association analysis on a single sub-interval, the sub-interval is recorded as the target sub-interval; the sub-intervals outside the target sub-interval that are not recorded in the partitioned area are recorded as reference sub-intervals; the combination of the reference sub-intervals whose matching threshold with the target sub-interval is greater than the preset matching threshold and the target sub-interval is recorded as a partitioned area; and the association analysis continues to be performed on the sub-intervals that are not recorded in the partitioned area until all sub-intervals are recorded in the partitioned area, at which point the association analysis stops.

[0081] For any two sub-intervals, the matching threshold corresponding to the two sub-intervals = 1 - [1 / (absolute value of the difference between the development coefficients of the two sub-intervals + absolute value of the difference between the fluctuation comparison degree of the two sub-intervals + 1)]. It can be understood that the matching threshold reflects the consistency of fluctuation and abnormality between sub-intervals. By associating and partitioning according to the matching threshold, sub-intervals with similar changing trends can be divided into the same region, which can provide a clearer target for subsequent correction, thereby enhancing the self-consistency of data within the region.

[0082] The correlation partitioning is based on the degree of influence, including: performing correlation analysis on each sub-interval; when performing correlation analysis on a single sub-interval, the sub-interval is recorded as the target sub-interval; the sub-intervals outside the target sub-interval that are not recorded in the partitioned area are recorded as reference sub-intervals; the combination of the reference sub-intervals with an influence correlation greater than the preset influence correlation with the target sub-interval and the target sub-interval is recorded as a partitioned area; and the correlation analysis continues to be performed on the sub-intervals that are not recorded in the partitioned area until all sub-intervals are recorded in the partitioned area, at which point the correlation analysis stops.

[0083] For any two sub-intervals, the correlation between the two sub-intervals is 1 - [1 / (absolute value of the difference between the reference values ​​of the two sub-intervals + 1)]. It can be understood that the correlation reflects the inherent connection between different sub-intervals. By partitioning through this correlation, we can ensure that the sub-intervals in the divided area have a high degree of consistency in features. We can process the data in these areas in a unified manner, reduce the impact of noise, and thus improve the reliability of the data.

[0084] For a single sub-interval, the impact reference value is the average of the sub-impact reference values ​​corresponding to each time point within the single sub-interval, and the sub-impact reference value corresponding to a single time point is the average of the sub-anomaly degree of each target monitoring data at that time point;

[0085] The preset matching threshold and preset impact correlation values ​​can be determined by the user based on the actual application scenario. The greater the user's need to improve the accuracy of data compensation, the smaller the preset matching threshold and preset impact correlation values ​​should be. A preset matching threshold and preset impact correlation value are provided. The system detects the historical records of user-associated partitions based on the matching threshold, and records the average of the reference matching thresholds corresponding to each partition in the historical records that meet the user's needs as the preset matching threshold. The system detects the historical records of user-associated partitions based on the impact correlation, and records the average of the reference impact correlations corresponding to each partition in the historical records that meet the user's needs as the preset impact correlation. The reference matching threshold is the matching threshold corresponding to any two sub-intervals within a single partition, and the reference impact correlation is the impact correlation corresponding to any two sub-intervals within a single partition.

[0086] Specifically, if the region status is that the number of abnormal factors is greater than the standard number and the degree of influence stacking is greater than or equal to the preset degree of influence stacking, then the data processing method is multi-factor iterative correction.

[0087] In multi-factor iterative correction, the correction method is determined based on the correction dependency and factor interaction degree, including:

[0088] If the correction dependency is greater than or equal to the preset correction dependency or the factor interaction degree is greater than or equal to the preset factor interaction degree, the correction method is serial correction based on the correction response threshold.

[0089] If the correction dependency is less than the preset correction dependency and the factor interaction degree is less than the preset factor interaction degree, then the correction method is to perform parallel correction based on the influence coverage.

[0090] The region status includes a first region status and a second region status. The first region status is when the number of abnormal factors is greater than the standard number and the influence stacking degree is greater than or equal to the preset influence stacking degree. The second region status is when the number of abnormal factors is equal to the standard number or the influence stacking degree is less than the preset influence stacking degree.

[0091] The number of outliers corresponding to a single partition is equal to the total number of outliers corresponding to that partition.

[0092] For a single influencing factor in a single partitioned region, the maximum value of the sub-anomaly coefficients of the target monitoring data corresponding to the influencing factor at each time point in the partitioned region is detected and recorded as the maximum sub-anomaly coefficient. If the maximum sub-anomaly coefficient corresponding to the influencing factor is greater than the preset maximum sub-anomaly coefficient, the influencing factor is recorded as an anomalous factor.

[0093] The value of the preset maximal anomaly coefficient can be determined by the user according to the actual application scenario. The greater the user's need to improve the correction accuracy, the smaller the value of the preset maximal anomaly coefficient. One preset maximal anomaly coefficient value is to be the average value of the maximal anomaly coefficients corresponding to each division region in the historical record that can meet the user's needs.

[0094] The standard number is 1, and the influence stacking degree corresponding to a single partition region = the factor interaction degree corresponding to the partition region × the correction dependency degree corresponding to the partition region;

[0095] The preset value of the stacking degree of influence can be determined by the user according to the actual application scenario. The smaller the preset value of the stacking degree of influence, the greater the user's need for multi-factor iterative correction. A preset value of the stacking degree of influence is provided, the user's historical records of multi-factor iterative correction are detected, and the average value of the stacking degree of influence corresponding to the historical records that can meet the user's needs is recorded as the preset stacking degree of influence.

[0096] The corrected dependency is the average of the dependency reference values ​​corresponding to each anomalous factor in a single partitioned region. The dependency reference value corresponding to a single anomalous factor is the maximum value among the sub-dependencies of that anomalous factor and all other anomalous factors.

[0097] The detection process involves serially correcting historical records based on the correction response threshold, and recording these records as reference historical records. For two anomalous factors, designated as the first factor and the second factor, the order in which the two anomalous factors are corrected is determined. The first order is defined as when the first factor is corrected prior to the second factor in the reference historical records, and the second order is defined as when the second factor is corrected prior to the first factor. The number of reference historical records that satisfy the first order is denoted as a1, and the number of reference historical records that satisfy the second order is denoted as a2. The sub-dependency is calculated as the larger of a1 and a2 / (the smaller of a1 and a2 + 1).

[0098] Factor interaction degree is the average of the sub-factor interaction degrees corresponding to each anomalous factor in a single partitioned region, and the sub-factor interaction degree corresponding to a single anomalous factor is the average of the interaction coefficients corresponding to that anomalous factor and all other anomalous factors.

[0099] The formula for calculating the interaction coefficient r between any two outliers is:

[0100] ,

[0101] m represents the number of time points in a single partitioned region. and These are the values ​​of the target monitoring data corresponding to the two abnormal factors at the k-th time point in the divided region. for The average value of the target monitoring data for the corresponding abnormal factor at each time point in the divided region. for The average value of the target monitoring data for the corresponding abnormal factor at each time point in the divided region, k = 1, 2, 3, ..., m;

[0102] The user can determine the values ​​of preset correction dependency and preset factor interaction degree according to the actual application scenario. The smaller the values ​​of preset correction dependency and preset factor interaction degree, the greater the user's need for serial correction based on the correction response threshold. The system provides a preset correction dependency and preset factor interaction degree value, detects the user's historical records of serial correction based on the correction response threshold, and records the average correction dependency degree corresponding to the historical records that meet the user's needs as the preset correction dependency degree, and records the average factor interaction degree corresponding to the historical records that meet the user's needs as the preset factor interaction degree.

[0103] The method for determining the correction response threshold is as follows: for a single abnormal factor, the abnormal factor is recorded as the target abnormal factor, and other abnormal factors other than the target abnormal factor are recorded as reference abnormal factors. The correction response threshold corresponding to the target abnormal factor is the average of the sub-response thresholds corresponding to the target abnormal factor and each reference abnormal factor. The sub-response threshold corresponding to the target abnormal factor and a single reference abnormal factor = the number of reference historical records that the target abnormal factor is preferentially corrected relative to the reference abnormal factor / the total number of reference historical records.

[0104] The method for confirming the impact coverage is as follows: for a single segmented region, each abnormal factor in the segmented region is recorded as a reference factor. The impact coverage of a single reference factor is calculated as: the average value of the sub-anomaly degree of the target monitoring data corresponding to the reference factor at each time point in the segmented region / (the average value of the interaction coefficients of the reference factor and other reference factors + 1). The preset impact coverage value can be determined by the user according to the actual application scenario. The greater the user's need to improve the correction accuracy, the smaller the preset impact coverage value will be. A preset impact coverage value is provided. The historical records of parallel correction based on the impact coverage are detected, and the average value of the impact coverage corresponding to each feature factor in the historical records that can meet the user's needs is recorded as the preset impact coverage.

[0105] In the parallel correction of influence coverage, abnormal factors with influence coverage greater than the preset influence coverage are recorded as characteristic factors. Correction is performed on each characteristic factor at the same time. When correcting a single characteristic factor, the energy consumption readings corresponding to each time point in a single division interval are used as the basis to adjust the energy consumption readings corresponding to each time point according to the instability offset to obtain the initial correction data value of the time point corresponding to the characteristic factor.

[0106] For a single point in time,

[0107] If the instability offset is greater than the preset instability offset, the power consumption reading will be reduced according to the sub-anomaly degree.

[0108] If the instability offset is less than the preset instability offset, the power consumption reading will be increased according to the sub-anomaly degree.

[0109] If the instability offset is equal to the preset instability offset, no adjustment is made;

[0110] If the instability offset is greater than the preset instability offset, the initial correction data value corresponding to a single feature factor = energy consumption reading - decrease in energy consumption reading; if the instability offset is less than the preset instability offset, the initial correction data value corresponding to a single feature factor = energy consumption reading + increase in energy consumption reading; if the instability offset is equal to the preset instability offset, the energy consumption reading remains unchanged.

[0111] When adjusting the energy consumption reading at a time point corresponding to a single characteristic factor, the increase or decrease in the energy consumption reading at that single time point is positively correlated with the absolute value of the instability offset.

[0112] After all feature factors have been corrected simultaneously, the corrected data value at a single time point is equal to the initial corrected data value of each feature factor at that time point divided by the number of feature factors.

[0113] Specifically, serial correction based on the correction response threshold includes:

[0114] The priority coefficient for correction of each abnormal factor is determined based on the correction response threshold, and the energy consumption readings at each time point are adjusted according to the instability offset of each characteristic factor.

[0115] For a single point in time,

[0116] If the instability offset is greater than the preset instability offset, the energy consumption reading at that time point will be reduced according to the degree of dependence.

[0117] If the instability offset is less than the preset instability offset, the energy consumption reading at that time point will be increased according to the degree of dependence.

[0118] The priority coefficient for correcting a single feature factor is positively correlated with the correction response threshold corresponding to that anomalous factor.

[0119] It can be understood that the larger the priority coefficient of a single feature factor, the higher the priority of the correction order for that feature factor.

[0120] When adjusting the energy consumption reading at a time point corresponding to a single characteristic factor, the energy consumption reading is adjusted based on the adjustment data corresponding to that characteristic factor and the degree of instability offset.

[0121] When adjusting the energy consumption reading at a time point corresponding to a single characteristic factor, the increase or decrease in the energy consumption reading at that single time point is positively correlated with the absolute value of the instability offset.

[0122] If the instability offset is equal to the preset instability offset, no adjustment is made;

[0123] It is understandable that if there are q characteristic factors, then the energy consumption readings at each time point will be adjusted q times.

[0124] After the correction of each feature factor is completed, the correction data value corresponding to a single time point is the energy consumption reading obtained after the feature factor with the lowest priority coefficient is corrected.

[0125] The method for confirming the adjustment data is as follows: for a single feature factor, the feature factor is recorded as the first target factor; the feature factor in the reference sequence that is adjacent to the first target factor and located before the first target factor is recorded as the second target factor; the power consumption readings at each time point in a single division interval, after being adjusted according to the dependence influence degree corresponding to the second target factor, are recorded as the adjustment data corresponding to the first target factor. It should be noted that if the first target factor is located at the beginning of the reference sequence, the adjustment data corresponding to the first target factor is the power consumption readings at each time point in a single division region.

[0126] The sequence after sorting each feature factor in descending order of priority coefficient is denoted as the reference sequence;

[0127] The formula for calculating the instability offset w is: , This represents the energy consumption reading at the k-th time point within a single partitioned region. This represents the average of the energy consumption readings at each time point within a single defined region.

[0128] The user can determine the value of the preset instability offset according to the actual application scenario. The larger the value of the preset instability offset, the greater the user's need to increase the adjustment of the power consumption reading. One preset instability offset value is provided, with the preset instability offset being 0.

[0129] The method for confirming the dependency impact is as follows: for a single feature factor, the feature factor is designated as the target feature factor, and the feature factors whose priority coefficient for correction is greater than that of the target feature factor are designated as reference feature factors. The dependency impact at a single time point is calculated as follows: 0.5 × [(average of sub-dependencies of the target feature factor and each reference feature factor / dependency reference value of the target feature factor) + (average of interaction coefficients of the target feature factor and each reference feature factor / interaction coefficient of the sub-factors of the target feature factor)] × sub-anomaly degree of the target monitoring data corresponding to the target feature factor at that time point. It should be noted that if the priority coefficient for correction of a single feature factor is the largest, then the dependency impact of that feature factor is equal to the sub-anomaly degree.

[0130] Specifically, if the correction comparison coefficient is greater than or equal to the preset correction comparison coefficient, the adjustment method is to increase the range of the correction factor.

[0131] The increase in the range of the correction factor is positively correlated with the predicted deviation coefficient.

[0132] For a single segmented region, other abnormal factors besides the characteristic factors corresponding to the segmented region are recorded as non-characteristic factors, and the sub-reference value corresponding to a single abnormal factor is the average value of the sub-abnormality of the target monitoring data corresponding to the abnormal factor at each time point in the segmented region.

[0133] Correction comparison coefficient = average of sub-reference values ​​corresponding to each non-feature factor / average of sub-reference values ​​corresponding to each anomalous factor; it should be noted that if there are no non-feature factors, the correction comparison coefficient is 0.

[0134] The user can determine the value of the preset correction comparison coefficient according to the actual application scenario. The smaller the value of the preset correction comparison coefficient, the greater the user's need to increase the adjustment range of the correction factor. The system provides a preset correction comparison coefficient value, detects the historical records of the user's adjustment of the correction factor range, and records the average value of the correction comparison coefficients corresponding to the historical records that meet the user's needs as the preset correction comparison coefficient.

[0135] The correction factor range includes all abnormal factors to be corrected. When increasing the correction factor range, non-characteristic factors are selected and recorded in the correction factor range in descending order of influence coverage.

[0136] Specifically, if the correction comparison coefficient is less than the preset correction comparison coefficient, the adjustment method is to increase the number of sub-data.

[0137] The increase in the number of sub-data points is positively correlated with the prediction deviation coefficient.

[0138] The number of sub-data is the total amount of sub-data obtained by dividing the target fault data into equal parts.

[0139] Specifically, if the region status is such that the number of abnormal factors is equal to the standard number or the influence stacking degree is less than the preset influence stacking degree, the data processing method is to correct the data based on the sub-abnormality degree.

[0140] In the data correction based on sub-anomaly degree, each anomaly factor is corrected simultaneously. When correcting a single anomaly factor, the energy consumption readings corresponding to each time point in a single division interval are used as the basis to adjust the energy consumption readings corresponding to each time point according to the instability offset to obtain the initial correction data value of the time point corresponding to the anomaly factor.

[0141] For a single time point corresponding to a single anomalous factor,

[0142] If the instability offset is greater than the preset instability offset, the power consumption reading will be reduced based on the sub-anomaly degree corresponding to the target monitoring data of the anomaly factor at that time point.

[0143] If the instability offset is less than the preset instability offset, the power consumption reading will be increased based on the sub-anomaly degree corresponding to the target monitoring data of the anomaly factor at that time point.

[0144] If the instability offset is equal to the preset instability offset, no adjustment is made;

[0145] If the instability offset is greater than the preset instability offset, the initial correction data value corresponding to a single abnormal factor = energy consumption reading - decrease in energy consumption reading; if the instability offset is less than the preset instability offset, the initial correction data value corresponding to a single abnormal factor = energy consumption reading + increase in energy consumption reading; if the instability offset is equal to the preset instability offset, the energy consumption reading remains unchanged.

[0146] After all abnormal factors have been corrected simultaneously, the corrected data value at a single time point is equal to the initial corrected data value of each abnormal factor at that time point divided by the number of abnormal factors.

[0147] Data correction is performed based on the degree of instability comparison, including: correcting the energy consumption readings corresponding to each time point in each anomalous sub-data set.

[0148] When correcting a single time point within a single anomalous subdata set,

[0149] If the instability reference value is greater than the preset instability reference value, the energy consumption reading at that time point will be reduced according to the instability comparison degree.

[0150] If the instability reference value is less than the preset instability reference value, the energy consumption reading at that time point will be increased according to the instability comparison degree.

[0151] If the instability reference value is equal to the preset instability reference value, no adjustment will be made;

[0152] The increase or decrease in the energy consumption reading at a single time point is positively correlated with the instability comparison degree at that time point;

[0153] Each time point in the abnormal sub-data is marked as an abnormal time point, and all other time points are marked as normal time points.

[0154] The extreme value of an anomaly at a single time point is the maximum value of the sub-anomaly degree in the target monitoring data corresponding to each influencing factor at that time point.

[0155] The instability comparison degree corresponding to a single abnormal time point = |abnormal extreme value corresponding to the abnormal time point - average value of abnormal extreme values ​​corresponding to each normal time point| / standard deviation of abnormal extreme values ​​corresponding to each normal time point;

[0156] If the instability reference value is greater than the preset instability reference value, the correction data value at a single time point = energy consumption reading - decrease in energy consumption reading; if the instability reference value is less than the preset instability reference value, the correction data value at a single time point = energy consumption reading + increase in energy consumption reading; if the instability reference value is equal to the preset instability reference value, the energy consumption reading remains unchanged.

[0157] The formula for calculating the instability reference value p is: Where i represents the number of time points contained in a single outlier subdata set, and j represents 1, 2, 3, ..., i. For the j-th time point in a single subdata set, [the value is] the outlier extreme value. This represents the average of the outlier extreme values ​​at each time point within a single subdata set. This represents the energy consumption reading at the j-th time point within a single data set. This represents the average of the energy consumption readings at each time point within a single sub-data set.

[0158] The user can determine the preset instability reference value according to the actual application scenario. The larger the preset instability reference value, the greater the user's need to increase the adjustment of the power consumption reading. One preset instability reference value is provided, and the preset instability reference value is 0.

[0159] Please see Figure 4 The diagram shown is a module connection diagram of the smart energy meter fault data analysis system of the present invention. The present invention also provides a smart energy meter fault data analysis system, comprising:

[0160] The correction analysis module is used to determine the data status of the target fault data based on the estimated anomaly degree and the impact characterization value, and to determine the data correction method based on the data status. The data correction method is either partition comparison correction or local correction.

[0161] The comparison and correction module, which is connected to the correction analysis module, is used to determine the partitioning method based on the fault development coefficient and the period difference degree in the partition comparison and correction to obtain several partitioned regions, determine the regional status of each partitioned region according to the number of abnormal factors and the degree of influence stacking, and determine the data processing method according to the regional status.

[0162] The partitioning method is to perform associative partitioning based on a matching threshold or influence relevance, and the data processing method is to perform multi-factor iterative correction or data correction based on sub-anomaly degree;

[0163] The correction optimization module, which is connected to the comparison correction module, is used to determine the adjustment method based on the correction comparison coefficient under preset correction conditions, whether to adjust the range of correction factors or the number of sub-data.

[0164] A local correction module, which is connected to the correction analysis module, is used to determine abnormal sub-data based on fluctuation comparison degree and perform data correction based on instability comparison degree during local correction.

[0165] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0166] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for analyzing fault data of a smart energy meter, characterized in that, include: The data status of the target fault data is determined based on the estimated anomaly degree and the impact characterization value, and the data correction method is determined based on the data status. The data correction method is either partition comparison correction or local correction. Wherein, the estimated anomaly degree = the number of abnormal sub-data in the target fault data / the total number of sub-data in the target fault data; the target fault data is divided into n equal parts to obtain several sub-data, and the abnormal sub-data is the sub-data with a fluctuation comparison degree greater than the preset fluctuation comparison degree. For a single sub-data, the time period between any two adjacent time points is recorded as a sub-interval. The fluctuation comparison degree corresponding to a single sub-data is = the standard deviation of the power consumption corresponding to each sub-interval in the sub-data / the average power consumption corresponding to each sub-interval in the sub-data. The impact characterization value is the maximum value among the anomaly coefficients of each target monitoring data corresponding to the target fault data; the anomaly coefficient of a single target monitoring data is the maximum value among the sub-anomalies corresponding to each time point in the target monitoring data; the sub-anomaly corresponding to a single time point in a single target monitoring data = |the monitoring data value corresponding to that time point in the target monitoring data - the average value of the monitoring data corresponding to each time point in the standard monitoring data corresponding to the analysis impact factor| / (the standard deviation of the monitoring data values ​​corresponding to each time point in the standard monitoring data corresponding to the analysis impact factor × the average value of the monitoring data values ​​corresponding to each time point in the standard monitoring data corresponding to the analysis impact factor); If the data status of the target fault data is that the estimated abnormality is greater than or equal to the preset estimated abnormality or the impact characterization value is greater than or equal to the preset impact characterization value, then the data correction method is partition comparison correction. In partition comparison correction, the partitioning method is determined based on the fault development coefficient and the periodic difference of the target fault data to obtain several partitioned regions. The region status of each partitioned region is determined according to the number of abnormal factors in the partitioned region and the impact stacking degree of the partitioned region, and the data processing method is determined according to the region status. The partitioning method is to perform association partitioning based on the matching threshold of any two sub-intervals or the influence correlation of any two sub-intervals. The data processing method is to perform multi-factor iterative correction on the partitioned regions or to perform data correction based on the sub-anomaly degree of each time point in the partitioned regions. Wherein, the influence stacking degree corresponding to a single partition region = the factor interaction degree corresponding to that partition region × the correction dependency degree corresponding to that partition region; the factor interaction degree is the average of the sub-factor interaction degrees corresponding to each anomalous factor in a single partition region, and the sub-factor interaction degree corresponding to a single anomalous factor is the average of the interaction coefficients corresponding to that anomalous factor and all other anomalous factors; the formula for calculating the interaction coefficient r between any two anomalous factors is: ， m represents the number of time points in a single partitioned region. and These are the values ​​of the target monitoring data corresponding to the two abnormal factors at the k-th time point in the divided region. for The average value of the target monitoring data for the corresponding abnormal factor at each time point in the divided region. for The average value of the target monitoring data for the corresponding abnormal factor at each time point in the divided region, k = 1, 2, 3, ..., m; The matching threshold between the two sub-intervals = 1 - [1 / (absolute value of the difference between the development coefficients of the two sub-intervals + absolute value of the difference between the fluctuation comparison degree of the two sub-intervals + 1)]; In multi-factor iterative correction, the correction method is determined based on the correction dependency and factor interaction degree, including: If the correction dependency is greater than or equal to the preset correction dependency or the factor interaction degree is greater than or equal to the preset factor interaction degree, the correction method is serial correction based on the correction response threshold. If the correction dependency is less than the preset correction dependency and the factor interaction degree is less than the preset factor interaction degree, then the correction method is to perform parallel correction based on the influence coverage. Wherein, the correction dependency is the average of the dependency reference values ​​corresponding to each abnormal factor in a single partitioned region, and the dependency reference value corresponding to a single abnormal factor is the maximum value among the sub-dependencies corresponding to that abnormal factor and other abnormal factors; the detection performs serial correction according to the correction response threshold and records the historical records that can meet the requirements as reference historical records. For two abnormal factors, the first factor is corrected in the reference historical records with priority over the second factor and is recorded as the first order, and the second factor is corrected in the reference historical records with priority over the first factor and is recorded as the second order. The number of reference historical records that meet the first order is recorded as a1, and the number of reference historical records that meet the second order is recorded as a2. Sub-dependency = the larger value of a1 and a2 / (the smaller value of a1 and a2 + 1). The impact coverage of a single reference factor = the average of the sub-anomalies of the target monitoring data corresponding to the reference factor at each time point in the divided region / (the average of the interaction coefficients between the reference factor and other reference factors + 1). The correction response threshold corresponding to the target anomalous factor is the average of the sub-response thresholds corresponding to the target anomalous factor and each reference anomalous factor. The sub-response threshold corresponding to the target anomalous factor and a single reference anomalous factor = the number of reference historical records for which the target anomalous factor is preferentially corrected relative to that reference anomalous factor / the total number of reference historical records. Under the preset correction conditions, the adjustment method is determined based on the correction comparison coefficient, which is to adjust the range of correction factors or the number of sub-data. Wherein, the correction comparison coefficient = the average value of the sub-reference values ​​corresponding to each non-feature factor / the average value of the sub-reference values ​​corresponding to each abnormal factor; for a single division region, other abnormal factors besides the feature factors corresponding to the division region are recorded as non-feature factors, and the sub-reference value corresponding to a single abnormal factor is the average value of the sub-abnormality of the target monitoring data corresponding to the abnormal factor at each time point in the division region. The range of correction factors includes all abnormal factors that are being corrected. If the data status of the target fault data is that the estimated abnormality is less than the preset estimated abnormality and the impact characterization value is less than the preset impact characterization value, then the data correction method is local correction. In local correction, abnormal sub-data is determined based on the fluctuation comparison degree of sub-data, and data correction is performed based on the instability comparison degree at the time point. Wherein, the instability comparison degree corresponding to a single abnormal time point = |abnormal extreme value corresponding to the abnormal time point - average value of abnormal extreme values ​​corresponding to each normal time point| / standard deviation of abnormal extreme values ​​corresponding to each normal time point.

2. The method for analyzing fault data of smart energy meters according to claim 1, characterized in that, The partitioning method is determined based on the fault development coefficient and the periodicity difference, including: If the fault development coefficient is greater than or equal to the preset fault development coefficient or the period difference is greater than or equal to the preset period difference, the partitioning method is to perform associated partitioning based on the matching threshold. If the fault development coefficient is less than the preset fault development coefficient and the period difference is less than the preset period difference, then the partitioning method is to partition based on the correlation of influence.

3. The method for analyzing fault data of smart energy meters according to claim 2, characterized in that, If the region status is such that the number of abnormal factors is greater than the standard number and the degree of influence stacking is greater than or equal to the preset degree of influence stacking, then the data processing method is multi-factor iterative correction.

4. The method for analyzing fault data of smart energy meters according to claim 3, characterized in that, Serial correction based on the correction response threshold includes: The priority coefficient for correction of each abnormal factor is determined based on the correction response threshold, and the energy consumption readings at each time point are adjusted according to the instability offset of each characteristic factor. For a single point in time, If the instability offset is greater than the preset instability offset, the energy consumption reading at that time point will be reduced according to the degree of dependence. If the instability offset is less than the preset instability offset, the energy consumption reading at that time point will be increased according to the degree of dependence. The priority coefficient for correcting a single feature factor is positively correlated with the correction response threshold corresponding to that anomalous factor; The formula for calculating the instability offset w is: , This represents the energy consumption reading at the k-th time point within a single partitioned region. This represents the average of the energy consumption readings at each time point within a single defined region.

5. The method for analyzing fault data of a smart energy meter according to claim 3, characterized in that, If the correction comparison coefficient is greater than or equal to the preset correction comparison coefficient, the adjustment method is to increase the range of correction factors. The increase in the range of the correction factor is positively correlated with the predicted deviation coefficient.

6. The method for analyzing fault data of a smart energy meter according to claim 5, characterized in that, If the correction comparison coefficient is less than the preset correction comparison coefficient, the adjustment method is to increase the number of sub-data. The increase in the number of sub-data points is positively correlated with the prediction deviation coefficient.

7. The method for analyzing fault data of a smart energy meter according to claim 3, characterized in that, If the region status is such that the number of abnormal factors is equal to the standard number or the impact stacking degree is less than the preset impact stacking degree, the data processing method is to correct the data based on the sub-abnormality degree.

8. An analysis system applying the fault data analysis method for smart energy meters according to any one of claims 1 to 7, characterized in that, include: The correction analysis module is used to determine the data status of the target fault data based on the estimated anomaly degree and the impact characterization value, and to determine the data correction method based on the data status. The data correction method is either partition comparison correction or local correction. The comparison and correction module, which is connected to the correction analysis module, is used to determine the partitioning method based on the fault development coefficient and the period difference degree in the partition comparison and correction to obtain several partitioned regions, determine the regional status of each partitioned region according to the number of abnormal factors and the degree of influence stacking, and determine the data processing method according to the regional status. The partitioning method is to perform associative partitioning based on a matching threshold or influence relevance, and the data processing method is to perform multi-factor iterative correction or data correction based on sub-anomaly degree; The correction optimization module, which is connected to the comparison correction module, is used to determine the adjustment method based on the correction comparison coefficient under preset correction conditions, whether to adjust the range of correction factors or the number of sub-data. A local correction module, which is connected to the correction analysis module, is used to determine abnormal sub-data based on fluctuation comparison degree and perform data correction based on instability comparison degree during local correction.

Images