An efficient method for processing electricity meter monitoring data
By using a multivariate physical rule constraint and confidence-weighted fusion method, the false alarm problem of the CUSUM algorithm in strong fluctuation scenarios is solved, and efficient processing and reliability improvement of electricity meter monitoring data are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANDONG DEYUAN ELECTRICITY TECH CO LTD
- Filing Date
- 2026-05-18
- Publication Date
- 2026-07-17
AI Technical Summary
Traditional CUSUM algorithms are prone to misjudging false residuals generated by changes in normal operating conditions as anomalies in scenarios with strong fluctuations, leading to false alarms and reducing the reliability of power grid monitoring systems.
Consistency analysis is performed by introducing multivariate physical rule constraints, calculating the fluctuation matching degree and the operating condition deviation interference degree, and combining the confidence weighted fusion to form a comprehensive physical residual. The input sequence of the CUSUM algorithm is then used for anomaly monitoring.
It effectively distinguishes between real anomalies and false residuals, improves the reliability of electricity meter monitoring data, reduces false alarm rate, and achieves efficient anomaly signal capture.
Smart Images
Figure CN122218304B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology, and in particular to an efficient method for processing electricity meter monitoring data. Background Technology
[0002] With the intelligent development of power systems, a large number of user-side devices have begun to install smart meters to monitor electricity consumption parameters. As the number of deployed meters continues to increase, the scale of monitoring data generated by the power system every day is enormous. However, some abnormal electricity consumption behaviors or aging deviations of metering chips often manifest as cumulative changes in small deviations over a long period of time. Their abnormal signals are extremely hidden, and long-term small deviations are mixed in with the normal load fluctuations of the power grid, making them extremely difficult to process. False alarms or missed alarms are likely to occur, making it difficult to accurately identify early weak abnormal data.
[0003] To capture hidden anomalies, namely weak signal problems, in massive amounts of electricity meter monitoring data, current monitoring systems typically obtain deviations that violate electrical laws based on physical constraints, and then use the CUSUM algorithm to capture the anomalies. This algorithm effectively amplifies weak abnormal signals by continuously accumulating small physical deviations in the same direction over time.
[0004] In actual electricity meter monitoring scenarios, due to issues such as voltage sampling errors, current transformer proportional errors, and unstable power factor measurements, the data calculated based on physical relationships may have certain deviations. In particular, when the power grid experiences strong fluctuations such as the start-up and shutdown of large equipment, multiple electrical parameters will fluctuate synchronously and drastically, causing temporary deviations in physical relationships and generating a large number of false residuals. The traditional CUSUM algorithm cannot effectively distinguish between local real measurement anomalies and synchronous fluctuation errors caused by sudden changes in overall operating conditions. As a result, the algorithm treats these normal operating condition errors as anomalies and continues to accumulate, leading to extremely serious misjudgments and false alarms, which greatly weakens the reliability of the power grid monitoring system. Summary of the Invention
[0005] In view of this, embodiments of the present invention provide an efficient method for processing electricity meter monitoring data to solve the problem that the traditional CUSUM algorithm is sensitive to parameter errors and is prone to misjudging false residuals generated by changes in normal operating conditions as abnormalities and causing frequent false alarms in strong fluctuation scenarios.
[0006] This invention provides an efficient method for processing electricity meter monitoring data, comprising the following steps: real-time acquisition of various monitoring parameters of the electricity meter; for any physical rule at the current moment: extracting the parameters associated with the physical rule and calculating theoretical values, and calculating physical residuals based on the difference between actual recorded values and theoretical values; calculating the local volatility of each parameter associated with the physical rule, and obtaining the volatility significance of each parameter in conjunction with its calculated historical volatility baseline; calculating the volatility matching degree based on the changes in the volatility significance of all parameters; extracting the physical residual sequence and the volatility significance sequence of each parameter within a second preset time window ending at the current moment, and calculating the p-value. The absolute value of the Wilson correlation coefficient is used as the synchronicity of each parameter; the operating condition deviation interference degree is calculated based on the average of the product of the synchronicity of each parameter and the fluctuation matching degree; the reliability priority is calculated based on the standard deviation of the physical residuals of the physical rules in the week prior to the current time, and the reliability priority is weighted with the operating condition deviation interference degree to obtain the confidence level; the comprehensive physical residual is calculated by weighting the corresponding physical residuals of the confidence levels of all physical rules; the CUSUM algorithm input sequence is constructed using the comprehensive physical residual for anomaly monitoring, and the comparison results of the output positive and negative cumulative statistics with the preset threshold are used to generate early warning signals, thus completing the efficient processing of the electricity meter monitoring data.
[0007] Preferably, the physical rules refer to: active power calculation rules for each phase, power triangular geometric constraint rules, three-phase power balance rules, and electrical energy integration rules.
[0008] Preferably, the calculation of physical residual includes: obtaining the actual recorded values of each parameter associated with the physical rule, substituting them into the physical rule to calculate the theoretical value; calculating the difference between the actual recorded value and the theoretical value, and standardizing the difference, using the result of the standardization as the physical residual of the physical rule.
[0009] Preferably, obtaining the volatility significance of each parameter includes: calculating the variance of each parameter associated with the physical rule at all times within a first preset time window with the current time as the endpoint, as the local volatility of each parameter; obtaining the mean of the local volatility of each parameter as the historical volatility baseline of each parameter; and calculating the volatility significance of each parameter. , In the formula, Indicates the current time's... The first physical rule associated with Local fluctuations of the parameters; Indicates the current time's... The first physical rule associated with Historical volatility baseline of the parameters.
[0010] Preferably, the calculation of the fluctuation matching degree includes: calculating the mean of the fluctuation significance of all parameters associated with the physical rule; calculating the absolute difference between the fluctuation significance of each parameter and the mean, calculating the average of all the absolute differences, and performing a negative natural exponent operation on the average to obtain the fluctuation matching degree of the physical rule.
[0011] Preferably, the deviation of the operating condition from the disturbance satisfies the expression: In the formula, Indicates the current time's... The degree of deviation from the operating conditions of the physical rules; Indicates the current time's... The first physical rule associated with Synchronization of various parameters; Indicates the current time's... The degree of fluctuation matching of the physical rules; , Indicates the first The parameter index values and total number associated with each physical rule.
[0012] Preferably, the calculation of reliability priority includes: calculating the standard deviation of the physical residual sequence of the physical rule for all times within the week prior to the current time, as the physical residual standard deviation of the physical rule; taking the average of the physical residual standard deviations of all physical rules as the system benchmark value; calculating the difference between the system benchmark value and the physical residual standard deviation of the physical rule, and calculating the ratio of the difference to the system benchmark value and normalizing it, and taking the normalization result as the reliability priority of the physical rule.
[0013] Preferably, the confidence level satisfies the expression: In the formula, Indicates the current time's... The confidence level of each physical rule; Indicates the current time's... The reliability priority of each physical rule; Indicates the current time's... The degree of deviation from the operating conditions of the physical rules; This represents the natural exponential function.
[0014] Preferably, the calculation of the comprehensive physical residual includes: calculating the product of the confidence level of each physical rule and its physical residual, and summing all the products to obtain a first summation value; summing the confidence levels of each physical rule to obtain a second summation value; and using the ratio of the first summation value to the second summation value as the comprehensive physical residual at the current moment.
[0015] Preferably, the step of generating an early warning signal by comparing the output positive and negative cumulative statistics with a preset threshold includes: using the CUSUM algorithm to calculate the deviation between the current comprehensive physical residual and the normal baseline, and extracting the absolute values of the positive and negative cumulative statistics in real time, wherein the normal baseline is the mean of the comprehensive physical residual under the condition of no abnormality during the testing phase; if either the absolute value of the positive or negative cumulative statistics is greater than the preset threshold, an early warning signal is generated and the current time is marked as a potential abnormal state; if the absolute values of both the positive and negative cumulative statistics are less than or equal to the preset threshold, it is determined that the system is in a normal working state.
[0016] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:
[0017] This invention firstly overcomes the shortcomings of single physical constraints, which are easily affected by local interference or measurement errors and produce false residuals, by introducing multivariate physical rule constraints for consistency analysis, thereby improving the reliability of the underlying data. Secondly, by calculating the fluctuation matching degree and operating condition deviation interference degree of feature data, it can distinguish between false residuals caused by synchronous changes in operating conditions and real anomalies caused by local electrical path damage, effectively suppressing noise interference in complex fluctuation environments. Furthermore, based on confidence-weighted fusion to form a comprehensive physical residual as the input of the CUSUM algorithm, it enhances the sensitivity to early weak anomaly signals while significantly reducing the false alarm rate caused by oversensitivity, thus achieving efficient processing of electricity meter monitoring data. Attached Figure Description
[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0019] Figure 1 This is a flowchart of an efficient method for processing electricity meter monitoring data provided in Embodiment 1 of the present invention. Detailed Implementation
[0020] Embodiments of this disclosure are described in detail below, with examples of these embodiments illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this disclosure, and should not be construed as limiting it.
[0021] To illustrate the technical solution of the present invention, specific embodiments are described below.
[0022] This invention provides an efficient method for processing electricity meter monitoring data, such as... Figure 1As shown, the method includes the following steps:
[0023] Step S101: Obtain various monitoring parameters of the electricity meter in real time.
[0024] It should be noted that the electricity meter monitors the user's electricity consumption status in real time through the metering module, and obtains various electrical operating parameters to form time series data. Due to communication abnormalities, equipment fluctuations or sampling errors, the raw data inevitably contains missing values, duplicate records or abnormal jump data. In order to ensure the accuracy of subsequent data physical relationship analysis and processing efficiency, the raw status data must be cleaned and standardized to obtain a reliable monitoring data sequence, which will provide basic data support for subsequent calculation of physical residuals.
[0025] Specifically, the electricity meter acquires data including voltage, current, active power, reactive power, apparent power, and power factor for each phase, as well as the system's total active power, total reactive power, total apparent power, and cumulative energy, according to a set sampling period. The completeness of the collected monitoring parameters is checked, and any missing data is supplemented using a linear interpolation algorithm. In this embodiment, the sampling period is once per minute, but implementers can adjust it based on actual needs.
[0026] At this point, data sequences of various monitoring parameters have been obtained.
[0027] Step S102: For any physical rule at the current moment: extract the parameters associated with the physical rule and calculate the theoretical value, and calculate the physical residual based on the difference between the actual recorded value and the theoretical value.
[0028] It should be noted that, in order to achieve efficient identification of early, minute anomalies in electricity meter monitoring data, a solid foundation in physics is essential. The various monitoring parameters collected by the electricity meter do not exist in isolation but follow strict fundamental laws of electricity. When the electricity meter and power grid are operating normally, the data relationships between the parameters remain highly consistent. However, once there are issues such as aging and offsetting of the metering chip, electricity theft and tampering, or sampling channel failures, the constraints of these physical laws are disrupted. Therefore, by introducing electrical physics rules to verify the consistency of the monitoring data and comparing the actual monitoring data with theoretically derived data, the degree to which the current state deviates from the true physical laws can be intuitively assessed.
[0029] Specifically, this embodiment clearly defines and applies four physical rules between the monitored data of the electricity meter, including: First, the active power calculation rule for each phase, that is, the active power of each phase is equal to the product of the voltage, current and power factor of the corresponding phase; Second, the power triangular geometric constraint rule, that is, the square of the total apparent power is equal to the sum of the square of the total active power and the square of the total reactive power; Third, the three-phase power balance rule, that is, the total active power of the system is equal to the sum of the active power of phase A, the active power of phase B and the active power of phase C; Fourth, the electrical energy integration rule, that is, the accumulated electrical energy is equal to the integral of the total active power over time.
[0030] Obtain the parameters associated with any physical rule at the current time, and calculate the theoretical value of any physical rule at the current time using the physical rule.
[0031] For example, taking the first physical rule in this embodiment, namely the active power calculation rule for each phase, as an example for the calculation of phase A: First, extract the parameters associated with the first physical rule at the current moment, that is, obtain the phase A voltage, phase A current, and phase A power factor at the current moment; then, according to the electrical laws of the first physical rule, multiply the above phase A voltage, phase A current, and phase A power factor to obtain the theoretical value of phase A active power at the current moment; and at the same sampling moment, the phase A active power data directly measured and recorded by the energy meter is the actual recorded value of the first physical rule.
[0032] Calculate the physical residual of any physical rule at the current moment based on the difference between the actual recorded value and the theoretical value; the specific calculation formula is as follows:
[0033]
[0034] In the formula, Indicates the current time's... The physical residuals of the physical rules; Indicates the current time's... The actual recorded value of each physical rule; Indicates the current time's... The theoretical value of each physical rule; This represents the standardized function.
[0035] in, Reflects the current moment's first The actual recorded value of the electricity meter deviates from the numerical deviation of a specific electrical physical rule under that rule. The larger the difference, the greater the deviation of the monitoring data at the current moment from the specified rule. The more severe the normality of a physical rule, the more likely there is a measurement offset or abnormal behavior in the local measurement link at the current moment. By standardizing it, the differences in dimensions and value ranges between different physical rules are eliminated, so that the physical residuals generated based on different electrical laws have a unified scale for comparison and fusion.
[0036] At this point, the physical residuals of each physical rule at the current moment are obtained.
[0037] Step S103: Calculate the local volatility of each parameter associated with the physical rule, and obtain the volatility significance of each parameter by combining it with the calculated historical volatility baseline; calculate the volatility matching degree based on the changes in the volatility significance of all parameters.
[0038] It should be noted that in actual power consumption scenarios, drastic load changes can cause instantaneous strong fluctuations in monitoring data such as local grid voltage, current, and power. This transient process can lead to short-term instability in the measurement results of the electricity meter. In order to accurately identify fluctuations caused by sudden changes in actual operating conditions, it is necessary to use the current moment as an anchor point and compare the current local instantaneous data fluctuation characteristics with the historical fluctuation baseline of the equipment during long-term operation. Only when the fluctuation amplitude at the current moment is significantly and abnormally higher than the historical normal level can it be confirmed that the current data is in a strong fluctuation condition, thereby ruling out the possibility of it being a real electrical anomaly.
[0039] Specifically, the variance of any parameter associated with any physical rule at all times within the first preset time window with the current time as the endpoint is obtained and calculated as the local volatility of any parameter associated with any physical rule at the current time; at the same time, the mean of the local volatility of any parameter associated with any physical rule at all times within the week prior to the current time is obtained as the historical volatility baseline of any parameter associated with any physical rule at the current time.
[0040] It should be added that the first preset time window is used to define the size of the time window for calculating local volatility. In actual power systems, transient fluctuations caused by the start-up and shutdown of large industrial equipment or typical impact loads usually take several minutes to more than ten minutes to gradually subside and enter a steady state. Combined with the conventional 1-minute sampling frequency of smart meters, if the time window is too small, only a local slice of the fluctuation can be captured, which cannot reflect the complete transient characteristics. If the first preset time window is too large, too much steady-state smoothing data will be introduced, resulting in the dilution of high-frequency fluctuation characteristics. Therefore, the empirical value range is 5 to 15, and 7 is used in this implementation. Implementers can adjust it according to the duration characteristics of typical impact loads in the actual power grid. Furthermore, the time statistical window for obtaining the historical volatility baseline is selected as one week, which takes into account that the operating conditions of power grid equipment and the electricity load of users have extremely significant periodic evolution characteristics, thereby constructing a highly robust recent steady-state reference system. Implementers can also make corresponding dynamic fine adjustments according to the specific production scheduling cycle of the monitoring object.
[0041] Based on the relative difference between the local volatility and the historical volatility baseline, the volatility significance of any parameter associated with any physical rule at the current moment is calculated; the specific calculation formula is as follows:
[0042]
[0043] In the formula, Indicates the current time's... The first physical rule associated with The significance of fluctuations in various parameters; Indicates the current time's... The first physical rule associated with Local fluctuations of the parameters; Indicates the current time's... The first physical rule associated with Historical volatility baseline of the parameters.
[0044] in, Reflects the current moment's first The first physical rule associated with The relative change in the local volatility of a parameter compared to a historical volatility benchmark; the larger this value, the stronger the change in the local volatility of the parameter at the current moment. The first physical rule associated with This parameter is experiencing drastic fluctuations far exceeding normal levels, meaning that the power grid operating conditions may be undergoing a sudden change at the current moment, and the parameter is in an extremely unstable transient measurement environment.
[0045] It should be further noted that under highly fluctuating operating conditions, multiple parameters of the electricity meter are often affected synchronously and fluctuate simultaneously, causing deviations from the original physical relationships and forming false residuals. Conversely, when a real equipment failure or electricity theft occurs, the anomaly is usually limited to specific measurement links or local parameters and will not cause synchronous fluctuations of all related parameters at the current moment. Therefore, by analyzing the consistency of the fluctuation degree of all related parameters under the same physical law, the source of the current deviation can be effectively distinguished. If the fluctuation significance of all parameters at the current moment is highly consistent, it fully indicates that the deviation is an illusion caused by changes in the overall operating conditions.
[0046] Specifically, the fluctuation matching degree of any physical rule at the current moment is calculated and obtained. The fluctuation matching degree is calculated as follows: the mean of the fluctuation significance of all parameters associated with any physical rule at the current moment is calculated, the absolute difference between the fluctuation significance of any parameter associated with any physical rule at the current moment and the mean is calculated, the average of all absolute differences is calculated, and the negative natural exponent is applied to the average value to obtain the fluctuation matching degree of any physical rule at the current moment.
[0047] The average of all absolute differences reflects the average absolute difference in the intensity of fluctuations of the parameters associated with the same physical rule at the current moment. The smaller the value, the closer the fluctuation amplitudes of the parameters are. When the fluctuation matching degree obtained by performing negative natural exponent operation is closer to 1, it indicates that the electrical parameters associated with the physical rule are experiencing synchronous fluctuations with extremely consistent amplitudes at the current moment. This means that the root cause of the destruction of physical consistency at the current moment may be due to a sudden change in the overall operating conditions of the power grid, providing a core basis for subsequent identification and elimination of such false residuals.
[0048] At this point, the fluctuation matching degree of any physical rule at the current moment has been obtained.
[0049] Step S104: Extract the physical residual sequence and the fluctuation significance sequence of each parameter for all times within the second preset time window with the current time as the endpoint; calculate the synchronicity of each parameter; and obtain the operating condition deviation interference degree by combining the fluctuation matching degree; calculate the reliability priority based on the standard deviation of the physical residual of the physical rule within one week before the current time; and weight the reliability priority with the operating condition deviation interference degree to obtain the confidence level.
[0050] It should be noted that when the power grid's operating conditions change, if the physical residual of a certain physical rule and the fluctuation significance show a highly synchronized trend in recent time, that is, as the operating conditions fluctuate drastically, the parameter fluctuation significance increases and the residual also increases in the same direction, it fully demonstrates that the physical rule has been severely disturbed by the operating conditions at the current stage. In order to obtain the above-mentioned synchronization relationship, it is necessary to extract the features of the recent time period at the current moment to construct a time series, and then conduct trend correlation analysis.
[0051] Specifically, the physical residuals of any physical rule at all times within a second preset time window ending at the current time are extracted, and a physical residual sequence is constructed in chronological order. Simultaneously, the fluctuation significance of any parameter associated with any physical rule at all times within the second preset time window is extracted, and a fluctuation significance sequence of any parameter associated with any physical rule at the current time is constructed. The absolute value of the Pearson correlation coefficient between the physical residual sequence and the fluctuation significance sequence of any parameter associated with any physical rule at the current time is calculated, and this value is used as the synchronicity of any parameter associated with any physical rule at the current time.
[0052] It should be added that the second preset time window is used to define the local statistical sample size when constructing the sequence. In actual power grid operation monitoring, to confirm whether there is a real synchronous physical correlation between the evolution trend of the physical residual sequence and the power grid fluctuation significance sequence, it is necessary to rely on a data sequence with a sufficiently long time span. From the perspective of mathematical statistics, if the sequence sample points are too few, they are easily affected by individual random noise points, resulting in a high amount of occasional spurious correlation. On the other hand, if the sample points are too many, they may cross the single transient operating cycle of the power grid, thereby introducing interference from other steady-state intervals. Therefore, the empirical value range is 10 to 30. In this implementation, 15 is used to ensure that when performing correlation mathematical statistics, there is a continuous 15-minute data span with the current time as the endpoint to support the lower limit of confidence in the statistical sense. The implementers can adjust it according to the response delay and statistical confidence requirements of the system.
[0053] The operating condition deviation interference degree of any physical rule at the current moment is calculated by summing the products of the synchronicity of any parameter associated with any physical rule at the current moment and the fluctuation matching degree of any physical rule at the current moment; the specific calculation formula is as follows:
[0054]
[0055] In the formula, Indicates the current time's... The degree of deviation from the operating conditions of the physical rules; Indicates the current time's... The first physical rule associated with Synchronization of various parameters; Indicates the current time's... The degree of fluctuation matching of the physical rules; , Indicates the first The parameter index values and total number associated with each physical rule.
[0056] in, Reflects the current moment's first The increase in the trend of the physical residual sequence of the physical rule is the degree of confidence caused by the synchronous fluctuation of the associated parameters due to the operating condition. The larger the value, the higher the degree of interference of the operating condition deviation on the physical rule at the current stage. The deviation generated is likely to be a dynamic false deviation caused by strong operating condition fluctuation.
[0057] Furthermore, to evaluate the usability of a physical rule in a complex environment, in addition to dynamically assessing the deviation of the current operating condition from the disturbance, it is also necessary to consider the inherent physical stability of the rule at the underlying level. If a physical rule's theoretical calculation value and actual value are always highly consistent during historical steady-state operation, that is, the fluctuation of the physical residual is extremely small, it indicates that the rule is not easily affected by daily measurement noise and its inherent reliability is extremely high.
[0058] Specifically, the physical residual sequence of any physical rule is extracted from all times within the week prior to the current time, and the standard deviation of the sequence is calculated as the physical residual standard deviation of any physical rule. The average of the physical residual standard deviations of all physical rules is used as the system benchmark value. At the same time, the difference between the system benchmark value and the physical residual standard deviation of any physical rule at the current time is calculated, and the ratio of the difference to the system benchmark value is further calculated. The ratio is then subjected to max-min normalization, and the normalization result is used as the reliability priority of any physical rule at the current time.
[0059] Among them, the standard deviation of the physical residual of any physical rule reflects the underlying noise and inherent fluctuation amplitude of this rule in recent operation; while the system benchmark value is the macroscopic expectation of the error fluctuation of all physical rules in the entire monitoring system, representing the average measurement error baseline of the current equipment as a whole. By calculating the difference and ratio between the system benchmark value and the standard deviation of the physical residual of each physical rule, the relative static stability advantage of each physical rule compared with the overall average error level of the power grid system is reflected. The higher the reliability priority obtained, the smaller the standard deviation of the physical residual of the physical rule is compared with the system average level. This means that the physical rule has extremely high stability and is not easily affected by strong fluctuation conditions. Therefore, its underlying reliability is higher and it is not easily affected by daily random measurement noise.
[0060] It should be further noted that, when evaluating the usability of a physical rule in the current complex environment, both static and dynamic characteristics must be taken into account. Only when a physical rule itself has high historical stability and is subject to relatively small deviations from operating conditions can it be considered that the physical rule at the current moment can truly reflect the underlying electrical relationships, and the confidence level of the physical residuals calculated based on the physical rule is relatively higher.
[0061] Specifically, based on the reliability priority and operational condition deviation disturbance degree of any physical rule at the current moment, the confidence level of any physical rule at the current moment is calculated; the specific calculation formula is as follows:
[0062]
[0063] In the formula, Indicates the current time's... The confidence level of each physical rule; Indicates the current time's... The reliability priority of each physical rule; Indicates the current time's... The degree of deviation from the operating conditions of the physical rules; This represents the natural exponential function.
[0064] in, The first reflecting the current moment The inherent reliability benchmark established by the underlying physical logic of the physical rules; the current moment's first... The deviation of the operating condition of the physical rule from the disturbance is mapped to a dynamic penalty weight; if the current time is the first When the physical residual of a physical rule increases in the same direction as the power grid operating conditions fluctuate drastically, that is, the greater the deviation of the operating conditions from the interference, it indicates that there may be a transient impact or the same frequency drift of the measuring equipment in the current power grid, rather than a local electrical fault, which is prone to frequent false alarms. Therefore, the greater the deviation of the operating conditions from the interference, the faster the dynamic penalty weight will approach 0, thereby suppressing the reliability priority of the physical rule and effectively shielding the false physical residual caused by sudden environmental changes. Conversely, the smaller the deviation of the operating conditions from the interference, the better the physical rule is in terms of anti-interference ability. At this time, the dynamic penalty weight will approach 1, the less the confidence of the physical rule is penalized, the more its reliability priority is retained, and the more ensure that real abnormal signals can occupy the absolute dominant weight in the subsequent comprehensive evaluation.
[0065] At this point, the confidence level of each physical law at the current moment has been obtained.
[0066] Step S105: Calculate the comprehensive physical residual by weighted averaging of the corresponding physical residuals based on the confidence levels of all physical rules; construct the input sequence of the CUSUM algorithm using the comprehensive physical residual for anomaly monitoring, and generate an early warning signal by comparing the output positive and negative cumulative statistics with preset thresholds.
[0067] It should be noted that in complex power grid environments, different physical constraints are significantly affected by sudden changes in operating conditions and measurement errors. If a simple arithmetic average is used to fuse multiple physical rules, it is easy to cause confusion between the true deviation with high confidence and the noise artifact features with low confidence. Therefore, weighted fusion of multiple physical residuals based on confidence can adaptively amplify the true anomaly features and suppress environmental noise artifacts, providing a highly robust data source for subsequent anomaly detection.
[0068] Specifically, the calculation of the comprehensive physical residual of the multi-physical rules at the current moment includes: calculating the product of the confidence level and physical residual of any physical rule at the current moment, and summing the products of the confidence levels and physical residuals of all physical rules at the current moment to obtain a first summation value; summing the confidence levels of all physical rules at the current moment to obtain a second summation value; and using the ratio of the first summation value to the second summation value as the comprehensive physical residual at the current moment.
[0069] Among them, the higher the confidence level of each physical rule at the current moment, the higher the stability of all physical rules under the current moment, and the less interference from measurement errors or fluctuations in operating conditions. The comprehensive physical residual obtained at this time is more reliable and more likely to reflect the real abnormal changes, rather than false deviations caused by temporary fluctuations or noise. Therefore, the larger the comprehensive physical residual, the more likely that the current electricity meter system may have really experienced underlying substantive abnormalities such as metering chip aging or electricity theft tampering.
[0070] Furthermore, the input sequence of the CUSUM algorithm is constructed using the comprehensive physical residual. Considering that actual underlying electrical anomalies may exhibit two diametrically opposed physical trends—positive surge or negative decay—during continuous monitoring time steps, this accumulation process requires an independent two-way monitoring mechanism. Specifically, after calculating the deviation between the current comprehensive physical residual and the normal baseline, the CUSUM algorithm inputs the deviation into independent update logics for the positive and negative cumulative statistics: the positive update logic only accumulates positive deviations exceeding the normal baseline and filters out negative fluctuations; the negative update logic only accumulates negative deviations below the normal baseline and filters out positive fluctuations; and the absolute values of the positive and negative cumulative statistics are extracted in real time and compared with preset thresholds respectively.
[0071] If the absolute value of either the positive or negative cumulative statistics is greater than the preset threshold, the anomaly detection module is triggered, generating an early warning signal and marking the current time as a potential abnormal state, thereby completing the efficient processing of the electricity meter monitoring data; conversely, if the absolute values of both the positive and negative cumulative statistics are less than or equal to the preset threshold, it is determined that the current state is normal, the system does not trigger an early warning action, and smoothly enters the cyclic monitoring of the next time step.
[0072] It should be added that the normal baseline is the mean of the comprehensive physical residual sequence of the electricity meter under normal testing conditions. Simultaneously, the standard deviation of the comprehensive physical residual sequence of the electricity meter under normal testing conditions is calculated. Based on the confidence interval theory of normal distribution, the preset threshold is typically set between 3 and 5 times the standard deviation. In this embodiment, 4 times the standard deviation is preferably used as the preset threshold. Combined with the aforementioned two-way monitoring mechanism, this setting essentially constructs a two-way safety monitoring interval to ensure that while filtering out conventional measurement noise and accumulated interference from operating condition spikes to a great extent, it can also intercept real, continuous, small offset signals. Implementers can dynamically calibrate this based on the actual tolerance index for risk identification at the power grid site.
[0073] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.
Claims
1. A method for efficiently processing electricity meter monitoring data, characterized in that, include: Real-time acquisition of various monitoring parameters from the electricity meter; For any physical rule at the current moment, where the physical rule refers to: active power calculation rule for each phase, power triangular geometric constraint rule, three-phase power balance rule, and electrical energy integration rule, the parameters associated with the physical rule are extracted and theoretical values are calculated. The physical residual is calculated based on the difference between the actual recorded value and the theoretical value. The local volatility of each parameter associated with the physical rule is calculated, and the volatility significance of each parameter is obtained by combining it with the calculated historical volatility baseline. The mean of the volatility significance of all parameters associated with the physical rule is calculated. The absolute difference between the volatility significance of each parameter and the mean is calculated, and all... The average of the absolute differences is used, and a negative natural exponent is applied to the average to obtain the fluctuation matching degree of the physical rule; the physical residual sequence and the fluctuation significance sequence of each parameter are extracted for all times within the second preset time window ending at the current time, and the absolute value of the Pearson correlation coefficient is calculated as the synchronicity of each parameter; the operating condition deviation interference degree is calculated based on the average of the product of the synchronicity of each parameter and the fluctuation matching degree; the reliability priority is calculated based on the standard deviation of the physical residuals of the physical rule in the week prior to the current time, and the reliability priority is weighted with the operating condition deviation interference degree to obtain the confidence level; The comprehensive physical residual is calculated by weighting the corresponding physical residuals with the confidence levels of all physical rules. Anomaly monitoring is performed by constructing the input sequence of the CUSUM algorithm using comprehensive physical residuals. The positive and negative cumulative statistics output are compared with preset thresholds to generate early warning signals, thus completing the efficient processing of electricity meter monitoring data.
2. The efficient processing method for electricity meter monitoring data according to claim 1, characterized in that, The calculation of the physical residual includes: Obtain the actual recorded values of each parameter associated with the physical rule, substitute them into the physical rule to calculate the theoretical value; calculate the difference between the actual recorded value and the theoretical value, and standardize the difference, using the standardized result as the physical residual of the physical rule.
3. The efficient processing method for electricity meter monitoring data according to claim 1, characterized in that, The acquisition of the significance of fluctuations in each parameter includes: Calculate the variance of each parameter associated with the physical rule for all moments within a first preset time window, with the current moment as the endpoint, as the local volatility of each parameter; obtain the mean of the local volatility of each parameter as the historical volatility baseline of each parameter; calculate the volatility significance of each parameter. , In the formula, Indicates the current time's... The first physical rule associated with Local fluctuations of the parameters; Indicates the current time's... The first physical rule associated with Historical volatility baseline of the parameters.
4. The efficient processing method for electricity meter monitoring data according to claim 1, characterized in that, The deviation of the operating condition from the disturbance degree satisfies the expression: ； In the formula, Indicates the current time's... The degree of deviation from the operating conditions of the physical rules; Indicates the current time's... The first physical rule associated with Synchronization of various parameters; Indicates the current time's... The degree of fluctuation matching of the physical rules; , Indicates the first The parameter index values and total number associated with each physical rule.
5. The efficient processing method for electricity meter monitoring data according to claim 1, characterized in that, The calculation of reliability priority includes: Calculate the standard deviation of the physical residual sequence of the physical rule for all times within the week prior to the current time, and use it as the physical residual standard deviation of the physical rule; take the average of the physical residual standard deviations of all physical rules as the system benchmark value; calculate the difference between the system benchmark value and the physical residual standard deviation of the physical rule, calculate the ratio of the difference to the system benchmark value and normalize it, and use the normalization result as the reliability priority of the physical rule.
6. The efficient processing method for electricity meter monitoring data according to claim 1, characterized in that, The confidence level satisfies the expression: ； In the formula, Indicates the current time's... The confidence level of each physical rule; Indicates the current time's... The reliability priority of each physical rule; Indicates the current time's... The degree of deviation from the operating conditions of the physical rules; This represents the natural exponential function.
7. The efficient processing method for electricity meter monitoring data according to claim 1, characterized in that, The calculation of the comprehensive physical residual includes: Calculate the product of the confidence level and the physical residual for each physical rule, and sum all products to obtain the first result. Summation value; sum the confidence values of all physical rules to obtain a second summation value; use the ratio of the first summation value to the second summation value as the comprehensive physical residual at the current moment.
8. The efficient processing method for electricity meter monitoring data according to claim 1, characterized in that, The step of generating an early warning signal by comparing the output positive and negative cumulative statistics with a preset threshold includes: The CUSUM algorithm calculates the deviation between the current physical residual and the normal baseline, and extracts the absolute values of the positive and negative cumulative statistics in real time. The normal baseline is the mean of the physical residual under the condition of no abnormality during the testing phase. If either the absolute value of the positive or negative cumulative statistics is greater than a preset threshold, an early warning signal is generated and the current time is marked as a potential abnormal state. If the absolute values of both the positive and negative cumulative statistics are less than or equal to the preset threshold, it is determined that the system is in normal working condition.