An electric energy meter operation quality early warning method based on intelligent fusion terminal

Abstract

The application relates to the field of data processing, in particular to a power meter operation quality early warning method based on an intelligent fusion terminal, which comprises the following steps: obtaining total power consumed by a user in each day from a target power meter as daily power consumption, and arranging all the daily power consumptions according to time sequences to obtain an original sequence; processing the original sequence to obtain a first-order difference sequence; presetting an initial value of a cumulative sum, obtaining an improved cumulative sum of the current day by using a composite reference value calculated through the original sequence and the first-order difference sequence, and obtaining an improved decision threshold value through the composite reference value; and performing operation fault early warning of the target power meter in the current day according to the improved cumulative sum and the improved decision threshold value. The application can more accurately detect the operation fault of the power meter, such as metering drift, by comprehensively considering the fluctuation characteristics and scale factors of the power consumption, adaptively adjusting the decision threshold value and the cumulative sum calculation, and reducing false positives and false negatives.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing. More particularly, the present application relates to a power meter operation quality early warning method based on an intelligent fusion terminal. BACKGROUND

[0002] In the construction of new power systems, intelligent distribution areas are the core units supporting the lean operation and maintenance of power grids. The intelligent fusion terminal deployed at the area side is a key node for edge computing and data aggregation, which can collect, store and analyze localized data of hundreds of intelligent power meters in the jurisdiction.

[0003] It is crucial to conduct quality early warning on power meters. Through early warning, the accuracy of electricity fee calculation can be ensured, and users can avoid overpaying or underpaying electricity fees due to inaccurate power meters. At the same time, accurate power meter data also plays a key role in the stable operation of power grids.

[0004] The existing technology such as the CUSUM algorithm realizes real-time detection of power meter measurement drift by accumulating the deviation of electricity consumption data from the expected value. However, it has problems such as fixed parameters that are difficult to adapt to different user characteristics, and not distinguishing between normal fluctuations and real drift, which affects the accuracy of power meter quality detection. SUMMARY

[0005] To solve the above technical problems that the existing technology has certain limitations and reduces the accuracy of power meter operation quality detection, the present application provides the following technical solutions.

[0006] A power meter operation quality early warning method based on an intelligent fusion terminal, comprising:

[0007] Obtain the total power consumed by the user in each day from the target power meter as daily electricity consumption, and arrange all daily electricity consumption in time sequence to obtain an original sequence; process the original sequence to obtain a first-order difference sequence; preset an initial value of cumulative sum, obtain an improved cumulative sum of the current day by using a composite reference value calculated from the original sequence and the first-order difference sequence, and obtain an improved decision threshold value by using the composite reference value; and perform operation failure early warning on the target power meter of the current day according to the improved cumulative sum and the improved decision threshold value.

[0008] The calculation process of the composite reference value comprises:

[0009] Calculate the load fluctuation index of the first-order difference sequence and the scale correction coefficient of the original sequence, and take the product of the load fluctuation index and the scale correction coefficient as the composite reference value.

[0010] Preferably, the calculation process of the load fluctuation index comprises:

[0011] Calculate the fluctuation amplitude and the direction confusion degree of the first-order difference sequence;

[0012] Take the product of the fluctuation amplitude and the direction confusion degree as the load fluctuation index of the first-order difference sequence.

[0013] Preferably, the calculation process of the scale correction coefficient comprises:

[0014] Pre-set a minimum effective daily power consumption, when the median of the original sequence is less than or equal to the minimum effective daily power consumption, set the scale correction coefficient to 1; when the median of the original sequence is greater than the minimum effective daily power consumption, take the ratio of the median of the original sequence to the minimum effective daily power consumption as the scale correction coefficient.

[0015] Preferably, the obtaining process of the improved decision threshold comprises:

[0016] Calculate the square of the composite reference value as the first item, and calculate the product of the pre-set minimum detectable drift value and the median of the original sequence as the second item;

[0017] Take the ratio of the first item to the second item as the improved decision threshold.

[0018] Preferably, the obtaining step of the improved cumulative sum of the current day comprises:

[0019] Calculate the sum of the cumulative sum of the target electric energy meter of the previous day and the daily power consumption of the current day as the third item, subtract the composite reference value from the third item to obtain the first calculation result, if the first calculation result is negative, the cumulative sum of the current day is recorded as 0; if the first calculation result is positive or 0, the cumulative sum of the current day is equal to the first calculation result.

[0020] Preferably, the fluctuation amplitude is the median of the first-order difference sequence.

[0021] Preferably, the obtaining process of the direction confusion degree comprises:

[0022] Based on the direction change of the first-order difference sequence, a symbol sequence is extracted, the frequency of each element of the symbol sequence is counted, and the normalized Shannon entropy is calculated according to the frequency, and the normalized Shannon entropy is taken as the direction confusion degree.

[0023] Preferably, the fault early warning comprises:

[0024] If the improved cumulative sum of the current day is greater than or equal to the improved decision threshold, it is determined that the target electric energy meter has a positive drift, and a running quality early warning is triggered.

[0025] The beneficial effects of the present application are:

[0026] 1、The application can comprehensively capture the dynamic change characteristics of power consumption and user power consumption mode by constructing the original sequence and the first-order difference sequence, and calculating the composite reference value in combination with the load fluctuation index (including fluctuation amplitude and direction chaos degree) and the scale correction coefficient. The direction chaos degree quantifies the irregularity of power consumption behavior based on Shannon entropy, so that the detection is more sensitive to abnormal fluctuations, thereby more accurately identifying faults such as forward drift of electric energy metering, and reducing false positives and false negatives; the introduction of the scale correction coefficient considers the difference of user power consumption scale (compared by the median and the minimum effective daily power consumption), so that the method can adaptively adjust the detection sensitivity, for users with larger power consumption, the threshold is correspondingly increased to avoid triggering the warning due to normal power fluctuation; for users with smaller power consumption, a higher sensitivity is maintained to ensure that small-scale abnormalities can also be detected.

[0027] 2、The application dynamically adjusts the cumulative sum based on the composite reference value, avoiding the excessive sensitivity of the traditional cumulative sum method to noise and random fluctuations. The cumulative sum is corrected by the composite reference value, making the cumulative process more robust, triggering the warning only when there is a continuous anomaly, reducing the interference caused by short-term fluctuations, and calculating the improved decision threshold based on the composite reference value and the median of power consumption, so that the threshold changes adaptively with the data characteristics, further improving the stability and reliability of the detection. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 is a method flowchart of steps S1-S3 in an electric energy meter operation quality early warning method based on an intelligent fusion terminal according to an embodiment of the application. DETAILED DESCRIPTION

[0029] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are part of the embodiments of the application, rather than all the embodiments of the application.

[0030] The application scenario of the application is: using an improved one-sided CUSUM (Cumulative Sum Control Chart) algorithm to detect the forward drift of the daily power consumption of a single user newly collected by an electric energy meter every day, to realize automatic early warning of the operation quality of the electric energy meter.

[0031] REFERENCE Figure 1 An electric energy meter operation quality early warning method based on an intelligent fusion terminal includes steps S1-S3, and specifically as follows:

[0032] S1: Obtain the total power consumed by a user every day from a target electric energy meter as daily power consumption, and arrange all the daily power consumption in chronological order to construct an original sequence; and obtain a first-order difference sequence by processing the original sequence.

[0033] In one embodiment, with the intelligent fusion terminal, any power meter in the jurisdiction is randomly selected as a target power meter, and the total power consumed by the target power meter in each of the last N days, such as the last 30 days, is obtained as the daily power consumption of each day according to the set time (timing), and all the daily power consumptions obtained are arranged in time sequence to build an original sequence of the target power meter.

[0034] The above data acquisition method is that the power meter automatically freezes the power consumption data at zero o'clock every day and actively reports to the intelligent fusion terminal. These data reflect the actual power consumption of the user every day, and are essential basic input information for subsequent analysis of the running state of the power meter.

[0035] Due to the above collection process, there may be power outages, communication interruptions, meter transient faults, or extreme power consumption events (such as abnormal start or stop of equipment in a short period of time), etc., which may cause some abnormal points in the original sequence. Therefore, data points greater than the first threshold value in the original sequence are removed.

[0036] Among them, the median of the original sequence is calculated, and the absolute value of the difference between each data point in the original sequence and the median is calculated, and three times the median of all absolute values is taken as the first threshold value.

[0037] Based on the above cleaned original sequence, a first-order difference sequence is calculated. The first-order difference is obtained by calculating the difference between the daily power consumptions of adjacent two days.

[0038] By calculating the difference sequence, the focus can be shifted from the absolute size of the power consumption level to the dynamic fluctuation characteristics of the power consumption behavior.

[0039] S2: Obtain the improved cumulative sum of the current day using the composite reference value calculated from the original sequence and the first-order difference sequence, and obtain the improved decision threshold value through the composite reference value.

[0040] The one-sided CUSUM algorithm is a method based on cumulative sum control chart, which detects the small shift of sequence mean by monitoring the cumulative sum of sequence, and is suitable for detecting the positive drift of the daily power consumption of the newly collected power meter of a single user. The algorithm sets a target value and the allowed shift standard deviation multiple (usually represented by h and k, h is the decision threshold value, and k is the reference value), then calculates the cumulative sum and compares it with the decision threshold value.

[0041] However, it should be noted that the judgment of whether the target power meter is abnormal cannot simply rely on the size of the daily power consumption change. Because different users have different power consumption habits, some users have large daily power consumption fluctuations, but it is a normal random change; some users' power consumption seems stable, but there may be slow metering deviation.

[0042] Therefore, the application realizes the adaptive adjustment of the parameters by introducing the composite reference value and the improved decision threshold.

[0043] In one embodiment, it is considered that different power consumption behaviors have different load fluctuation characteristics. When the user power consumption is highly stable, such as the daily lighting load hardly changes, the first-order difference sequence is close to zero and the sign is consistent; when the user power consumption is frequently randomly started and stopped, such as the intermittent operation of household air conditioners and production equipment, the first-order difference sequence is large and the sign frequently switches. The metering drift of the electric energy meter usually shows a one-way, low-entropy slow cumulative deviation, which is different from the high-entropy, two-way characteristics of normal power consumption fluctuations.

[0044] Therefore, by calculating the load fluctuation index of the first-order difference sequence, the normal power consumption fluctuation and abnormal conditions such as the metering drift of the electric energy meter are distinguished, and further more, the CUSUM algorithm is provided with a more reasonable parameter setting basis, and the accuracy and reliability of the detection are improved.

[0045] The calculation process of the load fluctuation index of the first-order difference sequence is as follows:

[0046] Firstly, based on the first-order difference sequence S1, the sign sequence is extracted by the sign function. The sign function assigns +1, -1 and 0 to the positive, negative and zero values of the first-order difference sequence respectively, so as to clearly determine the direction of the daily power consumption change, i.e. +1 for rising, -1 for falling and 0 for unchanged.

[0047] Further, the normalized Shannon entropy is calculated by using the Shannon entropy formula and normalization, and the normalized Shannon entropy is used as the direction confusion degree of the first-order difference sequence.

[0048] The direction confusion degree is expressed by the following relation:

[0049]

[0050] In the formula, is the direction confusion degree, P (s) represents the probability of different signs in the sign sequence, log2 represents the logarithmic function with base 2, is the normalization coefficient.

[0051] When the frequencies of the three signs in the sign sequence are completely equal, i.e. the frequencies of the three signs are one-third, the original Shannon entropy is divided by the maximum Shannon entropy to obtain the normalized Shannon entropy. The smaller the value is, the more consistent the power consumption change direction is, and the larger the value is, the more random the power consumption increase and decrease are.

[0052] Then, the median of the first-order difference sequence is taken as the fluctuation amplitude of the first-order difference sequence. The median can directly reflect the typical amplitude of the daily power consumption.

[0053] Finally, the direction confusion degree and the fluctuation amplitude are fused to realize the joint quantification of "fluctuation size" and "fluctuation disorder".

[0054] The load fluctuation index of the first-order difference sequence is calculated by fusing the direction confusion degree and the fluctuation amplitude, which is expressed by a relationship as follows:

[0055]

[0056] In the formula, is the load fluctuation index of the first-order difference sequence, is the fluctuation amplitude of the first-order difference sequence, is the direction confusion degree of the first-order difference sequence.

[0057] In the formula, is designed to ensure that is 0, that is, the direction of the first-order difference sequence is completely consistent, and only reflects the single-direction fluctuation amplitude; is 1, that is, the direction of the first-order difference sequence is completely random, and the influence of random fluctuation is amplified.

[0058] In summary, when the user power consumption is highly stable or presents a single-direction trend (such as metering drift, that is, the power metering deviates but the direction is fixed), the value tends to be zero, in which case the system can maintain high sensitivity to small power consumption shifts and timely find possible problems; when the user power consumption is frequently and randomly started and stopped (for example, air conditioners and production equipment are intermittently operated), the value significantly increases, reflecting that the background noise is enhanced, that is, the power consumption fluctuation is complex and random, in which case the warning threshold needs to be correspondingly increased to suppress false positives and avoid misjudging normal power consumption fluctuation as a fault.

[0059] In an embodiment, the metering characteristics of different power consumption scales are also considered. If the same low-sensitivity decision threshold is used in the small-load area as in the large-load area, small shifts may be regarded as normal fluctuations, resulting in that the metering abnormality is not found for a long time; if the same high-sensitivity decision threshold is used in the large-load area as in the small-load area, the natural fluctuation may frequently exceed the decision threshold, resulting in invalid alarms.

[0060] Therefore, by calculating the scale correction coefficient of the original sequence, the metering characteristics and fluctuation rules of different power consumption scales are dynamically adapted.

[0061] The calculation process of the scale correction coefficient of the original sequence is as follows:

[0062] ​​First, the median of the original sequence is obtained, and the daily electricity consumption of the target electric energy meter under the minimum measurable load is defined. For example, according to JJG 596-2012 “Electronic AC Electric Energy Meter”, for the type of electric energy meter very common in the power system, the minimum load point of basic error detection is 0.05I, that is, when the load connected to the electric energy meter reaches 5% or more of the basic current, the measurement error of the electric energy meter must be controlled within ±2% to ensure the accuracy of the measurement. Therefore, the product of the basic current, rated voltage, time length of one day (i.e. 24 hours) and 5% rated load of the target electric energy meter is taken as the minimum effective daily electricity consumption.

[0063] Then, when the median of the original sequence is less than or equal to the minimum effective daily electricity consumption, the scale correction coefficient is set to 1; when the median of the original sequence is greater than the minimum effective daily electricity consumption, the ratio of the median of the original sequence to the minimum effective daily electricity consumption is taken as the scale correction coefficient of the original sequence.

[0064] In summary, when the typical daily electricity consumption of the user does not exceed the minimum effective daily electricity consumption, the target electric energy meter works in the small load area, at which time the measurement reliability decreases. In order to avoid sacrificing the sensitivity to electricity consumption abnormalities due to small electricity consumption, the scale correction coefficient is set to 1; when working in the large load area, the reference value is scaled up in proportion to meet the analysis requirements under large load conditions.

[0065] In one embodiment, the fluctuation characteristics (reflected by the load fluctuation index of the first-order difference sequence) and the physical boundary of the load (reflected by the scale correction coefficient of the original sequence) are combined into a composite reference value k to achieve accurate modeling of the individual electricity consumption behavior of the user. Specifically:

[0066] It can not only avoid misjudging high fluctuation users as having electricity problems due to normal random fluctuations (because the above-mentioned load fluctuation index reflects normal fluctuations, and the scale correction coefficient is reasonably normalized), but also ensure that low load users can still be effectively detected when there is a weak drift (i.e. a small electricity consumption anomaly) (the sensitivity is guaranteed by setting the scale correction coefficient to 1).

[0067] For example, the product of the above-mentioned calculated load fluctuation index and scale correction coefficient is taken as one of the parameters required by the CUSUM algorithm, i.e. the composite reference value.

[0068] After obtaining the above-mentioned composite reference value, it is introduced into the decision threshold to obtain an improved decision threshold, so that the decision threshold can be adjusted according to the distribution characteristics of the electricity consumption and the expected drift amplitude, so that when detecting positive drift, both small drifts can be captured sensitively and false positives caused by data fluctuations can be avoided.

[0069] In one embodiment, square the composite reference value as the first term; calculate the product of the preset minimum detectable drift value and the median of the original sequence as the second term; take the ratio of the first term and the second term as the improved decision threshold. The setting of the minimum detectable drift value needs to balance the false positive rate and the false negative rate, and the value is usually determined by experiment or prior knowledge, which can be initially set to 0.02 here.

[0070] Specifically, the first term reflects the cumulative effect of the drift, which amplifies the impact of long-term or cumulative drift by squaring, avoids false positives caused by short-term fluctuations, and is more sensitive to persistent anomalies. The second term ensures that the decision threshold is self-adaptive to the size of the electricity consumption data and the detection requirements.

[0071] When analyzing the positive drift of daily electricity consumption, the process of calculating the cumulative sum using the CUSUM algorithm can be divided into several key steps: determining the parameters, initializing, iterative calculation, and judging the drift.

[0072] Therefore, after calculating the composite reference value and the improved decision threshold of the algorithm, the following operations are performed:

[0073] First, initialize the cumulative sum, i.e. This means that there is no accumulated deviation before starting monitoring.

[0074] Then, starting from the first day, collect the new daily electricity consumption every day, calculate the sum of the cumulative sum of the previous day and the daily electricity consumption of the current day, as the third term, subtract the composite reference value from the third term to get the first calculation result, if the first calculation result is negative, the current day's cumulative sum is 0; if the first calculation result is positive or 0, the current day's cumulative sum is equal to the first calculation result.

[0075] Iterate to the current day to calculate the improved cumulative sum of the daily electricity consumption of the current day.

[0076] When the first calculation result is negative, it means that the electricity consumption of the first day has changed to some extent, but the accumulated deviation is not enough to make the cumulative sum positive relative to the improved reference value, so the cumulative sum is still 0; when the first calculation result is greater than 0, it means that the electricity consumption has a certain positive deviation relative to the reference.

[0077] It should be noted that the traditional CUSUM algorithm usually assumes that the process has a stable mean, such as the target mean, to directly measure the deviation of the current observation value from the normal state. However, in the embodiment of the present application, considering that the user's daily electricity consumption is affected by factors such as season, weather, and living habits, it does not have a fixed stable mean, so the daily electricity consumption of the current day is directly used as input here.

[0078] S3: Running fault early warning of the target electric energy meter in the current day is made according to the improved cumulative sum and the improved decision threshold.

[0079] In one embodiment, if the improved cumulative sum in the current day is greater than or equal to the improved decision threshold, it is determined that the target electric energy meter has a metering positive drift, and a running quality early warning is triggered.

[0080] In another embodiment, on the basis of the above-mentioned preliminary process of triggering the running quality early warning, further trend discrimination is made. Specifically as follows:

[0081] A cumulative sum sequence ending at the current day, such as 7 days, is obtained, linear trend fitting is performed on the cumulative sum sequence, the fitting slope is calculated, and a threshold, such as 0.05h / d (indicating a change rate of 0.05 units per hour per day), is set. When the fitting slope is greater than the threshold, it indicates that the change amplitude of the cumulative sum sequence in unit time exceeds the normal fluctuation range, and shows a relatively obvious upward or downward trend. In addition, the goodness of fit is calculated, and when the goodness of fit is greater than 0.6, it indicates that the linear fitting can accurately describe the trend change of the cumulative sum sequence, and the fitting result has high reliability.

[0082] Only when the fitting slope is greater than 0.05h / d and the goodness of fit is greater than 0.6 at the same time, it is determined that the cumulative sum sequence shows a continuous and stable upward or downward trend, which meets the typical characteristics of metering drift, and the quality early warning is triggered at this time. If it is only a transient peak followed by a rapid decline, it is considered as interference, and the final running quality early warning is not triggered, the normal state is maintained or it is transferred to the suspicious observation state, and further verified by subsequent data.

[0083] It should be pointed out that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the protection scope of the present application. Therefore, the protection scope of the present application patent should be subject to the appended claims.

Claims

1. A method for early warning of the operational quality of electricity meters based on a smart fusion terminal, characterized in that, include: The total electricity consumed by the user in each day of the target electricity meter is obtained as the daily electricity consumption, and all the daily electricity consumption is arranged in chronological order to construct the original sequence; the original sequence is processed to obtain the first-order difference sequence; the initial value of the cumulative sum is preset, and the improved cumulative sum of the current day is obtained by using the composite reference value calculated by the original sequence and the first-order difference sequence, and the improved decision threshold is obtained by using the composite reference value. Based on the improved cumulative sum and the improved decision threshold, an early warning of operational faults in the target energy meter for the current day is provided. The calculation process for the composite reference value includes: Calculate the load fluctuation index of the first-order difference sequence and the scale correction coefficient of the original sequence, and use the product of the load fluctuation index and the scale correction coefficient as the composite reference value.

2. The method for early warning of electricity meter operation quality based on a smart fusion terminal according to claim 1, characterized in that, The calculation process of the load fluctuation index includes: Calculate the fluctuation amplitude and directional disorder of a first-order difference sequence; The product of the amplitude and direction of the fluctuation is used as the load fluctuation index of the first-order difference sequence.

3. The method for early warning of electricity meter operation quality based on a smart fusion terminal according to claim 1, characterized in that, The calculation process for the scale correction factor includes: A minimum effective daily electricity consumption is preset. When the median of the original sequence is less than or equal to the minimum effective daily electricity consumption, the scale correction factor is set to 1. When the median of the original sequence is greater than the minimum effective daily electricity consumption, the ratio of the median of the original sequence to the minimum effective daily electricity consumption is used as the scale correction factor.

4. The method for early warning of electricity meter operation quality based on a smart fusion terminal according to claim 1, characterized in that, The process of obtaining the improved decision threshold includes: Calculate the square of the composite reference value and use it as the first term; calculate the product of the preset minimum detectable drift value and the median of the original sequence and use it as the second term; The ratio of the first term to the second term is used as the improved decision threshold.

5. The method for early warning of electricity meter operation quality based on a smart fusion terminal according to claim 1, characterized in that, The steps for obtaining the current day's improved cumulative sum include: The sum of the previous day's accumulated electricity consumption and the current day's daily electricity consumption is calculated as the third item. The composite reference value is subtracted from the third item to obtain the first calculation result. If the first calculation result is negative, the current day's accumulated electricity consumption is recorded as 0; if the first calculation result is positive or 0, the current day's accumulated electricity consumption is equal to the first calculation result.

6. The method for early warning of electricity meter operation quality based on a smart fusion terminal according to claim 2, characterized in that, The fluctuation amplitude is the median of the first-order difference sequence.

7. A method for early warning of electricity meter operation quality based on a smart fusion terminal according to claim 2, characterized in that, The process of obtaining the directional disorder includes: The symbol sequence is extracted based on the directional change of the first-order difference sequence. The frequency of each element in the symbol sequence is counted, and the normalized Shannon entropy is calculated based on the frequency. This normalized Shannon entropy is used as the directional disorder.

8. The method for early warning of electricity meter operation quality based on a smart fusion terminal according to claim 1, characterized in that, Fault warning includes: If the cumulative sum of improvements made today is greater than or equal to the decision threshold for improvement, then the target electricity meter is determined to have a positive metering drift, triggering an operational quality warning.

Images