Electric energy meter fault detection method based on multi-sensor data fusion

Abstract

The present application relates to the technical field of data processing, and more particularly to a kind of electric energy meter fault detection method based on multi-sensor data fusion, the method according to the data variation trend of each monitoring index in current period, judge whether electric energy meter occurs data drift anomaly in current period;If electric energy meter occurs data drift anomaly in current period, the historical abnormal data of each monitoring index that occurs data drift anomaly is obtained, according to the performance of each historical abnormal data itself, the historical abnormal data of each monitoring index under the influence of temperature is obtained;Comprehensive temperature and the relationship curve of each monitoring index under the influence of temperature historical abnormal data, and the real-time data performance of each monitoring index in current period, determine the reason of electric energy meter occurs data drift anomaly in current period, and then carry out fault detection to electric energy meter, improve the maintenance efficiency of electric energy meter.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a power meter fault detection method based on multi-sensor data fusion. BACKGROUND

[0002] The power meter is an important device for measuring the user's electricity usage, and its stability and accuracy directly affect the user's electricity bill settlement and power grid management. Power meter failure mainly includes data deviation from the true value caused by hardware aging, data loss caused by communication module abnormalities, etc., mainly manifested as data jitter, unreasonable large or small data, fixed data, data drift (referring to the deviation of data from the true value along a certain trend and continuously increasing, although the short-term monitoring value is not obvious, but it will cause the data to deviate from the original change rule) and other data anomalies, affecting the accuracy of user electricity metering and the identification of circuit anomalies, therefore, it is crucial to monitor the power meter state in real time and identify power meter faults in a timely manner.

[0003] The existing power meter fault detection method mainly sets a fixed threshold to identify the moment of abnormal fluctuation of monitoring data such as current and voltage, and alarms the power meter failure. However, for data drift anomalies, data drift anomalies may be caused by internal component failure of the power meter, or may be caused by the power meter affected by environmental factors such as temperature, and the increase of environmental temperature may cause changes in resistance value in the sampling circuit of the power meter, etc., and further may cause data drift anomalies, such anomalies will return to normal after cooling, and are not real failures of internal components of the power meter, only need to introduce a temperature compensation model or improve the heat dissipation system, therefore, the fixed threshold in the traditional way cannot well distinguish the causes of data drift anomalies, and further affects the maintenance efficiency of the power meter.

[0004] Therefore, how to more accurately identify the real failure of the power meter and further improve the maintenance efficiency of the power meter has become a problem to be solved. SUMMARY

[0005] Therefore, the embodiments of the present application provide a power meter fault detection method based on multi-sensor data fusion to solve the problem of how to more accurately identify the real failure of the power meter and further improve the maintenance efficiency of the power meter.

[0006] The power meter fault detection method based on multi-sensor data fusion provided in the embodiments of the present application comprises the following steps:

[0007] obtaining a real-time data sequence of at least two monitoring indexes of the electric energy meter in a current period, judging whether the electric energy meter has a data drift anomaly in the current period according to a change trend of a residual value of each real-time data in the real-time data sequence of each monitoring index;

[0008] If the electric energy meter has a data drift anomaly in the current period, a data sequence is formed by real-time data of each monitoring index in the current period and historical data before the current period, at least three historical abnormal data of data drift anomaly in the data sequence of each monitoring index are obtained, and at least two historical abnormal data of temperature influence are obtained according to data similarity of each monitoring index at a moment corresponding to each historical abnormal data and repeatability of each historical abnormal data in the data sequence of each monitoring index.

[0009] The temperature at the moment corresponding to each historical abnormal data of temperature influence is obtained, a relationship curve between the temperature and the historical abnormal data of temperature influence of each monitoring index is comprehensively considered, and real-time data performance of each monitoring index in the current period is considered to determine a cause of the data drift anomaly of the electric energy meter in the current period, and the electric energy meter is detected for fault according to the cause of the data drift anomaly of the electric energy meter in the current period.

[0010] Preferably, the judging whether the electric energy meter has a data drift anomaly in the current period according to the change trend of the residual value of each real-time data in the real-time data sequence of each monitoring index comprises:

[0011] For a real-time data sequence of any monitoring index, a residual value of each real-time data in the real-time data sequence of the any monitoring index is obtained, and residual values of all real-time data in the real-time data sequence of the any monitoring index are fitted to obtain a real-time residual fitting curve.

[0012] A root mean square error of the real-time residual fitting curve is obtained, an inverse of a sum between a preset constant and the root mean square error is calculated to obtain a fitting credibility, an average value of slopes of all data points on the real-time residual fitting curve is calculated and recorded as an average slope of the real-time residual fitting curve, and a product between an absolute value of the average slope of the real-time residual fitting curve and the fitting credibility is linearly normalized to obtain an anomaly possibility index of the any monitoring index in the current period.

[0013] The anomaly possibility index of each monitoring index of the electric energy meter in the current period is obtained, and whether the electric energy meter has a data drift anomaly in the current period is judged according to the anomaly possibility index of all monitoring indexes in the current period.

[0014] Preferably, the abnormal possibility index of each monitoring index in the current period is used to determine whether the electric energy meter has data drift anomaly in the current period, including:

[0015] If the abnormal possibility index of any monitoring index in the current period is greater than the preset abnormal possibility index threshold, it is determined that the electric energy meter has data drift anomaly in the current period.

[0016] Preferably, the real-time data of each monitoring index in the current period and the historical data before the current period are combined to form a data sequence, and at least three historical abnormal data of data drift anomaly in the data sequence of each monitoring index are obtained, including:

[0017] For any historical data under any monitoring index, in the data sequence of the any monitoring index, according to the time length of the current period, all data in the period of the any historical data are obtained, and the time point corresponding to the any historical data is the first time point in the period of the any historical data;

[0018] According to the residual value of all data in the period of the any historical data, the abnormal possibility index of the period of the any historical data is obtained, and if the abnormal possibility index of any historical data in the period is greater than the preset abnormal possibility index threshold, the any historical data is taken as historical abnormal data.

[0019] Preferably, the data similarity of each monitoring index at the time point corresponding to each historical abnormal data and the repeatability of each historical abnormal data in the data sequence of each monitoring index are used to obtain at least two temperature-affected historical abnormal data under each monitoring index, including:

[0020] For the data sequence of any monitoring index, the residual fitting curve of all data in the period of each historical data in the data sequence of the any monitoring index is obtained by fitting the residual value of all data in the period of each historical data, and the average value of the slope of all data points on the residual fitting curve in the period of each historical data is calculated to obtain the drift characteristic value of each historical data of the any monitoring index, which is used to represent the drift characteristic value in the period of each historical data of the any monitoring index.

[0021] The drift characteristic value of each historical data under each monitoring index is obtained, the time point corresponding to any historical abnormal data under the any monitoring index is recorded as a target time point, and the correlation index of all monitoring indexes at the target time point is obtained according to the drift characteristic value similarity of each monitoring index at the target time point.

[0022] obtaining a repetition degree of the any historical abnormal data under the any monitoring index according to a similarity between the drift characteristic value of the any historical abnormal data and other historical abnormal data under the any monitoring index;

[0023] linearly normalizing a product between the correlation index and the repetition degree to obtain a temperature influence degree of the any historical abnormal data;

[0024] obtaining a temperature influence degree of each historical abnormal data under each monitoring index, and recording historical abnormal data corresponding to a temperature influence degree greater than a preset temperature influence degree threshold as temperature-affected historical abnormal data in all temperature influence degrees.

[0025] Preferably, the obtaining of the correlation index of all monitoring indexes at the target moment according to the similarity between the drift characteristic values of each monitoring index at the target moment comprises:

[0026] obtaining a drift characteristic value of each monitoring index at the target moment, denoted as a target drift characteristic value, calculating a difference absolute value between the absolute values of each two target drift characteristic values, and calculating an inverse of an accumulation value of all difference absolute values to obtain the correlation index of all monitoring indexes at the target moment.

[0027] Preferably, the obtaining of the repetition degree of the any historical abnormal data under the any monitoring index according to the similarity between the drift characteristic values of the any historical abnormal data and other historical abnormal data under the any monitoring index comprises:

[0028] calculating a difference absolute value between the drift characteristic values of the any historical abnormal data and any other historical abnormal data under the any monitoring index, calculating an inverse of a sum between the difference absolute value and a preset constant to obtain a similarity between the any historical abnormal data and the any other historical abnormal data;

[0029] respectively obtaining a root mean square error of a residual fitting curve of a period in which the any historical abnormal data and the any other historical abnormal data are located, calculating an inverse of a sum of root mean square errors between the any historical abnormal data and the any other historical abnormal data to obtain a similarity confidence between the any historical abnormal data and the any other historical abnormal data;

[0030] calculating a product between the similarity and the similarity confidence to obtain a comprehensive similarity between the any historical abnormal data and the any other historical abnormal data;

[0031] Accumulate the comprehensive similarity between the any historical abnormal data under the any monitoring index and each other historical abnormal data to obtain the repetition degree of the any historical abnormal data under the any monitoring index.

[0032] Preferably, the relationship curve of the comprehensive temperature and the historical abnormal data affected by temperature under each monitoring index, and the real-time data performance of each monitoring index in the current period, determine the reason for the data drift anomaly of the electric energy meter in the current period, including:

[0033] For any monitoring index, according to the temperature of each historical abnormal data affected by temperature under the any monitoring index corresponding to the moment, map the drift characteristic value of all historical abnormal data affected by temperature under the any monitoring index to a two-dimensional coordinate system, the horizontal coordinate of the two-dimensional coordinate system represents temperature, and the vertical coordinate represents the drift characteristic value;

[0034] Curve fitting is performed on the data points in the two-dimensional coordinate system of the any monitoring index to obtain the relationship curve between temperature and the drift characteristic value of historical abnormal data under the any monitoring index;

[0035] Obtain the average temperature of the electric energy meter in the current period, and in the relationship curve corresponding to the any monitoring index, obtain the drift characteristic value corresponding to the average temperature;

[0036] Fitting is performed on the residual values of all data in the real-time data sequence of the any monitoring index in the current period to obtain a real-time residual fitting curve, and the average slope of the real-time residual fitting curve is calculated to obtain the drift characteristic value of the any monitoring index in the current period;

[0037] Calculate the absolute value of the difference between the drift characteristic value of the any monitoring index in the current period and the drift characteristic value corresponding to the average temperature to obtain the compliance degree of the relationship curve corresponding to the any monitoring index;

[0038] Obtain the compliance degree of each monitoring index corresponding to the relationship curve of the electric energy meter, linearly normalize the reciprocal of the cumulative product of all compliance degrees to obtain a temperature interference possibility index of the electric energy meter in the current period, and subtract the temperature interference possibility index from the constant 1 to obtain a fault possibility index of the electric energy meter in the current period;

[0039] According to the fault possibility index of the electric energy meter in the current period, determine the reason for the data drift anomaly of the electric energy meter in the current period.

[0040] Preferably, the determining the reason for the data drift anomaly of the electric energy meter in the current period according to the fault possibility index of the electric energy meter in the current period includes:

[0041] If the fault possibility index is greater than a preset fault possibility index threshold, it is determined that the reason for the data drift anomaly of the electric energy meter in the current period is electric energy meter failure.

[0042] If the fault possibility index is less than or equal to the preset fault possibility index threshold, it is determined that the reason for the data drift anomaly of the electric energy meter in the current period is that the ambient temperature is too high.

[0043] Preferably, the fault detection of the electric energy meter according to the reason for the data drift anomaly of the electric energy meter in the current period comprises:

[0044] If the reason for the data drift anomaly of the electric energy meter in the current period is electric energy meter failure, a fault alarm is given to the electric energy meter.

[0045] Compared with the prior art, the embodiment of the present application has the following beneficial effects:

[0046] The present application obtains real-time data sequences of at least two monitoring indexes of the electric energy meter in the current period, judges whether the data drift anomaly of the electric energy meter in the current period occurs according to the change trend of the residual value of each real-time data in the real-time data sequence of each monitoring index, if the data drift anomaly of the electric energy meter in the current period occurs, the real-time data of each monitoring index in the current period and the historical data before the current period are combined to form a data sequence, at least three historical abnormal data of data drift anomaly in the data sequence of each monitoring index are obtained, at least two temperature-affected historical abnormal data under each monitoring index are obtained according to the data similarity of each monitoring index at the corresponding time of each historical abnormal data and the repeatability of each historical abnormal data in the data sequence of each monitoring index, the temperature at the corresponding time of each temperature-affected historical abnormal data is obtained, the relationship curve between the temperature and the temperature-affected historical abnormal data under each monitoring index is comprehensively considered, and the real-time data performance of each monitoring index in the current period is considered to determine the reason for the data drift anomaly of the electric energy meter in the current period, and the fault detection of the electric energy meter according to the reason for the data drift anomaly of the electric energy meter in the current period is performed. Among them, by identifying the abnormal data of the data drift anomaly of the electric energy meter, combining the performance of the abnormal data and the temperature at the monitoring time, the influence relationship between the temperature and the abnormal data is obtained to distinguish the data drift anomaly data caused by the electric energy meter failure and the data drift anomaly data caused by the temperature influence in the current period, so that the management personnel can take different maintenance measures according to different situations, and the maintenance efficiency of the electric energy meter is improved. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.

[0048] Figure 1 is a method flowchart of a power meter fault detection method based on multi-sensor data fusion provided by the first embodiment of the present application. DETAILED DESCRIPTION

[0049] The embodiments of the present disclosure will be described in detail below, and examples of the embodiments are shown in the drawings. The embodiments described below by referring to the drawings are exemplary and are intended to explain the present disclosure, and should not be understood as a limitation of the present disclosure.

[0050] It should be noted that the terms "first", "second", etc. in the specification of the present disclosure and the above drawings are used to distinguish similar objects, and do not necessarily mean a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The implementation described in the following exemplary embodiments does not represent all implementations consistent with the present disclosure. Rather, they are only examples of devices and methods consistent with some aspects of the present disclosure.

[0051] In order to illustrate the technical solutions of the present application, the following will be described by specific embodiments.

[0052] Referring to Figure 1 is a method flowchart of a power meter fault detection method based on multi-sensor data fusion provided by the first embodiment of the present application, as shown in Figure 1 , the method can include:

[0053] Steps and S101, obtaining real-time data sequence of at least two monitoring indexes of the power meter in the current period, and determining whether the data drift anomaly of the power meter occurs in the current period according to the change trend of the residual value of each real-time data in the real-time data sequence of each monitoring index.

[0054] The main fault of the electric energy meter is that the monitored data deviates from the true value due to the aging of the hardware, thereby affecting the accuracy of user electricity fee measurement and the identification of abnormal circuit conditions. The main manifestations of the electric energy meter fault include severe data jitter, unreasonable large or small data, fixed data value, and data drift. The existing electric energy meter fault detection method mainly sets a fixed threshold to identify the abnormal fluctuation time of the monitored data such as current and voltage, and alarms the electric energy meter fault. However, for data drift anomalies, the monitored data of the electric energy meter will present different degrees of data drift phenomenon with the increase of temperature, which is similar to the data drift phenomenon caused by the aging of the electric energy meter and other fault conditions, thereby reducing the accuracy of the electric energy meter fault detection result and affecting the maintenance efficiency of the electric energy meter.

[0055] Therefore, in the embodiment of the present application, firstly, it is judged whether the electric energy meter has a data drift anomaly in the current period. If it is judged that the electric energy meter has a data drift anomaly in the current period, the cause of the data drift anomaly is further analyzed, and then it is detected whether the electric energy meter has a real fault such as internal component aging according to the cause of the data drift anomaly, so as to improve the maintenance efficiency of the electric energy meter.

[0056] Since the most core monitored data of the electric energy meter is voltage, current and power, in the embodiment of the present application, voltage, current and power are taken as the monitored indicators of the electric energy meter, and the voltage, current and power monitored by the electric energy meter are recorded once per second to obtain the voltage value, current value and power value. Here, the monitored indicators are not limited, and the implementer can set the monitored indicators according to the specific scene.

[0057] Firstly, the 10 minutes before the current time are recorded as the current period, that is, the current time is the last time of the current period, and the recorded values (voltage value, current value and power value) of each monitored indicator (voltage, current and power) in the current period are linearly normalized to obtain the real-time data sequence of each monitored indicator (voltage, current and power) in the current period, so as to analyze the real-time data sequence of each monitored indicator in the current period and judge whether the electric energy meter has a data drift anomaly in the current period. Here, linear normalization is a prior art and will not be described again.

[0058] Under normal circumstances, the monitored data of each monitored indicator of the electric energy meter presents a periodic change rule according to the operation of the user's electrical equipment and will not continuously increase or decrease, so the residual term of the real-time data sequence of each monitored indicator is obtained by using the STL time series decomposition algorithm, that is, the residual value of each data in the real-time data sequence of each monitored indicator is obtained by using the STL time series decomposition algorithm, which is used to reflect the abnormal condition of each monitored indicator of the electric energy meter in the current period. Here, the STL time series decomposition algorithm is a prior art and will not be described again.

[0059] Because in the initial stage of the data drift anomaly, the deviation degree between the monitoring value of each monitoring index and the true value is relatively small, similar to the normal fluctuation of data, and with the increase of time, the deviation degree of data will gradually increase, that is, the residual value of data will gradually deviate from the normal change rule along a certain trend (increase or decrease), therefore, in the embodiment of the present application, whether the current in the current period has a data drift anomaly is judged according to the change trend of the residual value of each real-time data in the real-time data sequence of each monitoring index of the electric energy meter, specifically:

[0060] For the real-time data sequence of any monitoring index, the residual value of each real-time data in the real-time data sequence of the any monitoring index is obtained, and the least square method is used to fit the residual values of all real-time data in the real-time data sequence of the any monitoring index to obtain a real-time residual fitting curve, wherein the least square method is a prior art, which will not be described here;

[0061] The root mean square error of the real-time residual fitting curve is obtained, the reciprocal of the sum between a preset constant and the root mean square error is calculated to obtain a fitting credibility, the average value of the slopes of all data points on the real-time residual fitting curve is calculated and recorded as the average slope of the real-time residual fitting curve, and the product between the absolute value of the average slope of the real-time residual fitting curve and the fitting credibility is linearly normalized to obtain the anomaly possibility index of the any monitoring index in the current period;

[0062] The anomaly possibility index of each monitoring index of the electric energy meter in the current period is obtained, and if the anomaly possibility index of any monitoring index in the current period is greater than a preset anomaly possibility index threshold, it is determined that the electric energy meter has a data drift anomaly in the current period.

[0063] In an embodiment, taking the current as an example, the calculation formula of the anomaly possibility index of the current in the current period is:

[0064]

[0065] Wherein, represents the anomaly possibility index of the current in the current period, represents the average slope of the real-time residual fitting curve, represents the root mean square error of the real-time residual fitting curve, represents a preset constant, which is used to prevent the denominator from being 0, in the embodiment of the present application, is set to This is not limited here, and the implementer can set it according to the specific scene, represents a linear normalization function, represents an absolute value symbol.

[0066] It should be noted that, The smaller, the better the fitting degree of the real-time residual fitting curve corresponding to the real-time data sequence of the current, The higher the credibility, The larger, the more the current data deviates from the normal change rule in the current period, and then The larger, the more likely the monitoring data of the current is to have a data drift anomaly.

[0067] Similarly, the abnormal possibility index of each monitoring index of the electric energy meter in the current period is obtained. According to experimental statistics, the preset abnormal possibility index threshold is set to 0.7, which is not limited here, and the implementer can set it according to the specific scene. If the abnormal possibility index of each monitoring index of the electric energy meter in the current period is less than or equal to 0.7, it is determined that the electric energy meter does not have a data drift anomaly in the current period; if there is any monitoring index in the current period. The abnormal possibility index is greater than 0.7, it is determined that the electric energy meter has a data drift anomaly in the current period, at which time the cause of the data drift anomaly of the electric energy meter in the current period needs to be further analyzed to distinguish between data drift anomalies caused by electric energy meter failure and data drift anomalies caused by temperature effects, so that the management personnel can take different maintenance measures according to different situations to improve the maintenance efficiency of the electric energy meter and ensure the accuracy of user electricity metering and the safety of electricity use.

[0068] Steps and S102, if the electric energy meter has a data drift anomaly in the current period, the real-time data of each monitoring index in the current period and the historical data before the current period are combined to form a data sequence, at least three historical abnormal data of data drift anomaly in the data sequence of each monitoring index is obtained, and at least two historical abnormal data affected by temperature under each monitoring index is obtained according to the data similarity of each monitoring index at the time corresponding to each historical abnormal data and the repeatability of each historical abnormal data in the data sequence of each monitoring index.

[0069] Through step S101, it is determined whether the electric energy meter has a data drift anomaly in the current period. If it is determined that the electric energy meter has a data drift anomaly in the current period, the historical data of each monitoring index of the electric energy meter in the previous week of the current period is obtained, which is not limited here. The implementer can set it according to the specific scene, combine the real-time data of each monitoring index of the electric energy meter in the current period and the historical data to form a data sequence, and obtain at least three historical abnormal data of data drift anomaly in the data sequence of each monitoring index, which is used to analyze the cause of the data drift anomaly of the electric energy meter in the current period according to the historical abnormal data. The specific way to obtain the historical abnormal data of each monitoring index is:

[0070] For any historical data under any monitoring index, in the data sequence of the any monitoring index, according to the time length of the current period, all data in the period where the any historical data is located is obtained, and the time corresponding to the any historical data is the first time in the period where the any historical data is located.

[0071] According to the method for obtaining the abnormal possibility index of the current period of the current in step S101, the abnormal possibility index of the period where the any historical data is located is obtained according to the residual value of all data in the period where the any historical data is located, and if the abnormal possibility index of the period where the any historical data is located is greater than 0.7, the any historical data is taken as historical abnormal data.

[0072] Since the increase of the environmental temperature may cause the change of the resistance value in the sampling circuit of the electric energy meter, and then the data drift anomaly is generated, the data drift anomaly caused by the high-temperature environment is temporary (it will return to normal after cooling) and has regularity, and the real data value can be obtained by a compensation algorithm and the like; and the data drift anomaly caused by the electric energy meter fault cannot be recovered itself and has no obvious regularity. Therefore, in the embodiment of the present application, according to the regularity characteristic of the data drift anomaly caused by the high-temperature environment, at least two historical abnormal data affected by temperature are obtained from the historical abnormal data of each monitoring index, and then the relationship between the temperature and the historical abnormal data affected by temperature is analyzed, so that the data drift anomaly caused by the temperature and the data drift anomaly caused by the electric energy meter fault are effectively distinguished.

[0073] Among them, taking the u-th historical abnormal data of the D-th monitoring index of the electric energy meter as an example, the steps of judging whether the u-th historical abnormal data of the D-th monitoring index is the historical abnormal data affected by temperature are as follows:

[0074] (1) According to the data similarity of each monitoring index at the time corresponding to the u-th historical abnormal data, the correlation index of all monitoring data at the time corresponding to the u-th historical abnormal data is obtained.

[0075] Since the temperature usually affects multiple collection circuits, the data drift anomaly caused by the temperature usually appears in each monitoring index and may occur simultaneously; and the electric energy meter fault has strong randomness, so the data drift anomaly between different monitoring indexes has no strong correlation, so in the embodiment of the present application, taking the u-th historical abnormal data as an example, according to the data similarity of each monitoring index at the time corresponding to the u-th historical abnormal data, the correlation index of all monitoring data at the time corresponding to the u-th historical abnormal data is obtained, so as to preliminarily judge whether the u-th historical abnormal data is the historical abnormal data affected by temperature, and specifically:

[0076] The residual error values of all data in the period in which each historical data of the Dth monitoring index is located are fitted by using the least square method to obtain a residual fitting curve in the period in which each historical data is located, and an average value of the slopes of all data points on the residual fitting curve in the period in which each historical data is located, that is, an average slope of the residual fitting curve corresponding to each historical data, is calculated to obtain a drift characteristic value of each historical data of the Dth monitoring index, which is used to represent the drift characteristic value in the period in which each historical data of the Dth monitoring index is located;

[0077] The time corresponding to the u th historical abnormal data under the D th monitoring index is recorded as a target time, the drift characteristic values of each monitoring index at the target time are obtained and recorded as target drift characteristic values, the absolute values of all target drift characteristic values are pairwise subtracted to obtain the difference absolute values between the absolute values of each two target drift characteristic values, and the reciprocal of the cumulative value of all difference absolute values is calculated to obtain the correlation index of all monitoring indexes at the target time.

[0078] In an embodiment, the calculation formula of the correlation index of all monitoring indexes at the time corresponding to the u th historical abnormal data under the D th monitoring index (target time) is:

[0079]

[0080] Wherein, The correlation index of all monitoring indexes at the time corresponding to the u th historical abnormal data under the D th monitoring index (target time) is represented by D u, since the monitoring indexes of the electric energy meter in the embodiment of the application are current, voltage and power, and The drift characteristic value of the current at the time corresponding to the u th historical abnormal data (target time) is represented by C u, that is, the target drift characteristic value of the current at the target time, The drift characteristic value of the voltage at the time corresponding to the u th historical abnormal data (target time) is represented by V u, that is, the target drift characteristic value of the voltage at the target time, The drift characteristic value of the power at the time corresponding to the u th historical abnormal data (target time) is represented by P u, that is, the target drift characteristic value of the power at the target time, The absolute value symbol is represented by | |.

[0081] It should be noted that since the collection circuits of different monitoring indexes are different, the drift directions may be inconsistent, that is, the positive and negative attributes of the target drift characteristic values of different monitoring indexes at the target time are different, so the data drift degree of each monitoring index at the target time is represented by the absolute value, that is, The data drift degree of the current at the target time is represented by C u, The data drift degree of the voltage at the target time is represented by V u, represents the degree of data drift of power at the target moment, the greater the absolute value, the more the historical data at the target moment deviates from the normal change rule, the greater the degree of drift; the smaller the absolute value of the difference, the smaller, the smaller, the smaller, the stronger the correlation of the drift degree of current, voltage and power at the target moment, and the greater, the more likely the u-th historical abnormal data under the D-th monitoring index is the historical abnormal data affected by temperature.

[0082] (2) According to the similarity of the drift characteristic value between the u-th historical abnormal data under the D-th monitoring index and other historical abnormal data, the repetition degree of the u-th historical abnormal data in the D-th monitoring index is obtained.

[0083] Because the collection circuits of current and voltage are different, the degree of influence of the collected data by temperature is also different, that is, the drift degree of current, voltage and power data under the same temperature may not be completely consistent, so if only the correlation index of all monitoring indexes (i.e. current, voltage and power) is used, the u-th historical abnormal data cannot be completely distinguished as abnormal data affected by temperature or electric energy meter fault abnormal data.

[0084] Considering that the data drift phenomenon caused by temperature has a certain regularity and is easy to repeat, while the data drift phenomenon caused by electric energy meter fault is difficult to reproduce, in the embodiment of the present application, according to the similarity of the drift characteristic value between the u-th historical abnormal data under the D-th monitoring index and other historical abnormal data, the repetition degree of the u-th historical abnormal data under the D-th monitoring index is obtained, the higher the repetition degree, the more regular the abnormality, that is, the u-th historical abnormal data is more likely to be historical abnormal data affected by temperature, specifically:

[0085] The absolute value of the difference between the drift characteristic value of the u-th historical abnormal data under the D-th monitoring index and any other historical abnormal data is calculated, and the reciprocal of the sum of the difference absolute value and a preset constant is calculated to obtain the similarity between the u-th historical abnormal data and the any other historical abnormal data;

[0086] The root mean square error of the residual fitting curve of the period in which the u-th historical abnormal data and the any other historical abnormal data are located is obtained respectively, and the reciprocal of the sum of the root mean square error between the u-th historical abnormal data and the any other historical abnormal data is calculated to obtain the similarity confidence between the u-th historical abnormal data and the any other historical abnormal data;

[0087] calculating a product between the similarity and the similar confidence, to obtain a comprehensive similarity between the u-th historical abnormal data and any other historical abnormal data;

[0088] accumulating the comprehensive similarity between the u-th historical abnormal data under the D-th monitoring index and each other historical abnormal data, to obtain a repetition degree of the u-th historical abnormal data under the D-th monitoring index.

[0089] In an embodiment, a calculation formula of the repetition degree of the u-th historical abnormal data under the D-th monitoring index is as follows:

[0090]

[0091] wherein, represents the repetition degree of the u-th historical abnormal data under the D-th monitoring index, represents a number of all historical abnormal data under the D-th monitoring index, represents a drift eigenvalue of the u-th historical abnormal data under the D-th monitoring index, represents a drift eigenvalue of the v-th other historical abnormal data under the D-th monitoring index, represents a preset constant, used for preventing a denominator from being 0, in an embodiment of the present application, the preset constant is set to 1, which is not limited here, and an implementer can set it according to a specific scene, represents a root mean square error of a residual fitting curve of a time period in which the u-th historical abnormal data under the D-th monitoring index is located, represents a root mean square error of a residual fitting curve of a time period in which the v-th other historical abnormal data under the D-th monitoring index is located, represents an absolute value symbol.

[0092] It should be noted that, the smaller the value is, the smaller the error of the residual fitting curve of the time period in which the u-th historical abnormal data under the D-th monitoring index and the v-th other historical abnormal data are located is, the higher the fitting degree is, and thus the greater the value is, the higher the confidence of the value is; the smaller the value is, the more similar the data drift phenomenon between the two historical abnormal data under the D-th monitoring index is, and thus the greater the value is, the higher the repetition degree of the u-th historical abnormal data under the D-th monitoring index is, and the u-th historical abnormal data under the D-th monitoring index is more likely to be historical abnormal data affected by temperature.

[0093] (3) judging whether the u-th historical abnormal data under the D-th monitoring index is temperature-affected historical abnormal data according to the correlation index of all monitoring indexes at the time corresponding to the u-th historical abnormal data under the D-th monitoring index and the repetition degree of the u-th historical abnormal data under the D-th monitoring index.

[0094] Firstly, the temperature influence degree of the u-th historical abnormal data under the D-th monitoring index is obtained according to the correlation index of all monitoring indexes at the time corresponding to the u-th historical abnormal data under the D-th monitoring index and the repetition degree of the u-th historical abnormal data under the D-th monitoring index, and then whether the u-th historical abnormal data under the D-th monitoring index is temperature-affected historical abnormal data is judged according to the temperature influence degree, specifically:

[0095] The product between the correlation index and the repetition degree is linearly normalized to obtain the temperature influence degree of the u-th historical abnormal data. If the temperature influence degree of the u-th historical abnormal data is greater than a preset temperature influence degree threshold, the u-th historical abnormal data is determined as temperature-affected historical abnormal data.

[0096] In an embodiment, the calculation formula of the temperature influence degree of the u-th historical abnormal data under the D-th monitoring index is as follows:

[0097]

[0098] wherein, represents the temperature influence degree of the u-th historical abnormal data under the D-th monitoring index, represents the correlation index of all monitoring indexes at the time (target time) corresponding to the u-th historical abnormal data under the D-th monitoring index, represents the repetition degree of the u-th historical abnormal data under the D-th monitoring index, represents a linear normalization function.

[0099] It should be noted that, The greater the correlation index is, the stronger the correlation between the drift degrees of the monitoring indexes (current, voltage and power) of the electric energy meter at the time (target time) corresponding to the u-th historical abnormal data is, and the u-th historical abnormal data under the D-th monitoring index is more likely to be temperature-affected historical abnormal data. The greater the repetition degree is, the more similar the data drift phenomenon between the historical abnormal data under the D-th monitoring index is, the higher the repetition degree is, and the more regular the abnormality is, and the u-th historical abnormal data under the D-th monitoring index is more likely to be temperature-affected historical abnormal data. The greater the repetition degree is, the more similar the data drift phenomenon between the historical abnormal data under the D-th monitoring index is, the higher the repetition degree is, and the more regular the abnormality is, and the u-th historical abnormal data under the D-th monitoring index is more likely to be temperature-affected historical abnormal data. The greater the repetition degree is, the more similar the data drift phenomenon between the historical abnormal data under the D-th monitoring index is, the higher the repetition degree is, and the more regular the abnormality is, and the u-th historical abnormal data under the D-th monitoring index is more likely to be temperature-affected historical abnormal data.

[0100] According to the experimental statistics, the preset temperature influence degree threshold is set to 0.5, which is not limited here, and the implementer can set the temperature influence degree threshold according to the specific scene. If , it is determined that the u-th historical abnormal data under the D-th monitoring index is temperature-affected historical abnormal data.

[0101] Similarly, the temperature influence degree of each historical abnormal data under each monitoring index of the electric energy meter is obtained, and then the temperature-affected historical abnormal data under each monitoring index is obtained.

[0102] Steps and S103, the temperature of each temperature-affected historical abnormal data corresponding to the moment is obtained, the relationship curve between the temperature and the temperature-affected historical abnormal data under each monitoring index is comprehensively obtained, and the real-time data performance of each monitoring index in the current period is determined. The reason for the data drift anomaly of the electric energy meter in the current period is determined, and the electric energy meter is detected according to the reason for the data drift anomaly of the electric energy meter in the current period.

[0103] Since the increase of the environmental temperature may cause the resistance value in the sampling circuit of the electric energy meter to change, thereby generating data drift anomaly, the data drift anomaly caused by high temperature environment is temporary (it will return to normal after cooling) and has regularity. Therefore, after obtaining at least two temperature-affected historical abnormal data under each monitoring index through step S102, the temperature of each temperature-affected historical abnormal data corresponding to the moment is obtained, and then the relationship between the temperature and the temperature-affected historical abnormal data is analyzed to obtain the relationship curve between the temperature and the drift characteristic value of the historical abnormal data under each monitoring index of the electric energy meter, which is used to represent the influence law of the temperature on the data drift anomaly. In order to more accurately detect the reason for the data drift anomaly of the electric energy meter in the current period by using the relationship curve combined with the temperature of the electric energy meter in the current period, it is convenient for the management personnel to take different maintenance measures according to different situations, and the maintenance efficiency of the electric energy meter is improved.

[0104] Among them, taking the D-th monitoring index as an example, the specific way of obtaining the relationship curve between the temperature and the drift characteristic value of the historical abnormal data under the D-th monitoring index is as follows:

[0105] According to the temperature of each temperature-affected historical abnormal data corresponding to the moment under the D-th monitoring index, the drift characteristic values of all temperature-affected historical abnormal data under the D-th monitoring index are mapped into a two-dimensional coordinate system, wherein the abscissa of the two-dimensional coordinate system represents the temperature, and the ordinate represents the drift characteristic value.

[0106] Using the least square method, the data points in the two-dimensional coordinate system of the D-th monitoring index are curve-fitted to obtain the relationship curve between the temperature and the drift characteristic value of the historical abnormal data under the D-th monitoring index.

[0107] Similarly, the relationship curve between the temperature and the drift characteristic value of the historical abnormal data of each monitoring index of the electric energy meter is obtained, and is respectively denoted as a "temperature-current drift characteristic value" relationship curve, a "temperature-voltage drift characteristic value" relationship curve, and a "temperature-power drift characteristic value" relationship curve.

[0108] Further, the relationship curve corresponding to each monitoring index of the electric energy meter is used in combination with the temperature of the electric energy meter in the current period to obtain a fault possibility index of the electric energy meter in the current period, so as to determine the cause of the data drift anomaly of the electric energy meter in the current period. Specifically:

[0109] The average temperature of the electric energy meter in the current period is obtained, and the drift characteristic value corresponding to the average temperature is obtained in the relationship curve corresponding to the Dth monitoring index.

[0110] The residual values of all data in the real-time data sequence of the Dth monitoring index in the current period are fitted by using the least square method to obtain a real-time residual fitting curve, the average slope (i.e., the average value of the slopes of all data points on the real-time residual fitting curve) on the real-time residual fitting curve is calculated, and the drift characteristic value of the Dth monitoring index in the current period is obtained.

[0111] The absolute value of the difference between the drift characteristic value of the Dth monitoring index in the current period and the drift characteristic value corresponding to the average temperature is calculated to obtain the compliance degree of the relationship curve corresponding to the Dth monitoring index.

[0112] The compliance degree of the relationship curve corresponding to each monitoring index of the electric energy meter is obtained, the reciprocal of the cumulative multiplication of all compliance degrees is linearly normalized to obtain a temperature interference possibility index of the electric energy meter in the current period, and the temperature interference possibility index is subtracted by a constant 1 to obtain a fault possibility index of the electric energy meter in the current period.

[0113] If the fault possibility index is greater than a preset fault possibility index threshold, it is determined that the cause of the data drift anomaly of the electric energy meter in the current period is the fault of the electric energy meter.

[0114] In an embodiment, the calculation formula of the fault possibility index of the electric energy meter in the current period is:

[0115]

[0116] Wherein, E represents the fault possibility index of the electric energy meter in the current period, represents the drift characteristic value obtained by inputting the average temperature of the electric energy meter in the current period into the "temperature-current drift characteristic value" relationship curve, i.e., the drift characteristic value corresponding to the average temperature in the relationship curve corresponding to the current, represents the average slope of the real-time residual fitting curve corresponding to the current, that is, the drift characteristic value of the current in the current period, represents the drift characteristic value obtained by bringing the average temperature of the electric energy meter in the current period into the "temperature-voltage drift characteristic value" relationship curve, that is, the drift characteristic value corresponding to the average temperature in the relationship curve corresponding to the voltage, represents the average slope of the real-time residual fitting curve corresponding to the voltage, that is, the drift characteristic value of the voltage in the current period, represents the drift characteristic value obtained by bringing the average temperature of the electric energy meter in the current period into the "temperature-power drift characteristic value" relationship curve, that is, the drift characteristic value corresponding to the average temperature in the relationship curve corresponding to the power, represents the average slope of the real-time residual fitting curve corresponding to the power, that is, the drift characteristic value of the power in the current period, represents the absolute value symbol, represents a linear normalization function.

[0117] It should be noted that, that is, the temperature interference possibility index of the electric energy meter in the current period, The smaller, the more the drift characteristic phenomenon of the current in the current period conforms to the influence law of temperature on data drift anomaly, The smaller, the more the drift characteristic phenomenon of the voltage in the current period conforms to the influence law of temperature on data drift anomaly, The smaller, the more the drift characteristic phenomenon of the power in the current period conforms to the influence law of temperature on data drift anomaly, and the larger the temperature interference possibility index, the smaller E, the smaller the possibility that the electric energy meter in the current period has a data drift anomaly due to internal device failure of the electric energy meter, that is, the larger the possibility that the electric energy meter in the current period has a data drift anomaly due to high ambient temperature.

[0118] According to experimental statistics, the preset fault possibility index threshold is set to 0.5, which is not limited here, and the implementer can set it according to the specific scene. If It is determined that the electric energy meter in the current period has a data drift anomaly due to electric energy meter failure, and at this time, fault alarm needs to be performed on the electric energy meter; if It is determined that the electric energy meter in the current period has a data drift anomaly due to high ambient temperature, and at this time, a temperature compensation model needs to be introduced or the electric energy meter heat dissipation system needs to be improved to improve the accuracy of the monitoring data. The introduction of the temperature compensation model or the improvement of the electric energy meter heat dissipation system is prior art, which will not be described here.

[0119] In summary, the present application obtains real-time data sequences of at least two monitoring indexes of the electric energy meter in the current period, judges whether the data drift anomaly of the electric energy meter occurs in the current period according to the change trend of the residual value of each real-time data in the real-time data sequence of each monitoring index, if the data drift anomaly of the electric energy meter occurs in the current period, the real-time data of each monitoring index in the current period and the historical data before the current period are combined to form a data sequence, at least three historical abnormal data of data drift anomaly in the data sequence of each monitoring index are obtained, the data similarity of each monitoring index at the corresponding time of each historical abnormal data is obtained, and the repeatability of each historical abnormal data in the data sequence of each monitoring index is obtained. At least two temperature-affected historical abnormal data under each monitoring index are obtained; the temperature corresponding to the time of each temperature-affected historical abnormal data is obtained, the relationship curve between the temperature and the temperature-affected historical abnormal data under each monitoring index is comprehensively obtained, and the real-time data performance of each monitoring index in the current period is obtained. The cause of the data drift anomaly of the electric energy meter in the current period is determined, and the electric energy meter is detected according to the cause of the data drift anomaly of the electric energy meter in the current period. Wherein, by identifying the abnormal data of the electric energy meter, the influence relationship between the temperature and the abnormal data is obtained by combining the performance of the abnormal data and the temperature at the monitoring time, so as to distinguish the data drift anomaly data caused by the fault of the electric energy meter in the current period and the data drift anomaly data caused by the temperature influence, so as to facilitate the management personnel to take different maintenance measures according to different situations, and improve the maintenance efficiency of the electric energy meter.

[0120] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit it; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced by equivalent; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for detecting faults in an electric energy meter based on multi-sensor data fusion, characterized by, The power meter fault detection method based on multi-sensor data fusion comprises: Obtaining real-time data sequence of at least two monitoring indexes of the power meter in the current period, judging whether the power meter has data drift anomaly in the current period according to the change trend of residual value of each real-time data in the real-time data sequence of each monitoring index; If the power meter has data drift anomaly in the current period, the real-time data of each monitoring index in the current period and the historical data before the current period are combined to form a data sequence, at least three historical abnormal data of data drift anomaly in the data sequence of each monitoring index are obtained, and at least two temperature-affected historical abnormal data under each monitoring index are obtained according to the data similarity of each monitoring index at the time corresponding to each historical abnormal data and the repeatability of each historical abnormal data in the data sequence of each monitoring index; Obtaining the temperature at the time corresponding to each temperature-affected historical abnormal data, comprehensively determining the reason for the data drift anomaly of the power meter in the current period according to the relationship curve between the temperature and the temperature-affected historical abnormal data under each monitoring index and the real-time data performance of each monitoring index in the current period, and performing fault detection on the power meter according to the reason for the data drift anomaly of the power meter in the current period; The method comprises the following steps: For the real-time data sequence of any monitoring index, the residual value of each real-time data in the real-time data sequence of the any monitoring index is obtained, and the residual values of all real-time data in the real-time data sequence of the any monitoring index are fitted to obtain a real-time residual fitting curve; The root mean square error of the real-time residual fitting curve is obtained, the reciprocal of the sum between a preset constant and the root mean square error is calculated to obtain a fitting credibility, the average value of the slope of all data points on the real-time residual fitting curve is calculated and recorded as the average slope of the real-time residual fitting curve, and the product between the absolute value of the average slope of the real-time residual fitting curve and the fitting credibility is linearly normalized to obtain an anomaly possibility index of the any monitoring index in the current period; Obtaining the anomaly possibility index of each monitoring index of the power meter in the current period, and judging whether the power meter has data drift anomaly in the current period according to the anomaly possibility index of all monitoring indexes in the current period; The method comprises the following steps: For any monitoring index, the residual values of all data in the period in which each historical data of the data sequence of the any monitoring index are fitted to obtain a residual fitting curve in the period in which each historical data is located, the average value of the slopes of all data points on the residual fitting curve in the period in which each historical data is located is calculated to obtain the drift characteristic value of each historical data of the any monitoring index, which is used to represent the drift characteristic value in the period in which each historical data of the any monitoring index is located; The drift characteristic value of each historical data under each monitoring index is obtained, the time corresponding to any historical abnormal data under the any monitoring index is recorded as a target time, and the correlation index of all monitoring indexes at the target time is obtained according to the drift characteristic value similarity of each monitoring index at the target time; The repetition degree of the any historical abnormal data under the any monitoring index is obtained according to the drift characteristic value similarity between the any historical abnormal data and other historical abnormal data under the any monitoring index; The product between the correlation index and the repetition degree is linearly normalized to obtain the temperature influence degree of the any historical abnormal data; The temperature influence degree of each historical abnormal data under each monitoring index is obtained, and the historical abnormal data corresponding to the temperature influence degree greater than a preset temperature influence degree threshold is recorded as temperature-affected historical abnormal data in all temperature influence degrees; The relationship curve between the comprehensive temperature and the temperature-affected historical abnormal data under each monitoring index, and the real-time data performance of each monitoring index in the current period, determine the reason for the data drift anomaly of the electric energy meter in the current period, including: For any monitoring index, the drift characteristic values of all temperature-affected historical abnormal data under the any monitoring index are mapped into a two-dimensional coordinate system according to the temperature corresponding to the time of each temperature-affected historical abnormal data under the any monitoring index, the horizontal coordinate of the two-dimensional coordinate system represents the temperature, and the vertical coordinate represents the drift characteristic value; The data points in the two-dimensional coordinate system of the any monitoring index are curve-fitted to obtain the relationship curve between the temperature and the drift characteristic value of the historical abnormal data under the any monitoring index; The average temperature of the electric energy meter in the current period is obtained, and the drift characteristic value corresponding to the average temperature is obtained in the relationship curve corresponding to the any monitoring index; The residual values of all data in the real-time data sequence of the any monitoring index in the current period are fitted to obtain a real-time residual fitting curve, and the average slope of the real-time residual fitting curve is calculated to obtain the drift characteristic value of the any monitoring index in the current period; The absolute value of the difference between the drift characteristic value of the any monitoring index in the current period and the drift characteristic value corresponding to the average temperature is calculated to obtain the compliance degree of the relationship curve corresponding to the any monitoring index. Obtaining the degree of coincidence of each monitoring index of the electric energy meter with the relationship curve, linearly normalizing the reciprocal of the product of all degrees of coincidence, obtaining the temperature interference possibility index of the electric energy meter in the current period, and obtaining the failure possibility index of the electric energy meter in the current period by subtracting the temperature interference possibility index from 1. According to the failure possibility index of the electric energy meter in the current period, the reason for the data drift anomaly of the electric energy meter in the current period is determined.

2. The method for detecting the fault of the electric energy meter based on the multi-sensor data fusion according to claim 1, characterized in that, According to the abnormal possibility index of all monitoring indexes in the current period, whether the electric energy meter has a data drift anomaly in the current period is determined. If the abnormal possibility index of any monitoring index in the current period is greater than the preset abnormal possibility index threshold, it is determined that the electric energy meter has a data drift anomaly in the current period.

3. The method for detecting fault in an electricity meter based on multi-sensor data fusion as claimed in claim 1, wherein, The real-time data and historical data before the current period of each monitoring index in the current period are combined to form a data sequence, and at least three historical abnormal data of data drift anomaly in the data sequence of each monitoring index are obtained. For any historical data under any monitoring index, all data in the period of the any historical data are obtained in the data sequence of the any monitoring index according to the time length of the current period, and the time point corresponding to the any historical data is the first time point in the period of the any historical data. According to the residual value of all data in the period of the any historical data, the abnormal possibility index of the period of the any historical data is obtained, and if the abnormal possibility index of the period of the any historical data is greater than the preset abnormal possibility index threshold, the any historical data is taken as historical abnormal data.

4. The method for detecting fault of electric energy meter based on multi-sensor data fusion according to claim 1, characterized in that, According to the similarity of the drift characteristic value of each monitoring index at the target time, the correlation index of all monitoring indexes at the target time is obtained. The drift characteristic value of each monitoring index at the target time is obtained, denoted as target drift characteristic value, the absolute values of all target drift characteristic values are subtracted from each other to obtain the difference absolute value between the absolute values of each two target drift characteristic values, the reciprocal of the cumulative value of all difference absolute values is calculated to obtain the correlation index of all monitoring indexes at the target time.

5. The method for detecting fault in an electricity meter based on multi-sensor data fusion as claimed in claim 1, wherein, According to the similarity of the drift characteristic value between the any historical abnormal data and other historical abnormal data under the any monitoring index, the repetition degree of the any historical abnormal data under the any monitoring index is obtained. The difference absolute value of the drift characteristic value between the any historical abnormal data and any other historical abnormal data under the any monitoring index is calculated, and the reciprocal of the sum between the difference absolute value and a preset constant is calculated to obtain the similarity between the any historical abnormal data and the any other historical abnormal data. Residual fitting curves of the time period where the any historical abnormal data and the any other historical abnormal data are located are respectively acquired, and the reciprocal of the sum of the root mean square errors between the any historical abnormal data and the any other historical abnormal data is calculated to obtain the similarity confidence between the any historical abnormal data and the any other historical abnormal data; The product between the similarity and the similarity confidence is calculated to obtain the comprehensive similarity between the any historical abnormal data and the any other historical abnormal data; The comprehensive similarities between the any historical abnormal data and each other historical abnormal data under the any monitoring index are accumulated to obtain the repetition degree of the any historical abnormal data under the any monitoring index.

6. The method for detecting fault in an electricity meter based on multi-sensor data fusion as claimed in claim 1, wherein, The method for determining the cause of the data drift anomaly of the electric energy meter in the current period according to the fault possibility index of the electric energy meter in the current period, comprising: If the fault possibility index is greater than the preset fault possibility index threshold, it is determined that the cause of the data drift anomaly of the electric energy meter in the current period is the electric energy meter fault; If the fault possibility index is less than or equal to the preset fault possibility index threshold, it is determined that the cause of the data drift anomaly of the electric energy meter in the current period is the high environmental temperature.

7. The method for detecting faults in an electricity meter based on multi-sensor data fusion as claimed in claim 6, wherein, The method for performing fault detection on the electric energy meter according to the cause of the data drift anomaly of the electric energy meter in the current period, comprising: If the cause of the data drift anomaly of the electric energy meter in the current period is the electric energy meter fault, the electric energy meter is alarmed for fault.

Images