A method for assessing the status of smart energy metering boxes based on time-series data

Abstract

This invention relates to the field of electrical variable measurement technology, specifically to a method for assessing the status of intelligent energy metering boxes based on time-series data. The method includes: acquiring multi-parameter time-series data and ambient temperature data from the intelligent energy metering box, establishing historical baseline files for each parameter; extracting real abnormal components through multi-parameter correlation time-series analysis; identifying the current load scenario based on the historical baseline files and calculating abnormal deviations; providing fault warnings based on the variation patterns of multiple parameters in adjacent time periods and abnormal deviations; and outputting the status assessment results and warning information. This invention constructs a three-layer progressive assessment framework—correlation interference elimination, scenario identification and adaptation, and time-series change pattern warning—extending the metering box status assessment from a single-point judgment with a fixed threshold to a progressive assessment closed loop that is adaptive across all scenarios, capable of removing interference, and providing fault warnings. This improves the adaptability and reliability of status assessment in complex power consumption environments.

Description

Technical Field

[0001] This invention relates to the field of electrical variable measurement technology, and more specifically to a method for assessing the status of intelligent power metering boxes based on time-series data. Background Technology

[0002] As the terminal device for electricity metering, the operating status of intelligent electricity metering boxes directly affects the accuracy of electricity metering and electricity safety. During the operation of the metering box, the time-series data of multiple parameters such as voltage, current, temperature, and harmonics contain rich equipment status information. By analyzing the above time-series data, abnormal states of the metering box can be effectively identified. However, in actual power consumption environments, load condition changes constitute the main source of interference in metering box status identification. Current surges caused by the start-up and shutdown of high-power equipment can cause short-term spikes in voltage and temperature; nonlinear loads can cause harmonic distortion of the current waveform; and three-phase unbalanced loads can increase the neutral line current and cause local heating. These parameter fluctuations caused by load condition changes are highly similar to real faults such as poor contact and insulation degradation in their external manifestations in the time-series data. In addition, load patterns vary significantly in different time periods and seasons, and the range of parameter fluctuations under normal conditions varies, further increasing the complexity of status assessment.

[0003] Existing metering box status assessment methods mostly rely on fixed thresholds or single parameters for judgment, lacking the ability to adapt to the differences in time-series data characteristics under different load scenarios. These methods struggle to distinguish between normal parameter fluctuations caused by load changes and abnormal changes caused by equipment malfunctions by leveraging the correlation between multiple parameters, making the accuracy of status judgments susceptible to load fluctuations. Furthermore, they cannot adaptively identify different load scenarios and adjust assessment benchmarks by comparing with historical values, leading to frequent false alarms during scenario switching. Moreover, they cannot provide early warnings of faults by analyzing parameter changes over adjacent time periods, often only issuing alarms when the fault has reached a critical stage. Summary of the Invention

[0004] This invention provides a method for assessing the status of intelligent power metering boxes based on time-series data to solve the above-mentioned problems.

[0005] The intelligent energy metering box status assessment method based on time-series data of the present invention adopts the following technical solution: One embodiment of the present invention provides a method for assessing the status of a smart energy metering box based on time-series data. The method includes: acquiring multi-parameter time-series data and ambient temperature data of the smart energy metering box, and establishing historical baseline files for each parameter; extracting real abnormal components through multi-parameter correlation time-series analysis; identifying the current load scenario based on the historical baseline files and calculating the abnormal deviation; providing fault warning based on the variation patterns of multiple parameters in adjacent time periods and the abnormal deviation; and outputting the status assessment results and warning information.

[0006] Furthermore, the step of extracting true anomalous components through multi-parameter correlation time series analysis includes: calculating the correlation stability between current and temperature rise, the correlation stability between voltage and current, and the correlation stability between current and harmonics; determining the comprehensive correlation stability based on the correlation stability between current and temperature rise, the correlation stability between voltage and current, and the correlation stability between current and harmonics; and extracting true anomalous components based on the comprehensive correlation stability.

[0007] Further, the step of extracting real abnormal components based on the comprehensive correlation stability includes: comparing the comprehensive correlation stability with a preset correlation stability threshold; if the comprehensive correlation stability is lower than the preset correlation stability threshold, then it is determined that there is a real abnormality without load factors, and the parameter deviation within the current window is determined as a real abnormal component; if the comprehensive correlation stability is not lower than the preset correlation stability threshold, then the current state is determined to be normal.

[0008] Furthermore, the step of identifying the current load scenario based on the historical benchmark archive and calculating the abnormal deviation includes: extracting load features from the current time-series data; comparing the load features with historical load features in the historical benchmark archive to identify the current load scenario; and extracting the corresponding dynamic evaluation benchmark from the historical benchmark archive based on the current load scenario and calculating the abnormal deviation.

[0009] Further, the step of comparing the load characteristics with historical load characteristics in the historical benchmark archive to identify the current load scenario includes: extracting the mean current, coefficient of variation of current, and mean power factor from the current time-series data; calculating the deviation between the mean current, the coefficient of variation of current, and the mean power factor and the characteristics of the same historical period; if each deviation is less than a preset deviation threshold, then the load scenario label corresponding to the same historical period is determined as the current load scenario; if any deviation is greater than or equal to the preset deviation threshold, then an independent judgment is made based on the current load characteristics to determine the current load scenario.

[0010] Furthermore, the step of extracting the corresponding dynamic evaluation benchmark from the historical benchmark archive based on the current load scenario and calculating the abnormal deviation includes: extracting the parameter benchmark mean and benchmark standard deviation under the corresponding scenario from the historical benchmark archive based on the current load scenario; and calculating the abnormal deviation based on the parameter benchmark mean, the benchmark standard deviation and the current value of the parameter.

[0011] Furthermore, the fault warning based on the variation pattern of multiple parameters in adjacent time periods and the abnormal deviation includes: determining a consistency index of the direction of change of multiple parameters, a cumulative index of the magnitude of change of abnormal deviation, and a persistence index of the rate of change of abnormal deviation; determining a fault warning index based on the consistency index of the direction of change, the cumulative index of the magnitude of change of abnormal deviation, and the persistence index of the rate of change of abnormal deviation; and determining a fault warning level based on the fault warning index and a preset warning threshold.

[0012] Furthermore, the output status assessment results and early warning information include: summarizing the current load scenario, comprehensive correlation stability, abnormal deviation, fault early warning index, and fault early warning level into an assessment report; outputting the assessment report, and outputting corresponding early warning information according to the fault early warning level.

[0013] Furthermore, the method also includes: when the fault warning level reaches the warning level, determining the fault type based on the combination of the abnormal deviation amount and the performance of each of the associated stability values; if the abnormal temperature deviation amount exceeds the first preset deviation threshold, the current-temperature rise correlation stability is lower than the first preset stability threshold, and the voltage-current correlation stability is lower than the second preset stability threshold, then it is determined to be a poor contact fault; if the abnormal temperature deviation amount exceeds the second preset deviation threshold, the current-temperature rise correlation stability is lower than the third preset stability threshold, and the voltage-current correlation stability exceeds the fourth preset stability threshold, then it is determined to be an insulation degradation fault; if the abnormal voltage deviation amount is lower than the third preset deviation threshold, the abnormal current deviation amount is lower than the fourth preset deviation threshold, and the voltage-current correlation stability is lower than the fifth preset stability threshold, then it is determined to be a power supply line abnormal fault; if none of the above combination conditions are met, and the fault warning level reaches the warning level, then it is determined to be a comprehensive abnormal fault.

[0014] Furthermore, after outputting the status assessment results and early warning information, the method further includes: classifying the time-series data according to load scenarios according to a preset period, incorporating it into the statistical pool of the corresponding scenario in the historical benchmark archive, recalculating the daily periodic mean curve and standard deviation curve of each parameter to update the historical benchmark archive; monitoring the long-term trend of the comprehensive correlation stability, and if the mean of the comprehensive correlation stability is lower than the preset long-term stability threshold for a consecutive preset number of days, generating chronic degradation warning information.

[0015] The beneficial effects of the technical solution of the present invention are: In this embodiment of the invention, multi-parameter time-series data and ambient temperature data of the smart energy metering box are acquired, and historical baseline files for each parameter are established; real abnormal components are extracted through multi-parameter correlation time-series analysis; the current load scenario is identified based on the historical baseline files, and the abnormal deviation is calculated; fault warning is performed based on the change pattern of multiple parameters in adjacent time periods and the abnormal deviation; and the status assessment results and warning information are output.

[0016] This invention, by constructing a three-layer progressive evaluation framework of correlation interference elimination, scene identification and adaptation, and time-series change pattern early warning, expands the metering box status evaluation from a single-point judgment with a fixed threshold to a progressive evaluation closed loop that is fully adaptive to all scenarios, interference-free, and fault-predictable, thus improving the adaptability and reliability of status evaluation in complex power environments. On the other hand, by calculating the correlation stability between parameter pairs through multi-parameter correlation time-series analysis, and utilizing the characteristic that the correlation relationship remains stable when multiple parameters respond synchronously to changes in load conditions, the interference of load condition changes on status judgment is effectively eliminated, making anomaly identification more accurate. Furthermore, by comparing multi-parameter time-series data with historical values ​​to identify the current load scenario and matching the corresponding dynamic evaluation benchmark, the evaluation standard is automatically adjusted according to the power scenario, reducing false alarms caused by scenario switching. Moreover, by analyzing the change patterns of multiple parameters in adjacent time periods to provide fault early warning, a progressive tracking from early symptoms to fault confirmation is achieved, and the type of fault can be determined based on abnormal deviations, providing maintenance personnel with sufficient response time and clear handling directions. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, 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.

[0018] Figure 1 A flowchart illustrating the method for assessing the status of a smart energy metering box based on time-series data, provided in an embodiment of the present invention. Figure 2 This is a flowchart illustrating the real abnormal component extraction scheme provided in an embodiment of the present invention. Detailed Implementation

[0019] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following detailed description, in conjunction with the accompanying drawings and preferred embodiments, provides a detailed account of the specific implementation methods, structures, features, and effects of the method proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments may be combined in any suitable form. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0020] The specific scheme of the intelligent power metering box status assessment method based on time-series data provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0021] like Figure 1 As shown, this embodiment of the invention provides a method for assessing the status of a smart energy metering box based on time-series data, including: Step S110: Obtain multi-parameter time-series data and ambient temperature data from the smart energy metering box, and establish historical baseline files for each parameter.

[0022] The smart energy metering box incorporates voltage transformers, current transformers, temperature sensors, and humidity sensors, forming the hardware foundation for multi-parameter data acquisition. The voltage and current transformers output analog signals at a sampling frequency of at least once per minute. These analog signals are converted from analog to digital and then calculated using the root mean square (RMS) to obtain the effective values ​​of the voltage and current. The internal temperature is directly read by a temperature sensor located at the center of the metering box, and the internal humidity is directly read by a humidity sensor installed inside the metering box. Simultaneously, the power factor and harmonic distortion rate are calculated based on the output waveform of the current transformer. The power factor corresponds to the cosine of the phase difference between the fundamental voltage and current, and the harmonic distortion rate corresponds to the ratio of the sum of the effective values ​​of all harmonics to the effective value of the fundamental wave. The acquired time-series data of each parameter are timestamped and stored in local memory, forming a historical database with a rolling retention period of at least ninety days. In the historical database, each parameter has an independent historical baseline file, containing daily periodic mean curves and standard deviation curves categorized by load scenario over the past thirty days. The daily average curve is obtained by taking the arithmetic mean of the sampled values ​​at the same time within 30 days under the same load scenario, while the standard deviation curve is calculated based on the statistical standard deviation of the same time period alignment.

[0023] Historical benchmark archives are updated daily. New data added that day is incorporated into the statistical pool for the corresponding load scenario, and the mean and standard deviation are recalculated to track the slow drift of the benchmark caused by seasonal changes and equipment aging. The first thirty days after commissioning serve as the initial benchmark learning period. During this period, only the accumulation and statistics of benchmark data are completed, and no abnormal alarms are triggered, thereby avoiding evaluation bias caused by insufficient benchmark samples.

[0024] Step S120: Extract the real abnormal components through multi-parameter correlation time series analysis.

[0025] It is understandable that changes in load conditions will cause synchronous fluctuations in parameters such as current, voltage, and temperature. The characteristics of these fluctuations in the time-series data of a single parameter are highly similar to those of real faults such as poor contact. Judging solely based on the threshold of a single parameter is highly prone to misjudgment. However, fluctuations caused by changes in load conditions are essentially external disturbances, and the intrinsic correlations between the parameters remain relatively stable. When the metering box itself malfunctions, the normal correlations between the parameters are broken. Based on this, this invention extracts the true abnormal components through multi-parameter correlation time-series analysis to remove the interference caused by changes in load conditions. This shifts the focus of state assessment from the absolute numerical shift of a single parameter to the stability changes in the correlations between multiple parameters, thereby effectively distinguishing between external load disturbances and equipment malfunctions. The following describes one extraction scheme for the true abnormal components: like Figure 2 As shown, optionally, step S120 above includes: Step S121: Calculate the stability related to current and temperature rise, the stability related to voltage and current, and the stability related to current and harmonics; Step S122: Determine the comprehensive correlation stability based on the correlation stability between current and temperature rise, voltage and current, and current and harmonics. Step S123: Extract the true anomaly components based on the comprehensive correlation stability.

[0026] The correlations between current and temperature rise, voltage and current, and current and harmonics constitute the three core pairs of parameter correlations in multi-parameter correlation time series analysis. Specifically, the current-temperature rise correlation reflects the load's heat generation and heat dissipation balance characteristics; the voltage-current correlation reflects the power supply circuit impedance characteristics; and the current-harmonics correlation reflects the nonlinear load influence characteristics. These three pairs of correlations cover the most critical electrothermal coupling, impedance characteristics, and power quality features in metering box operation. Their joint analysis can effectively capture abnormal changes in equipment status. Each pair of correlations is calculated within a sliding time window of sixty sampling points, corresponding to a one-hour monitoring duration, ensuring both statistical significance and timely capture of status changes.

[0027] The temperature rise is obtained by subtracting the ambient temperature from the current internal temperature of the chamber. The ambient temperature is taken from the reading of an external ambient temperature sensor installed in the metering chamber. The linear correlation coefficient between the current sequence and the temperature rise sequence is calculated within a sliding window. It is obtained by dividing the covariance of the current sequence and the temperature rise sequence within the window by the product of their standard deviations. Current-temperature rise correlation stability. The calculation formula is:

[0028] In the formula, The threshold for determining the correlation can be set to 0.7. This applies to the stability of the correlation between current and temperature rise mentioned above. The numerator part of the calculation formula plus When When it is zero equal Divide by ,at this time The value is approximately 0.41, rather than zero, to avoid drastic fluctuations in correlation stability due to short-term disturbances; the denominator part Will Normalize to the interval [0.41, 1]. When Approaching 1 o'clock A value close to 1 indicates a strong positive correlation between current and temperature rise, normal heat dissipation, and a stable correlation; when... far below hour A decrease indicates that the temperature rise does not respond synchronously to the increase in current, or the temperature rise increases abnormally while the current remains unchanged, indicating an abnormal correlation.

[0029] The analysis of the voltage-current relationship is carried out under the same sliding window mechanism, and the linear correlation coefficient between voltage and current is calculated within the sliding window. In a normal power supply circuit, voltage and current should have a weak correlation or be independent, that is... The voltage is close to 0; if the line impedance increases abnormally, a significant negative correlation appears between voltage and current, manifested as an increase in line voltage drop and a decrease in voltage when current increases. Voltage-current correlation stability. The calculation formula is:

[0030] when When it is 0 An equal value of 1 indicates a normal association; when When it increases A decrease indicates a possible abnormality in line impedance. Voltage-current related stability. The purpose of the calculation formula is to quantify the independence between voltage and current; the stronger the independence, the better. The closer it is to 1.

[0031] The analysis of the relationship between current and harmonics is also based on a sliding window. Within the sliding window, the linear correlation coefficient between the current amplitude and the total harmonic distortion rate is calculated. Under normal nonlinear loads, the two are positively correlated; however, if the equipment itself generates harmonics, the correlation may deviate from the normal range. Current and harmonic correlation stability. The calculation formula is:

[0032] The above-mentioned current and harmonic correlation stability The calculation method and the stability related to current and temperature rise similar, The value is also set to 0.7. Current and harmonic stability. The calculation formula is used to quantify the stability of the positive correlation between current and harmonics; the more stable the positive correlation, the better. The closer it is to 1.

[0033] The three pairs of correlation stability are combined into a comprehensive correlation stability through a product. The calculation formula is as follows: When the load conditions change, parameters such as current, voltage, temperature, and harmonics respond synchronously, and the correlation stability of the three parameter pairs remains close to 1. A value close to 1 is considered normal load fluctuation; when the metering box itself malfunctions, the correlation between at least one parameter pair is broken, and the corresponding correlation stability decreases. Decrease.

[0034] Optionally, the above-mentioned extraction of true anomalous components based on comprehensive correlation stability includes: comparing the comprehensive correlation stability with a preset correlation stability threshold; if the comprehensive correlation stability is lower than the preset correlation stability threshold, then it is determined that there is a true anomalous component without load factors, and the parameter deviation within the current window is determined as the true anomalous component; if the comprehensive correlation stability is not lower than the preset correlation stability threshold, then the current state is determined to be normal.

[0035] The overall correlation stability is compared with a preset correlation stability threshold. When When the value is below 0.6, a true anomaly without load factors is identified, and the parameter deviation within the current window is determined as the true anomaly component; when... If the value is not lower than 0.6, the current state is considered normal.

[0036] Step S130: Identify the current load scenario based on historical benchmark files and calculate the abnormal deviation.

[0037] Optionally, step S130 above includes: extracting load features from the current time series data; comparing the load features with historical load features in the historical benchmark archive to identify the current load scenario; and based on the current load scenario, extracting the corresponding dynamic evaluation benchmark from the historical benchmark archive and calculating the abnormal deviation.

[0038] It is understandable that the characteristics of metering boxes, such as current amplitude, fluctuation frequency, and power factor, vary significantly under different power consumption scenarios, and the range of parameter fluctuations under normal conditions differs. If a uniform and fixed evaluation benchmark is used, normal parameter fluctuations are easily misjudged as abnormal during load mode switching at different times and seasons, while real abnormal changes may be ignored because they are within the normal fluctuation range, leading to false alarms or missed alarms. Therefore, this embodiment of the invention identifies the current load scenario in order to match an appropriate dynamic evaluation benchmark for subsequent status assessments, so that the evaluation standard is automatically adjusted according to the power consumption scenario, thereby accurately reflecting the actual operating status of the metering box under different scenarios. The following describes the scheme for determining the current load scenario: Optionally, the above-mentioned comparison of load characteristics with historical load characteristics in historical benchmark archives to identify the current load scenario includes: extracting the mean current, coefficient of variation of current, and mean power factor from the current time series data; calculating the deviation between the mean current, coefficient of variation of current, and mean power factor and the characteristics of the same historical period; if each deviation is less than a preset deviation threshold, then the load scenario label corresponding to the same historical period is determined as the current load scenario; if any deviation is greater than or equal to the preset deviation threshold, then an independent judgment is made based on the current load characteristics to determine the current load scenario.

[0039] The mean current, coefficient of variation of current, and mean power factor are extracted from time-series data within a preset time period as load feature vectors. The mean current reflects the overall power consumption level of the load during that period, the mean power factor reflects the impedance characteristics of the load, and the coefficient of variation of current reflects the severity of current fluctuations. These three features collectively characterize the load pattern of the current time period, providing a quantitative basis for subsequent comparisons with historical data. The formula for calculating the coefficient of variation of current is: In the formula, This is the average current. The standard deviation of the current can characterize the dispersion of the current within a statistical period. This is the coefficient of variation of the current. The larger the value, the more drastic the current fluctuation relative to its mean, and the higher the instability of the corresponding load.

[0040] The historical baseline archive stores a set of feature vectors for the same time point over the past thirty days, including the average current for the same historical period. Current variation coefficient and the mean power factor The deviation between the current load characteristics and historical characteristics is calculated using relative deviation to eliminate the impact of differences in the dimensions of different parameters. The formulas for calculating each deviation are as follows:

[0041]

[0042]

[0043] In the formula, The deviation of the mean current; The deviation of the coefficient of variation of current; This represents the deviation of the power factor from the mean. The formulas above calculate this by dividing the absolute difference between the current value and the historical average for the same period by the historical average, reflecting the degree of deviation of the current load characteristics from the historical normal range. It can be understood that the 0.01 in the above calculation formulas prevents the denominator from being zero.

[0044] The three deviations mentioned above are compared with preset deviation thresholds. When all three deviations are less than 0.5, it indicates that the current load characteristics are highly consistent with the historical period of the same time, and the load pattern has not changed significantly. In this case, the load scenario label corresponding to this historical period is directly used. As the current scenario .

[0045] When any deviation is greater than or equal to 0.5, it indicates a significant difference between the current load characteristics and historical data for the same period, suggesting a possible load mode shift. An independent judgment based on the current characteristics is required. Set load scenario judgment rules: When This is considered a low-load scenario. The low load current threshold is set to 5% of the rated current, read from the metering box nameplate parameters; when and and The scenario was determined to be a stable industrial load. The threshold for determining fluctuation is set to 0.3. The threshold for determining the industrial power factor is 0.85; when and The scenario is determined to be a residential fluctuating load scenario. The above determination rule classifies scenarios based on the combined characteristics of current level, fluctuation degree, and power factor, covering typical electricity consumption patterns commonly found in metering boxes.

[0046] Optionally, the above-mentioned extraction of the corresponding dynamic evaluation benchmark from the historical benchmark archive based on the current load scenario and calculation of the abnormal deviation includes: extracting the parameter benchmark mean and benchmark standard deviation under the corresponding scenario from the historical benchmark archive based on the current load scenario; and calculating the abnormal deviation based on the parameter benchmark mean, benchmark standard deviation and the current value of the parameter.

[0047] Historical benchmark archives for different load scenarios are maintained separately using independent statistical mechanisms to ensure that the statistical characteristics of parameter benchmarks in each scenario are not affected by data from other scenarios. Historical benchmark archives are stored categorized by load scenario, with each scenario containing the benchmark mean and standard deviation for each parameter. Once the current load scenario is identified, only the benchmark data corresponding to that scenario is retrieved from the historical benchmark archives, avoiding evaluation distortion caused by mixing data from different scenarios. Under low-load scenarios, the evaluation focuses on temperature drift trends, while industrial stable load scenarios and residential fluctuating load scenarios use their own independently statistically calculated parameter benchmark means and standard deviations.

[0048] The abnormal deviation is obtained by comparing the current value of the parameter with the baseline mean in the relevant scenario and normalizing it relative to the baseline standard deviation. Its calculation formula is as follows: In the formula, For parameters The real-time sampled value at the current moment; This is the baseline average value of this parameter under the current load scenario; This represents the baseline standard deviation of this parameter under the current load scenario; For parameters Abnormal deviation.

[0049] The sign and magnitude of the abnormal deviation reflect the direction and degree of the parameter's deviation from the normal baseline, respectively, providing a quantitative basis for subsequent fault early warning. When the value is positive, it indicates that the current value of the parameter is higher than the normal baseline level in its scenario; when... A negative value indicates that the current value of the parameter is lower than the normal baseline level. The larger the absolute value of the deviation, the more significant the deviation of the current parameter state from the normal range. If no abnormal components are extracted, the subsequent steps are skipped, the current state is directly determined to be normal, and the current scene recognition result is only used for subsequent historical baseline file updates.

[0050] The calculation of abnormal deviations is based on the premise that the true abnormal components have been extracted by multi-parameter correlation time series analysis, in order to avoid meaningless deviation quantification of normal load fluctuations. When the overall correlation stability is lower than the preset correlation stability threshold, it is determined that there is a real anomaly due to non-load factors. At this time, the calculation of abnormal deviations is initiated, quantifying the parameter deviations within the current window relative to the dynamic benchmark of the scene. When the overall correlation stability is not lower than the preset correlation stability threshold, it indicates that the current state is normal load fluctuation. At this time, the calculation of abnormal deviations and subsequent fault warning assessment are skipped, and the current state is directly determined to be normal. The current scene identification results are only used for subsequent updates of the historical benchmark archive.

[0051] Step S140: Perform fault warning based on the variation pattern and abnormal deviation of multiple parameters in adjacent time periods.

[0052] Optionally, step S140 includes: determining a consistency index for the direction of change of multiple parameters, a cumulative index for the magnitude of abnormal deviation, and a persistence index for the rate of change of abnormal deviation; determining a fault warning index based on the consistency index for the direction of change, the cumulative index for the magnitude of abnormal deviation, and the persistence index for the rate of change of abnormal deviation; and determining a fault warning level based on the fault warning index and a preset warning threshold.

[0053] Adjacent time periods are defined as time intervals within a sliding window that are equally divided into two halves. In a sliding window with a length of sixty sampling points, the first and second halves each contain thirty sampling points, corresponding to a monitoring duration of thirty minutes. The mean change of each parameter between the first and second halves is calculated using the following formula:

[0054] In the formula, This represents the average parameter value of the first thirty sampling points; This represents the average parameter value of the thirty sampling points in the latter half of the test. This represents the change in mean. (The change in mean current) ,in This represents the average current value of the first thirty sampling points. The average current value is calculated from the thirty sampling points in the latter half of the test. The current sampling value is the effective value of the current output by the current transformer in the metering box, and the sampling frequency is once per minute. The average temperature rise change is also considered. ,in and The values ​​represent the average temperature rise at thirty sampling points in the first and second halves of the test, respectively. The temperature rise is obtained by subtracting the ambient temperature from the internal temperature. The average voltage change is also shown. ,in and The values ​​are the average voltage values ​​of thirty sampling points in the first and second halves, respectively. The voltage sampling value is the effective value of the voltage output by the voltage transformer in the metering box, and the sampling frequency is once per minute. This indicates that the parameter increases in the latter half. It indicates a decrease.

[0055] The multi-parameter change direction consistency index is used to quantify the degree of coordination among the change directions of three parameters: current, temperature rise, and voltage. It is obtained by statistically analyzing the consistency of the signs of the mean changes in current, temperature rise, and voltage. If all three have the same sign, the multi-parameter change direction consistency index is considered good. The value is 1; if both have the same sign or one has a different sign, then... The value is 0.5; if the three values ​​have different signs, then... The value is 0. A larger value indicates that the changes in multiple parameters are more consistent, and the greater the possibility of load fluctuations. When And overall correlation stability When the load change is normal, it is determined and no warning is triggered.

[0056] The cumulative magnitude of abnormal deviation is used to reflect the cumulative fluctuation of abnormal deviation over a recent period. Its calculation formula is as follows:

[0057] In the formula, This represents the abnormal deviation at the current moment. to These represent the abnormal deviations at the first three time points; This is a cumulative indicator of the magnitude of abnormal deviation. The larger the value, the more the abnormal deviation has been fluctuating or monotonically increasing in the near future, reflecting the accumulation and development of the abnormality.

[0058] The rate of change persistence index for abnormal deviations is used to measure the degree of stability of the trend of abnormal deviations. Its calculation formula is as follows:

[0059] In the formula, The number of consecutive moments for the statistics is set to 6, corresponding to a monitoring duration of six minutes. The number of times during which the abnormal deviation remains in the same direction across the K consecutive time points, i.e., from to The number of times the symbol remains the same as the one in the previous time step; This is an indicator of the persistence of the rate of change of abnormal deviation. The larger the value, the more persistent the trend, and the less likely it is to be a random disturbance.

[0060] The fault warning index is obtained by integrating three factors: the degree of correlation anomaly, the ratio of the cumulative rate of anomaly change to its persistence, and the degree of non-load consistency. Its calculation formula is as follows:

[0061] In the formula, To assess overall correlation stability; This is a cumulative indicator of the magnitude of abnormal deviation. It serves as an indicator of the rate of change of abnormal deviations. It serves as an indicator of the consistency of the direction of change of multiple parameters; This is the fault early warning index. (The above fault early warning index...) In the calculation formula, the first term It can reflect the degree of correlation anomaly. The smaller the value, the more significant the anomaly of the non-loading factor; the larger the value of this item, the greater the anomaly. The ratio that reflects the cumulative rate of abnormal change to its persistence. Larger and A smaller ratio indicates a recent sharp, step-like change in the abnormal deviation, but it has not yet formed a stable unidirectional trend. An increase in the ratio reflects the risk of sudden failure. Larger and A larger ratio indicates that the abnormal deviation is deteriorating at a steady rate. A moderate ratio reflects the risk of progressive failure. Smaller and A larger value indicates that the abnormal deviation is continuously drifting in one direction with small steps; a smaller ratio reflects a slow degradation process. The third term... It can reflect the degree of non-load consistency. The smaller the value, the more inconsistent the changes in multiple parameters, and the more likely it is a fault in the equipment itself rather than a change in load. The higher the value, the higher the probability of a fault occurring.

[0062] Fault warning level The classification is based on the comparison between the fault warning index and the preset warning threshold. When this time, it is determined to be a normal state. ;when At that time, it was determined to be an early warning. ;when At that time, it was determined to be a mid-term warning. ;when At that time, it was determined to be an emergency alarm. During the alert period, the calculation is recalculated every hour. And update the fault warning level. If it continues for 24 hours Falling back below 1.0 and If the value recovers to above 0.8, the warning will be lifted and the system will be restored. .

[0063] Step S150: Output the status assessment results and early warning information.

[0064] Optionally, step S150 above includes: summarizing the current load scenario, overall correlation stability, abnormal deviation, fault warning index, and fault warning level into an evaluation report; outputting the evaluation report, and outputting corresponding warning information according to the fault warning level.

[0065] The aforementioned assessment report compiles core parameters from the entire status assessment chain, including the current load scenario, overall correlation stability, abnormal deviations of each parameter, fault warning index, and fault warning level. Specifically, the current load scenario reflects the power consumption pattern of the metering box; the overall correlation stability quantifies the health of the multi-parameter relationships; abnormal deviations characterize the magnitude of each parameter's offset from the dynamic baseline; and the fault warning index and fault warning level comprehensively reflect the probability and urgency of a fault. These parameters are integrated into the assessment report in a structured format, providing maintenance personnel with a complete picture of the metering box's operational status.

[0066] The assessment report is visualized through the monitoring terminal interface, with warning levels distinguished by color. Normal status, early warning, mid-term alarm, and emergency alarm correspond to different color levels, allowing maintenance personnel to intuitively perceive the degree of risk. The interface also displays a bar chart showing the correlation stability of various parameters, graphically illustrating the stability of the three pairs of correlations: current and temperature rise, voltage and current, and current and harmonics. It also displays trend curves of abnormal deviations, reflecting the historical trajectory of each parameter's deviation from the dynamic baseline.

[0067] Optionally, the above-mentioned intelligent energy metering box status assessment method based on time-series data further includes: when the fault warning level reaches the warning level, judging the fault type based on the abnormal deviation amount and the performance combination of each associated stability; If the abnormal temperature deviation exceeds the first preset deviation threshold, the current-temperature rise correlation stability is lower than the first preset stability threshold, and the voltage-current correlation stability is lower than the second preset stability threshold, then it is determined to be a contact failure. If the abnormal temperature deviation exceeds the second preset deviation threshold, the current-temperature rise correlation stability preset stability threshold, and the voltage-current correlation stability exceeds the fourth preset stability threshold, it is determined to be an insulation degradation type fault. If the voltage deviation is lower than the third preset deviation threshold, the current deviation is lower than the fourth preset deviation threshold, and the voltage-current correlation stability is lower than the fifth preset stability threshold, then it is determined to be a power supply line abnormality fault. If none of the above combined conditions are met, and the fault warning level reaches the warning level, then it is determined to be a comprehensive abnormal fault.

[0068] Fault type determination is initiated when the fault warning level reaches the warning level. At this time, the fault type is identified based on the combination of abnormal deviation and the performance of various associated stability values. Abstract numerical indicators are transformed into identifiable fault types, providing a directional basis for operation and maintenance.

[0069] The identification of contact-related faults relies on a combination of three conditions: abnormal temperature deviation, stability of the current-temperature rise correlation, and stability of the voltage-current correlation. For example, the criteria are: when the abnormal temperature deviation is greater than 1.5, the stability of the current-temperature rise correlation is less than 0.6, and the stability of the voltage-current correlation is less than 0.7, it is determined to be a contact-related fault. The physical mechanism of this judgment logic is that an abnormal temperature rise disrupts the current-temperature rise correlation, and the voltage-current correlation becomes abnormal. This indicates that there is localized heating caused by increased contact resistance in the circuit. The heat accumulation causes the temperature to deviate from the normal reference, and the increased contact resistance alters the impedance characteristics of the power supply circuit, thereby disrupting the normal independent relationship between voltage and current.

[0070] Criteria for identifying insulation degradation faults include: abnormal temperature deviation greater than 1.0, current-temperature rise correlation stability below 0.5, and voltage-current correlation stability above 0.8. In insulation degradation faults, abnormal temperature increases, but the voltage-current correlation remains normal, indicating that the heating is not caused by contact problems in the conductive circuit. Aging of the insulation material leads to increased leakage current. This leakage current generates Joule heat as it flows through the insulating medium, causing the temperature to rise. However, since the conductive circuit itself does not exhibit poor contact or abnormal line impedance, the voltage-current correlation remains intact. Therefore, the current-temperature rise correlation stability decreases, while the voltage-current correlation stability remains at a high level.

[0071] The criteria for determining abnormal faults in power supply lines include, for example, voltage deviation less than -1.0, current deviation less than -0.5, and voltage-current correlation stability below 0.6. These criteria indicate that both voltage and current are simultaneously low, and the voltage-current correlation is abnormal. This is caused by excessive voltage drop or a fault on the power supply side. An abnormal increase in line impedance or a voltage drop in the power supply leads to a decrease in load voltage, which in turn causes a decrease in current. This results in an abnormal negative correlation between voltage and current, significantly reducing the voltage-current correlation stability.

[0072] When the combined conditions of the above three typical faults are not met, but the fault warning level still reaches the warning level, it is judged as a comprehensive abnormal fault. This judgment result corresponds to an atypical abnormality caused by the intertwining of multiple factors. Its performance characteristics do not fall into the typical parameter combination range of a single fault type, indicating that there may be multiple concurrent faults or new abnormal modes inside the metering box.

[0073] The fault type determination result and warning level are output together, along with corresponding handling suggestions. For example, for faults with poor contact, it is recommended to check the wiring terminals and switch contacts; for faults with insulation deterioration, it is recommended to perform insulation resistance testing and assess whether insulation components need to be replaced; for faults with abnormal power supply lines, it is recommended to check the upstream power distribution lines and transformers; and for faults with comprehensive abnormalities, it is recommended to conduct a comprehensive inspection.

[0074] Optionally, after step S150 above, the above-mentioned intelligent energy metering box status assessment method based on time-series data further includes: classifying the time-series data according to load scenarios according to a preset period, incorporating it into the statistical pool of the corresponding scenario in the historical benchmark archive, recalculating the daily periodic mean curve and standard deviation curve of each parameter to update the historical benchmark archive; monitoring the long-term trend of comprehensive correlation stability, and if the mean of comprehensive correlation stability is lower than the preset long-term stability threshold within a consecutive preset number of days, generating chronic degradation warning information.

[0075] The aforementioned historical baseline archives are updated according to a preset cycle, which can be set to midnight every day. During the update process, the time-series data collected that day are categorized according to the identified load scenarios and included in the statistical pool of the corresponding scenario in the historical baseline archives. This categorization and inclusion mechanism ensures that data from different load scenarios accumulates independently, avoiding statistical distortion caused by data mixing between scenarios, and ensuring that the baseline archives for each scenario always reflect the true operational statistical patterns of that scenario.

[0076] After being included in the corresponding scenario statistical pool, the daily periodic mean curves and standard deviation curves of each parameter are recalculated. The daily periodic mean curve is obtained by taking the arithmetic mean of the sampled values ​​at the same time in the updated statistical pool, and the standard deviation curve is calculated based on the statistical standard deviation of the same time period. The recalculated mean curves and standard deviation curves replace the original data in the historical benchmark archive, completing the rolling update of the benchmark archive.

[0077] While updating historical baseline data, the long-term trend of the comprehensive correlation stability is monitored. As a core indicator for quantifying the health of multi-parameter correlations, the long-term trend of comprehensive correlation stability reflects the gradual deterioration of the metering box equipment's condition. By continuously recording and analyzing the historical sequence of comprehensive correlation stability, its slow decline over time is captured to identify potential chronic degradation risks. If the arithmetic mean of the comprehensive correlation stability is lower than a preset long-term stability threshold for thirty consecutive days, a chronic degradation warning message is generated. This preset long-term stability threshold can be set to 0.7. The thirty-day monitoring window covers the monthly operating cycle, effectively filtering out short-term random fluctuations and focusing the judgment on the continuous deterioration of the equipment's condition. The generated chronic degradation warning message can be included in the annual maintenance report, alerting maintenance personnel that the metering box may be experiencing chronic degradation, requiring targeted in-depth testing and preventative maintenance to prevent the fault from worsening undetected.

[0078] This invention is now complete.

[0079] In summary, in this embodiment of the invention, multi-parameter time-series data and ambient temperature data of the intelligent power metering box are acquired to establish historical baseline files for each parameter; real abnormal components are extracted through multi-parameter correlation time-series analysis; the current load scenario is identified based on the historical baseline files, and abnormal deviations are calculated; fault warnings are given based on the changing patterns of multiple parameters in adjacent time periods and abnormal deviations; and status assessment results and warning information are output.

[0080] This invention constructs a three-tiered progressive evaluation framework encompassing interference removal, scene identification and adaptation, and early warning of temporal change patterns. This expands metering box status assessment from a single-point judgment based on a fixed threshold to a progressive evaluation closed loop that is fully adaptive across all scenarios, capable of removing interference, and providing early warning of faults. This enhances the adaptability and reliability of status assessment in complex power environments. Furthermore, by calculating the correlation stability between parameter pairs through multi-parameter correlation time-series analysis, and leveraging the characteristic that multi-parameter synchronous responses and stable correlations remain stable when load conditions change, the interference of load condition changes on status judgment is effectively removed, making anomaly identification more accurate. Additionally, by comparing multi-parameter time-series data with historical values ​​to identify the current load scenario and matching the corresponding dynamic evaluation benchmark, the evaluation standard automatically adjusts with the power usage scenario, reducing false alarms caused by scenario switching. Finally, by analyzing the variation patterns of multiple parameters in adjacent time periods for fault early warning, a progressive tracking from early symptoms to fault confirmation is achieved. Furthermore, the type of fault can be determined based on abnormal deviations, providing maintenance personnel with sufficient response time and clear handling directions.

[0081] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for assessing the condition of intelligent energy metering boxes based on time-series data, characterized in that, The method includes: Acquire multi-parameter time-series data and ambient temperature data from the smart energy metering box, and establish historical baseline archives for each parameter. Extracting real anomaly components through multi-parameter correlation time series analysis; The current load scenario is identified based on the historical benchmark archives, and the abnormal deviation is calculated; Fault warning is based on the variation pattern of multiple parameters in adjacent time periods and the abnormal deviation amount; Output status assessment results and early warning information.

2. The method for assessing the status of intelligent energy metering boxes based on time-series data according to claim 1, characterized in that, The extraction of true anomaly components through multi-parameter correlation time series analysis includes: Calculate the stability related to current and temperature rise, the stability related to voltage and current, and the stability related to current and harmonics; Based on the current-temperature rise correlation stability, the voltage-current correlation stability, and the current-harmonic correlation stability, a comprehensive correlation stability is determined. Based on the comprehensive correlation stability, the true anomaly components are extracted.

3. The method for assessing the status of intelligent energy metering boxes based on time-series data according to claim 2, characterized in that, The extraction of true anomaly components based on the comprehensive correlation stability includes: The overall correlation stability is compared with a preset correlation stability threshold; If the overall correlation stability is lower than the preset correlation stability threshold, it is determined that there is a real anomaly without load factors, and the parameter deviation in the current window is determined as the real anomaly component. If the overall correlation stability is not lower than the preset correlation stability threshold, then the current state is determined to be normal.

4. The method for assessing the status of intelligent energy metering boxes based on time-series data according to claim 1, characterized in that, The process of identifying the current load scenario based on the historical benchmark archive and calculating the abnormal deviation includes: Extract load characteristics from the current time-series data; The load characteristics are compared with historical load characteristics in the historical benchmark archive to identify the current load scenario; Based on the current load scenario, the corresponding dynamic evaluation benchmark is extracted from the historical benchmark archive, and the abnormal deviation is calculated.

5. The method for assessing the status of intelligent energy metering boxes based on time-series data according to claim 4, characterized in that, The step of comparing the load characteristics with historical load characteristics in the historical benchmark archive to identify the current load scenario includes: Extract the mean current, coefficient of variation of current, and mean power factor from the current time series data; Calculate the deviation between the mean current, the coefficient of variation of the current, and the mean power factor from the characteristics of the same historical period; If all the deviations are less than the preset deviation threshold, then the load scenario label corresponding to the same historical period is determined as the current load scenario. If any of the deviations is greater than or equal to the preset deviation threshold, an independent determination is made based on the current load characteristics to determine the current load scenario.

6. The method for assessing the status of intelligent power metering boxes based on time-series data according to claim 4, characterized in that, The step of extracting the corresponding dynamic evaluation benchmark from the historical benchmark archive based on the current load scenario and calculating the abnormal deviation includes: Based on the current load scenario, extract the benchmark mean and benchmark standard deviation of the parameters under the corresponding scenario from the historical benchmark archive; The abnormal deviation is calculated based on the baseline mean of the parameter, the baseline standard deviation, and the current value of the parameter.

7. The method for assessing the status of intelligent energy metering boxes based on time-series data according to claim 1, characterized in that, The fault warning based on the variation pattern of multiple parameters in adjacent time periods and the abnormal deviation includes: Determine the consistency index of the direction of change of multiple parameters, the cumulative index of the magnitude of abnormal deviation, and the persistence index of the rate of change of abnormal deviation. Based on the consistency index of the direction of change, the cumulative index of the magnitude of the abnormal deviation, and the persistence index of the rate of change of the abnormal deviation, a fault warning index is determined. The fault warning level is determined based on the fault warning index and the preset warning threshold.

8. The method for assessing the condition of a smart energy metering box based on time-series data according to any one of claims 1-7, characterized in that, The output status assessment results and early warning information include: The current load scenario, overall correlation stability, abnormal deviation, fault warning index, and fault warning level are summarized into an evaluation report; Output the assessment report and output the corresponding warning information according to the fault warning level.

9. The method for assessing the condition of a smart energy metering box based on time-series data according to any one of claims 1-7, characterized in that, The method further includes: When the fault warning level reaches the warning level, the fault type is determined based on the abnormal deviation amount and the performance combination of each of the associated stability values. If the abnormal temperature deviation exceeds the first preset deviation threshold, the current-temperature rise correlation stability is lower than the first preset stability threshold, and the voltage-current correlation stability is lower than the second preset stability threshold, then it is determined to be a contact failure. If the abnormal temperature deviation exceeds the second preset deviation threshold, the current-temperature rise correlation stability is lower than the third preset stability threshold, and the voltage-current correlation stability exceeds the fourth preset stability threshold, then it is determined to be an insulation degradation type fault. If the voltage deviation is lower than the third preset deviation threshold, the current deviation is lower than the fourth preset deviation threshold, and the voltage-current correlation stability is lower than the fifth preset stability threshold, then it is determined to be a power supply line abnormality fault. If none of the above combined conditions are met, and the fault warning level reaches the warning level, then it is determined to be a comprehensive abnormal fault.

10. The method for assessing the condition of a smart energy metering box based on time-series data according to any one of claims 1-7, characterized in that, After outputting the status assessment results and early warning information, the method further includes: The time-series data is classified according to load scenarios according to a preset period and included in the statistical pool of the corresponding scenario in the historical benchmark archive. The daily periodic mean curve and standard deviation curve of each parameter are recalculated to update the historical benchmark archive. Monitor the long-term trend of the comprehensive correlation stability. If the average value of the comprehensive correlation stability is lower than the preset long-term stability threshold for a consecutive preset number of days, a chronic degradation warning message is generated.

Images