A power distribution network line loss node screening method and system

Abstract

The application discloses a power distribution network line loss node screening method and system, relates to the technical field of power distribution network safety, and is used for solving the technical problem that real physical loss nodes cannot be accurately identified due to line loss calculation deviation. The method comprises the following steps: monitoring the change of the line loss rate of each line section in the power distribution network within a continuous time period; when the change amplitude of the line loss rate exceeds a preset threshold and the change trend does not conform to the actual load change trend, the line section is preliminarily marked as a calculation fluctuation; for the line section preliminarily marked as the calculation fluctuation, line loss data of the line section under multiple time scales are extracted, and the line loss data under the multiple time scales are used to determine a continuously high line loss line section; line loss anomaly analysis is performed on the continuously high line loss line section, and a line loss anomaly result is obtained; and according to the line loss anomaly result, the line loss anomaly conditions of each line section are classified, and different types of investigation suggestions are generated.

Description

Technical Field

[0001] This application relates to the field of power distribution network safety technology, and in particular to a method and system for screening line loss nodes in power distribution networks. Background Technology

[0002] In the daily operation of a power distribution network, accurately identifying and locating nodes with high power loss is crucial for improving power supply efficiency and reducing operating costs. Typically, a large number of smart metering devices are deployed in a power distribution network to accurately collect electrical parameters such as voltage, current, and power, along with precise time stamps. Theoretically, all devices should adhere to a unified time calibration protocol to ensure precise data alignment on the time axis, thereby enabling accurate energy balance calculations. However, the reality is far more complex than theory suggests.

[0003] Many outdoor smart metering devices experience subtle but persistent time skews in their internal timing components due to drastic changes in ambient temperature. For example, the crystal oscillator frequency may deviate during day and night or seasonal temperature variations, causing a cumulative difference between the internal clock and the network's unified time standard. Although these deviations may be small, over extended periods of operation, the accumulated error can reach several seconds or even minutes. When the central data platform receives this data, its built-in algorithms typically treat this slightly time-skewed data as sufficiently synchronized and process it normally, as these deviations do not trigger the system's preset abnormal warning threshold.

[0004] Furthermore, in order to optimize operating costs, distribution companies may extend the intervals for on-site time calibration and firmware updates of smart meters distributed throughout the distribution network. This decision results in devices that have already begun to accumulate time errors not being calibrated for extended periods, causing internal clock skew to accumulate and widen continuously, thereby affecting the data flow integrity of a significant portion of the metering infrastructure in the distribution network.

[0005] These accumulated and uncalibrated time deviations introduce pervasive and difficult-to-detect "computational noise" when calculating distribution network line losses. Because smart meters at different metering points accumulate timing deviations of varying directions and degrees, energy balance calculations based on the perfect synchronization assumption will inevitably be distorted. The magnitude and pattern of this "computational noise" are directly related to the spatial distribution and accumulation of time synchronization errors, making it exceptionally difficult to distinguish any real physical phenomena from the calculation results. This noise is often comparable to, or even greater than, the signal produced by actual physical losses (e.g., the gradual increase in resistance caused by aging connection points), leading screening methods to frequently mark the areas with the most significant data synchronization differences as "high-loss" nodes, while masking the actual physical losses. This makes it difficult to achieve the fundamental goal of reducing overall network line losses, because screening methods cannot effectively distinguish between real physical problems and pervasive data artifacts. Summary of the Invention

[0006] This application provides a method and system for screening line loss nodes in a distribution network, aiming to solve the problem that line loss calculation deviations caused by factors such as metering equipment timing drift, data quality issues, distributed power source access, topology changes, and seasonal load effects in the distribution network make it impossible to accurately identify the actual physical loss nodes.

[0007] Firstly, to address the aforementioned technical problems, this invention provides a method for screening line loss nodes in a distribution network, comprising: monitoring the changes in the line loss rate of each line segment in the distribution network over a continuous time period; when the change in the line loss rate exceeds a preset threshold and the trend of change does not conform to the actual load change trend, initially marking the line segment as a calculated fluctuation; for the line segment initially marked as a calculated fluctuation, extracting line loss data of the line segment at multiple time scales, and determining the line segment with persistently high line loss based on the line loss data at multiple time scales; the line loss data of the line segment with persistently high line loss shows a level higher than normal and a consistent trend across all time scales; performing line loss anomaly analysis on the line segment with persistently high line loss to obtain line loss anomaly results; and classifying the line loss anomaly situation of each line segment according to the line loss anomaly results, generating different types of investigation suggestions.

[0008] Secondly, this application provides a distribution network line loss node screening system, which includes: a data receiving module for receiving power data from various metering terminals in the distribution network and calculating the initial line loss of each line segment based on the power data and the energy balance principle; a multi-time-scale analysis module for extracting line loss data of the line segments initially marked as having fluctuating values ​​at multiple time scales and determining line segments with persistently high line losses based on the line loss data at multiple time scales; the line loss data of the line segments with persistently high line losses at all time scales showing a level higher than normal and a consistent trend; an anomaly analysis module for performing line loss anomaly analysis on the line segments with persistently high line losses and obtaining line loss anomaly results; and an anomaly classification and suggestion generation module for classifying the line loss anomalies of each line segment and generating different types of investigation suggestions.

[0009] This application has at least the following beneficial effects: The distribution network line loss node screening method disclosed in this application monitors the changes in the line loss rate of each line segment in the distribution network over a continuous time period. When the change in the line loss rate exceeds a preset threshold and the trend does not conform to the actual load change trend, the line segment is initially marked as a calculated fluctuation. This step can effectively identify line segments with abnormal fluctuations in line loss data, avoiding blind analysis of the entire network and significantly improving the initial efficiency of screening. Based on this, for the line segments initially marked as calculated fluctuations, their line loss data at multiple time scales are extracted, and line segments with persistently high line losses are identified based on these data. The line loss data of these persistently high line loss line segments shows higher than normal levels at all time scales and the trend remains consistent. Through multi-time scale analysis, this application can effectively distinguish between instantaneous fluctuations and persistent high losses, thereby avoiding misjudging occasional data anomalies as long-term problems and making the screening target more focused on truly existing persistently high loss nodes. For these persistently high line loss segments, this application further conducts line loss anomaly analysis to obtain line loss anomaly results. Based on these results, the line loss anomalies of each segment are classified, and different types of troubleshooting suggestions are generated. This series of steps can deeply uncover the root causes of line loss anomalies, such as distinguishing between calculation deviations caused by non-physical factors like metering equipment timing drift, data quality issues, distributed power supply access, topology changes, or seasonal load effects, and actual physical losses. Through accurate attribution, this application overcomes the difficulty in effectively distinguishing between actual physical losses and commonly existing data artifacts in existing technologies, avoiding the mislabeling of areas with the most significant data synchronization differences as "high-loss" nodes, thereby achieving the fundamental goal of reducing overall network line loss. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating a method for screening line loss nodes in a distribution network provided in this application. Detailed Implementation

[0011] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0012] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0013] Traditional methods for screening line loss nodes in distribution networks suffer from "computational noise" when processing data from smart metering devices. This noise is caused by time deviations in the timing components within the devices due to ambient temperature variations, and by extended calibration cycles by distribution companies. This accumulated time error distorts the line loss calculations based on the energy balance principle, making it difficult to distinguish between actual physical losses and data artifacts. Consequently, it fails to effectively locate high-loss nodes, hindering the achievement of the goal of reducing overall network line losses.

[0014] In view of the above problems, this application provides a method for screening high-loss nodes in a distribution network. By introducing multi-timescale analysis, line loss anomaly analysis, and anomaly classification and suggestion generation, it aims to effectively distinguish between calculation deviations and actual physical losses, thereby more accurately identifying high-loss nodes in the distribution network and improving the accuracy and efficiency of line loss screening.

[0015] The distribution network line loss node screening method and system provided in this application will be described in detail and explained through the following specific embodiments.

[0016] Reference Figure 1 This application provides a method for screening distribution network line loss nodes, which may include the following steps:

[0017] S1. Monitor the changes in line loss rate of each line segment in the distribution network over a continuous time period. When the change in line loss rate exceeds the preset threshold and the trend of change does not conform to the actual load change trend, the line segment is initially marked as a calculated fluctuation.

[0018] In this context, a distribution network refers to the power network that transmits electrical energy from the transmission system to end users, typically including substations, lines, switching equipment, and metering equipment. A line segment is a physical line in the distribution network with a defined start and end point, serving as the basic unit for line loss calculation and analysis. The line loss rate is the ratio of energy loss to input energy on a line segment within a certain time period, a crucial indicator for measuring the degree of energy loss. A continuous time period can refer to, for example, 24 hours, a week, or a month, used to observe the dynamic changes in the line loss rate. A preset threshold is a reference value set to determine whether the change in the line loss rate is abnormal; exceeding this value is considered a significant change. The actual load change trend refers to the expected load change pattern under normal operating conditions, derived from historical data and predictive models.

[0019] This application allows for various methods to monitor the changes in line loss rates of different line segments in a distribution network over a continuous time period. For example, data acquisition units can be deployed to acquire real-time voltage, current, and power data for each line segment and calculate the line loss rate based on this data. These data acquisition units can be smart meters, remote terminal units, or distribution automation terminals, etc. The calculation of the line loss rate can be based on the energy balance principle, i.e., input power minus output power. The line loss rate of each line segment is calculated and recorded over a continuous time period, for example, hourly. When the change in the line loss rate exceeds a preset threshold—for example, if the line loss rate suddenly rises from 2% to 10% in a short period, and this trend is inconsistent with the historical or predicted load change trend of that line segment—then the line segment is initially marked as experiencing a fluctuation. The preset threshold can be set as a percentage of the average line loss rate, such as twice the standard deviation, based on statistical analysis of historical data.

[0020] S2. For the line segments initially marked as having fluctuations, extract the line loss data of the line segments at multiple time scales, and determine the line segments with persistently high line loss based on the line loss data at multiple time scales.

[0021] Among these, computational fluctuations refer to abnormal changes in line loss rates, which may be caused by various factors such as measurement errors, data quality issues, or actual physical losses, requiring further analysis. Multiple time scales can be included, such as minute-level, hour-level, daily-level, weekly-level, or monthly-level. Observing line loss data at different time granularities helps to identify different types of abnormal patterns. A line segment with consistently high line loss refers to a line segment whose line loss data shows above-normal levels and a consistent trend across all analyzed time scales. This indicates that the line segment may have actual physical losses or persistent computational biases.

[0022] For example, this application can extract line loss data for the line segment over different time scales, such as the past 24 hours, 7 days, and 30 days. At each time scale, the mean, maximum, minimum, and standard deviation of the line loss data are calculated, and their trends are analyzed. If the line segment's line loss data shows above-normal levels across all analyzed time scales, and this high line loss trend remains consistent across different time scales—for example, daily, weekly, and monthly line loss rates are all consistently high, and the change curves are similar—then it is identified as a line segment with persistently high line loss. This multi-time-scale analysis helps to eliminate occasional or short-term data anomalies, focusing on persistent problems.

[0023] S3. For line sections with persistently high line loss, conduct line loss anomaly analysis to obtain line loss anomaly results.

[0024] Line loss anomaly analysis refers to in-depth technical analysis of line sections with persistently high line loss to determine the specific causes of the anomalies. Line loss anomaly results refer to the conclusions drawn regarding the causes of the line loss anomalies after the line loss anomaly analysis.

[0025] For example, this application can further analyze the metering data quality, topology changes, distributed power supply access, or environmental factors of the line segment. Specifically, it can check whether the metering equipment has any abnormalities such as missing data, jumps, or freezes; verify whether the topology of the line segment has changed, such as line reconnection or switch switching; analyze whether the output fluctuation of the distributed power supply is related to the line loss fluctuation; or analyze the impact of environmental factors such as temperature and humidity on the timing drift of the metering equipment. Through these analyses, the specific causes of persistently high line losses can be identified, such as metering equipment failure, data transmission errors, topology mismatch, distributed power supply fluctuations, or actual physical losses (such as line aging or loose connections).

[0026] S4. Based on the abnormal line loss results, classify the abnormal line loss situations of each line segment and generate different types of investigation suggestions.

[0027] Among them, troubleshooting suggestions refer to targeted solutions or troubleshooting directions provided to operations and maintenance personnel based on abnormal line loss results.

[0028] For example, if abnormal line loss results indicate a faulty metering device, it is recommended that maintenance personnel conduct on-site inspection, calibration, or replacement of the relevant metering device; if it is attributed to data transmission errors, it is recommended to check the communication link and data acquisition system; if it is attributed to topology mismatch, it is recommended to update the distribution network topology model; if it is attributed to power flow fluctuations of distributed generation, it is recommended to optimize the dispatching strategy of distributed generation; and if it is ultimately attributed to actual physical losses, it is recommended to inspect, maintain, or upgrade the line segment. This classification and recommendation generation provides clear and targeted guidance for maintenance personnel, improving the efficiency and accuracy of line loss investigation.

[0029] In summary, this application effectively solves the problems of computational noise interference in line loss calculation and the difficulty in accurately distinguishing between computational bias and actual physical loss in existing technologies by constructing a systematic line loss node screening method. Its innovative features, such as multi-timescale analysis, refined anomaly attribution, and classification suggestion generation, significantly improve the accuracy, efficiency, and practicality of distribution network line loss screening, providing strong technical support for reducing overall distribution network line losses.

[0030] In some embodiments, this application further proposes the steps of extracting line loss data of line segments initially marked as having fluctuating performance at multiple time scales, and determining line segments with persistently high line loss based on the line loss data at the multiple time scales, including: extracting line loss data of line segments initially marked as having fluctuating performance at multiple time scales; when the line loss data shows a level higher than normal at all analyzed time scales and the trend remains consistent, further extracting line loss data of the line segment over the past several consecutive days to obtain the line loss rate of the time window of the high loss phenomenon; retrieving ambient temperature data of the area where the line segment is located, and calculating the time window of the high loss phenomenon. The correlation between the line loss rate within the time window and the ambient temperature data is analyzed. Based on the line loss rate within the time window of the high-loss phenomenon, the correlation, and the ambient temperature data, the cause of the line loss anomaly is attributed. Specifically, when the line loss rate within the time window of the high-loss phenomenon reaches a preset line loss rate threshold, and the correlation is higher than the correlation threshold, while the ambient temperature data is consistently higher than a specific high-temperature threshold, the cause of the line loss anomaly is attributed to timing drift calculation deviation caused by environmental thermal stress. Otherwise, the cause of the line loss anomaly is attributed to actual physical loss. The line segment for which the cause of the line loss anomaly is attributed to actual physical loss is identified as the line segment with persistently high line loss.

[0031] Specifically, line loss data for line segments initially marked for fluctuation calculation is extracted across multiple time scales. The aim is to comprehensively analyze the line loss performance of these segments and identify their persistently high loss characteristics. These multiple time scales may include, but are not limited to, hours, days, weeks, and months. By observing line loss data at different time granularities, the persistence and stability of line loss anomalies can be determined more accurately.

[0032] Specifically, when the line loss data shows above-normal levels across all analyzed time scales and the trend remains consistent, it indicates that the line segment may have a persistent high line loss problem. Based on this, line loss data for the line segment over several consecutive days is further extracted to obtain the line loss rate within the time window of the high-loss phenomenon. The purpose is to focus on the specific time period when the anomaly occurred, providing a more refined data foundation for subsequent attribution analysis.

[0033] In practical applications, ambient temperature data for the area where the line segment is located is retrieved, and the correlation between the line loss rate and the ambient temperature data within the time window of the high-loss phenomenon is calculated. The purpose is to explore the potential link between abnormal line loss and environmental factors (especially temperature). Significant changes in ambient temperature, especially sustained high temperatures, may cause thermal expansion or performance drift of internal components in metering equipment (such as electricity meters), thereby affecting their timing accuracy or measurement accuracy, ultimately manifesting as deviations in the calculated line loss value. The correlation can be calculated using statistical methods such as the Pearson correlation coefficient to quantify the strength of the linear relationship between the line loss rate and ambient temperature.

[0034] Furthermore, the cause of the abnormal line loss is attributed based on the line loss rate within the time window of the high-loss phenomenon, the correlation, and the ambient temperature data. Specifically, when the following three conditions are met—that is, the line loss rate within the time window of the high-loss phenomenon reaches a preset line loss rate threshold, the correlation between the line loss rate and the ambient temperature data is higher than a preset correlation threshold, and the ambient temperature data is consistently higher than a preset specific high-temperature threshold—then the cause of the abnormal line loss is attributed to timing drift calculation deviation caused by environmental thermal stress. The preset line loss rate threshold, correlation threshold, and specific high-temperature threshold can be set based on historical data analysis, equipment specifications, or expert experience. For example, the preset line loss rate threshold can be set to a certain percentage, the correlation threshold can be set to 0.7 or 0.8 or higher, and the specific high-temperature threshold can be set to 35°C or 40°C. If the above conditions are not met, the cause of the abnormal line loss is attributed to actual physical losses, such as line aging, loose connections, and insulation deterioration.

[0035] Therefore, the line segments for which the abnormal line loss is attributed to actual physical losses are identified as the persistently high line loss line segments. This attribution process allows subsequent troubleshooting and maintenance work to more accurately focus on line segments with actual physical defects, avoiding ineffective troubleshooting due to calculation errors.

[0036] In some embodiments, this application further proposes the following steps for performing line loss anomaly analysis on line segments with persistently high line losses: continuously monitoring real-time current and power data of the line segment with persistently high line losses; marking a significant load change event when the average rate of change of the real-time current or the real-time power exceeds a preset rate of change threshold; extracting the calculated line loss value, average current value, and line loss power value before, during, and after the significant load change event; performing a line loss positive value verification to obtain a line loss positive value verification result; the line loss positive value verification is used to check whether the extracted calculated line loss value is always positive; performing a square relationship response verification to obtain a square relationship. The response verification results are as follows: The square relationship response verification is used to verify whether the change in the extracted line loss calculation value is consistent with the square change in the average current value; an asymmetric response verification is performed to obtain the asymmetric response verification results; The asymmetric response verification is used to identify a pair of symmetrical load change events that occur in a short period of time, where the load first increases and then decreases or the load first decreases and then increases, and to verify whether the increase and decrease in the line loss power value in the symmetrical load change event both show significant asymmetry; Based on the positive line loss verification results, the square relationship response verification results, and the asymmetric response verification results, the abnormal line loss results are determined.

[0037] Specifically, when performing line loss anomaly analysis on line sections with persistently high line losses, it is first necessary to continuously monitor the real-time current and power data of that line section. Real-time current and power data are key indicators reflecting the line's operating status, and their changes directly indicate load fluctuations. When the average rate of change of real-time current or real-time power exceeds a preset rate of change threshold, it indicates a significant change in the line load, which is then marked as a significant load change event. This preset rate of change threshold can be set based on the operating characteristics and empirical values ​​of the distribution network to ensure that load fluctuations that may affect line loss calculations are captured.

[0038] Furthermore, for the marked significant load change events, it is necessary to extract the calculated line loss, average current, and line loss power values ​​before, during, and after the event. This data set provides the foundation for subsequent refined verification. The calculated line loss reflects the energy loss of the line, while the average current and line loss power values ​​provide specific quantitative information about the load change.

[0039] Based on this, this application introduces three key verification mechanisms: positive line loss verification, square relationship response verification, and asymmetric response verification.

[0040] The line loss positive value check is used to verify whether the extracted line loss calculation value is always positive. Physically speaking, line loss, as energy loss, should always be positive. A negative value strongly indicates a problem such as metering error, abnormal data acquisition, or a flawed calculation model, rather than actual physical loss.

[0041] Furthermore, the square relation response check is used to verify whether the change in the extracted calculated line loss value is consistent with the square change in the average current value. According to Joule's law, line loss power is proportional to the square of the current (P... loss =I 2 *R). Therefore, the actual physical loss should strictly follow this physical law. By checking the degree of agreement between the calculated line loss value and the square of the average current value, anomalies that do not conform to the physical law can be effectively identified, such as artificially high line losses caused by metering equipment failure or data transmission errors.

[0042] Furthermore, asymmetric response verification is used to identify symmetrical load change events occurring within a short period, where the load first increases and then decreases, or vice versa. It verifies whether the increase and decrease in line loss power values ​​during these symmetrical load change events both exhibit significant asymmetry. Ideally, for symmetrical load changes, the increase and decrease in line loss power should also exhibit some degree of symmetry. Significant asymmetry in line loss power values ​​may indicate anomalies such as nonlinear losses, delayed metering equipment response, or electricity theft.

[0043] Finally, based on the positive line loss verification results, the squared response verification results, and the asymmetric response verification results, the abnormal line loss results are comprehensively determined. The combined analysis of these verification results provides more comprehensive and accurate anomaly diagnostic information.

[0044] Through the above technical solutions, this application can significantly improve the accuracy and diagnostic depth of distribution network line loss node screening. Traditional line loss anomaly analysis often relies on simple threshold judgments or trend analysis, which is difficult to effectively distinguish between true physical losses and false anomalies caused by metering errors, data quality problems, or system transient changes. The positive value verification, square relationship response verification, and asymmetric response verification introduced in this application verify line loss data in a multi-dimensional and in-depth manner based on physical laws and operating characteristics.

[0045] In some embodiments, the above-mentioned extraction of line loss data for line segments initially marked as having fluctuating performance at multiple time scales, and the determination of line segments with persistently high line loss based on the line loss data at multiple time scales, includes the following steps: extracting line loss data for line segments initially marked as having fluctuating performance at multiple time scales, and analyzing the mean, standard deviation, and trend of the line loss data; when the line loss data shows above-normal levels at all analyzed time scales and the trend remains consistent, analyzing whether there are intermittent, non-random data gaps or anomalous mutations in the original measurement data of the line segment; performing pattern analysis on the data gaps or anomalous mutations in the original measurement data to obtain pattern analysis results. The pattern analysis is used to determine whether the missing or abnormal data changes have a temporal regularity or are related to specific events; based on the average value, standard deviation, and trend of the line loss data, the pseudo-persistence characteristics of the line loss trend are determined; based on the pattern analysis results and the pseudo-persistence characteristics of the line loss trend, the causes of the line loss anomalies are attributed, wherein when the missing or abnormal data changes have a specific regularity and highly match the pseudo-persistence trend, the cause of the line loss anomalies is attributed to calculation deviations caused by data quality issues; otherwise, the cause of the line loss anomalies is attributed to actual physical losses; the line segments for which the cause of the line loss anomalies is attributed to actual physical losses are identified as the line segments with persistently high line losses.

[0046] Multiple time scales can be used, including but not limited to hours, days, weeks, and months. By analyzing data at different time scales, a comprehensive understanding of the behavioral patterns of line loss data can be achieved. The average value is used to assess the overall level of line loss, the standard deviation measures the volatility of line loss, and the trend reveals the pattern of line loss evolution over time. Intermittent data loss refers to the irregular loss of data at certain points or time periods. Non-randomness means that this loss is not accidental and may be related to specific events or equipment status. Abnormal mutations refer to drastic changes in measurement data that far exceed the normal range within a short period, such as metering equipment failure, communication interruption, or data transmission errors. Pattern analysis can analyze whether data loss occurs at fixed times each day or whether it coincides with events such as maintenance of a specific piece of equipment or communication interruption. Pseudo-persistent characteristics refer to line loss data that superficially exhibits a continuous high loss trend, but this is not caused by actual physical loss; rather, it is a computational illusion caused by data quality issues such as data loss and abnormal mutations. For example, if a large amount of data is missing within a certain time period, the system may default to filling in zero values ​​or values ​​from the previous moment, resulting in falsely high values ​​or stable trends in the line loss calculation results.

[0047] Through the above technical solution, this application can significantly improve the accuracy and reliability of screening for line loss nodes in distribution networks. By conducting in-depth analysis and attribution of the quality of the original metering data, it effectively eliminates line loss calculation deviations caused by data quality issues such as missing data or abnormal mutations, avoiding the misjudgment of pseudo-persistent high losses as actual physical losses. This makes the identified persistently high line loss line segments more accurately reflect the actual physical loss problems in the distribution network, thus enabling more precise location of abnormal line loss nodes requiring investigation. Furthermore, this solution helps optimize resource allocation, concentrating limited investigation resources on line segments with actual physical losses, avoiding ineffective investigations of falsely high losses caused by data problems, and improving the efficiency and targeting of line loss management.

[0048] In some embodiments, this application further proposes a method for extracting line loss data of line segments initially marked as having fluctuating performance at multiple time scales, and determining line segments with persistently high line loss based on the line loss data at the multiple time scales. Specifically, this method includes: extracting line loss data of the line segments initially marked as having fluctuating performance at multiple time scales, and analyzing the average value, standard deviation, and trend of the line loss data; when the line loss data shows a level higher than normal at all analyzed time scales and the trend remains consistent, acquiring real-time distributed power output data and real-time load data of the line segment; calculating the volatility of the distributed power output data and the volatility of the load data of the line segment at different time scales; when the volatility of the distributed power output data and the volatility of the load data are highly correlated with a significant fluctuation trend at the same time scale, attributing the abnormal line loss to power flow fluctuations caused by distributed power access; otherwise, attributing the abnormal line loss to actual physical losses; and identifying line segments whose abnormal line loss is attributed to actual physical losses as the persistently high line loss line segments.

[0049] Specifically, when analyzing line segments initially marked for fluctuation calculation, the first step is to extract line loss data for these segments across multiple time scales, such as hourly, daily, and weekly. Subsequently, statistical analysis is performed on this line loss data, including calculating its mean and standard deviation, and observing its trends. This step aims to initially identify line segments that exhibit abnormally high line losses across different time dimensions with stable trends.

[0050] When a line segment is identified as exhibiting above-normal levels across all analyzed time scales with a consistent trend, real-time distributed generation (DG) output data and real-time load data for that segment are needed to further differentiate the true causes of the abnormal line losses. Real-time DG output data refers to the actual power generation or output of distributed power sources (e.g., solar, wind) connected to the line segment at a specific point in time; real-time load data refers to the actual electricity load carried by the line segment. This data can be obtained from the distribution network's SCADA system, metering system, or distributed power source monitoring system.

[0051] Based on this, the volatility of distributed generation output data and load data for the aforementioned line segment are calculated at different time scales. Volatility can be understood as the degree of change or instability of data within a certain time range, and can be quantified, for example, by the ratio of standard deviation to mean, or the difference between maximum and minimum values. By calculating the volatility of these two types of data, the impact of distributed generation output and load changes on the power flow stability of the line can be assessed.

[0052] Furthermore, when the volatility of the distributed power source output data and the volatility of the load data are highly correlated on the same time scale, it means that the output changes of the distributed power source and the changes in line load are closely synchronized or countersynchronized, which may lead to frequent fluctuations in line power flow. The process of determining whether there is a high correlation between the volatility of the distributed power source output data and the volatility of the load data on the same time scale can be done by calculating the statistical correlation (e.g., Pearson correlation coefficient) between these two volatility values ​​on the same time scale and comparing it with a preset correlation threshold. This power flow fluctuation may cause deviations in the line loss calculation model, misclassifying non-physical losses as high line losses. In this case, the cause of the abnormal line loss will be attributed to power flow fluctuations caused by the distributed power source connection, rather than actual physical losses. Conversely, if there is no significant correlation between the volatility of the distributed power source output data and the volatility of the load data, or the correlation is low, it indicates that the distributed power source has a small impact on the line loss calculation deviation, and the cause of the abnormal line loss will be attributed to actual physical losses.

[0053] Ultimately, only those line segments whose abnormal line loss is attributed to actual physical losses will be identified as the persistently high line loss line segments, so that targeted line loss anomaly analysis and investigation can be carried out subsequently.

[0054] Through the above technical solution, this application can more accurately identify the true causes of persistently high line loss sections in distribution networks. Especially in modern distribution networks with widespread distributed generation, this solution can effectively distinguish between calculation deviations caused by power flow fluctuations resulting from distributed generation and actual physical losses. This avoids misjudging calculation deviations as physical losses, thereby improving the accuracy of line loss node screening and reducing unnecessary on-site investigation workload and resource waste. Furthermore, more accurate attribution can provide more targeted decision support for the operation and maintenance of the distribution network. For example, for line loss anomalies attributed to power flow fluctuations, optimizing the dispatching strategy of distributed generation or improving the metering system can be considered; for line sections attributed to actual physical losses, equipment inspection and maintenance can be prioritized, thereby improving the overall operating efficiency and reliability of the distribution network.

[0055] In some embodiments, the above-mentioned extraction of line loss data for line segments initially marked as having fluctuating performance at multiple time scales, and the determination of line segments with persistently high line losses based on the line loss data at multiple time scales, includes the following steps: extracting line loss data for the line segments initially marked as having fluctuating performance at multiple time scales, and analyzing the average value, standard deviation, and trend of the line loss data. Multiple time scales may include, but are not limited to, hours, days, weeks, and months. Statistical analysis of line loss data at different time scales allows for a comprehensive understanding of the fluctuation characteristics and long-term trends of line losses. When the line loss data shows above-normal levels at all analyzed time scales and the trend remains consistent, historical topology data and real-time topology change event records for the line segment are obtained. The historical topology data can be stored in a distribution network geographic information system (GIS) or topology management system, recording the connection relationships and equipment configurations of the line segments at different historical points in time. The real-time topology change event records include events such as switching operations, line tripping, and reclosing that cause changes in the network topology and their occurrence times. Based on the historical topology data, the topology configuration of the line segment in different time periods is identified. For example, by querying the historical topology database, it can be determined whether the line segment was operating as a primary supply line, a backup line, or whether the number and type of its connected load points changed during a specific time period. Based on the real-time topology change event records, the line loss data is segmented to ensure that the line topology remains consistent within each segment. This means that when a topology change is detected, the line loss data stream is segmented at the point of change, ensuring that the line loss calculation within each data segment is based on a stable and unchanging topology. The average, standard deviation, and trend analysis of the segmented line loss data with consistent topology are performed at multiple time scales to determine whether there are significant discontinuous abrupt changes in the line loss data at the point of topology change. This step aims to check whether the line loss data exhibits sudden jumps or trend interruptions that are inconsistent with normal patterns when the topology changes; such abrupt changes are often a direct manifestation of the change in the calculation benchmark caused by the topology change. When a significant discontinuous abrupt change occurs in the line loss data at the point of topology change, the cause of the line loss anomaly is attributed to a calculation deviation due to the change in line topology; otherwise, the cause is attributed to the actual physical loss. This attribution logic is based on the premise that if the abnormal performance of line loss closely matches the topology change in time and exhibits a discontinuous abrupt change, it is highly likely due to changes in the calculation model or parameters caused by the topology change, rather than an increase in the physical loss of the line itself.

[0056] Finally, the line segments whose abnormal line loss is attributed to actual physical loss are identified as the persistently high line loss line segments. Only line segments that still show persistently high line loss after eliminating topology change factors are considered to have truly abnormal physical loss and require further investigation.

[0057] Through the above technical solution, this application can significantly improve the accuracy and efficiency of screening for line loss nodes in distribution networks. Specifically, by considering changes in line topology as an important factor affecting line loss calculation, this application can effectively distinguish between calculation deviations caused by topology changes and actual physical losses. This avoids misjudging fluctuations in line loss data caused by topology reconfiguration, load transfer, and other operations as abnormal physical losses in the line itself, thereby reducing unnecessary on-site investigation workload and resource waste. In addition, by more accurately identifying line segments with actual physical loss anomalies, this application enables subsequent line loss anomaly analysis and troubleshooting recommendations to be more focused on actual problems, improving the accuracy and efficiency of fault location and resolution, which is of great significance for improving the lean management level of distribution networks.

[0058] In some embodiments, this application further proposes the above-mentioned method of extracting line loss data of line segments initially marked as having fluctuating performance at multiple time scales, and determining line segments with persistently high line loss based on the line loss data at multiple time scales, including: extracting line loss data of the line segments initially marked as having fluctuating performance at multiple time scales, and analyzing the mean, standard deviation, and trend of the line loss data; when the line loss data shows above-normal levels at all analyzed time scales and the trend remains consistent, acquiring historical load data and calendar information of the line segment; identifying typical load patterns of the line segment during different seasons and holidays based on the historical load data and the calendar information; calculating the baseline value and fluctuation range of line loss of the line segment during different seasons and holidays based on the typical load patterns; and calculating the baseline value and fluctuation range of line loss based on the baseline value and the fluctuation range. The line loss data is corrected for seasonal load variations and holiday effects to eliminate or reduce the periodic or quasi-periodic characteristics caused by these variations. The corrected line loss data is then re-analyzed for averages, standard deviations, and trends across multiple time scales. The cause of the line loss anomaly is attributed based on whether the corrected line loss data still shows above-normal levels and a consistent trend across all analyzed time scales. If the corrected line loss data still shows above-normal levels and a consistent trend across all analyzed time scales, the anomaly is attributed to actual physical losses; otherwise, the anomaly is attributed to calculation bias caused by seasonal load variations or holiday effects. Line segments where the anomaly is attributed to actual physical losses are identified as the persistently high line loss line segments.

[0059] Specifically, when processing line segments initially marked for fluctuation calculations, it is first necessary to extract line loss data for these segments across multiple time scales and perform preliminary analysis on this data, including calculating their average value, standard deviation, and observing their trends. These time scales can cover hours, days, weeks, months, etc., to comprehensively understand the behavioral patterns of the line loss data. When the preliminary analysis shows that the line segment exhibits line loss data above normal levels across all analyzed time scales and that the trend remains consistent, further historical load data and calendar information for the line segment are needed to eliminate the influence of seasonal load variations and holiday effects. Historical load data can include hourly and daily load records from the past few years or even longer, while calendar information is used to identify specific seasonal divisions (such as spring, summer, autumn, and winter) and holidays (such as New Year's Day, Spring Festival, and National Day).

[0060] Based on the acquired historical load data and calendar information, typical load patterns of the line segment during different seasons and holidays can be identified. For example, load levels are generally high in summer due to increased air conditioning load; heating load may lead to load peaks in winter; and during holidays, commercial load may decrease while residential load may increase. By analyzing these typical load patterns, the baseline value and fluctuation range of line loss for the line segment during different seasons and holidays can be calculated. The baseline value of line loss represents the expected line loss level for a specific season or holiday under normal operating conditions, while the fluctuation range reflects the normal allowable fluctuation range around the baseline value.

[0061] Subsequently, using the calculated baseline and fluctuation range of line loss, the original line loss data is corrected for seasonal load variations and holiday effects. The purpose of this correction is to eliminate or mitigate the biases in line loss calculations caused by these periodic or quasi-periodic characteristics. For example, the original line loss data can be compared with the baseline values ​​for the corresponding season or holiday, and adjustments can be made accordingly, so that the corrected line loss data better reflects the true line loss level after excluding the effects of seasonality / holidays. The corrected line loss data then requires a new analysis of the mean, standard deviation, and trend over multiple time scales. This step aims to assess whether the line loss data still shows a consistently high level and trend after excluding seasonal load variations and holiday effects.

[0062] Finally, the cause of the line loss anomaly is attributed based on whether the corrected line loss data still shows above-normal levels and a consistent trend across all analyzed time scales. Specifically, if the corrected line loss data still shows above-normal levels and a consistent trend across all analyzed time scales, the cause of the line loss anomaly can be more reliably attributed to actual physical losses, such as line aging, equipment failure, or electricity theft. Conversely, if the corrected line loss data no longer shows a consistently high level or a consistent trend, it indicates that the original "constantly high line loss" phenomenon was mainly due to calculation errors caused by seasonal load changes or holiday effects. The line segments ultimately attributed to actual physical losses are identified as persistently high line loss line segments for subsequent focused investigation.

[0063] Through the above technical solution, this application can significantly improve the accuracy and efficiency of screening line loss nodes in distribution networks. In existing technologies, the failure to fully consider the impact of seasonal load changes and holiday effects on line loss calculations may lead to misjudging normal fluctuations caused by these periodic factors as sustained high line losses, resulting in wasted resources and reduced diagnostic efficiency. This application introduces historical load data and calendar information, and based on this, identifies typical load patterns, calculates baseline line loss values ​​and fluctuation ranges, and then corrects the line loss data, effectively eliminating or reducing the periodic or quasi-periodic characteristics caused by seasonal load changes and holiday effects. This correction mechanism makes subsequent line loss anomaly attribution more accurate, effectively distinguishing between actual physical losses and false high losses caused by calculation errors. Therefore, this application avoids misjudging line segments with normal load fluctuations and unnecessary on-site investigations, allowing distribution network maintenance personnel to focus their efforts on line segments with genuine physical defects or anomalies, thereby improving the accuracy of line loss anomaly location, reducing operating costs, and enhancing the overall operational reliability of the distribution network.

[0064] In some embodiments, this application further proposes an optimized method for determining line segments with persistently high line loss. By introducing calibration records of metering equipment and historical drift data of sensors, it aims to more accurately identify the true cause of line loss anomalies and distinguish between calculation deviations caused by metering equipment problems and actual physical losses.

[0065] The method for identifying persistently high line loss sections includes: extracting line loss data of the initially marked line sections for calculating fluctuations at multiple time scales, and analyzing the average value, standard deviation, and trend of the line loss data; when the line loss data shows above-normal levels at all analyzed time scales and the trend remains consistent, acquiring calibration records and historical sensor drift data of the associated metering equipment of the line section; identifying the most recent calibration time point of the metering equipment based on the calibration records, and segmenting the line loss data before and after the calibration time point; calculating the line loss data within each segment, in conjunction with the historical sensor drift data. The cumulative deviation or abrupt change at different time scales; based on the degree of temporal agreement between the sustained high level and consistent trend of the line loss data and the cumulative deviation or abrupt change, the cause of the line loss anomaly is attributed. Specifically, when the sustained high level and consistent trend of the line loss data highly agrees with the cumulative deviation or abrupt change over time, the cause of the line loss anomaly is attributed to calculation deviation caused by sensor drift or calibration error of the metering equipment; otherwise, the cause of the line loss anomaly is attributed to actual physical loss. The line segment for which the cause of the line loss anomaly is attributed to actual physical loss is identified as the line segment with sustained high line loss.

[0066] Specifically, after extracting and analyzing line loss data for the initially marked line segments to calculate fluctuations, if the line segment shows above-normal levels and a consistent trend across all analyzed time scales, it is necessary to further obtain calibration records and historical sensor drift data of the metering equipment associated with the line segment. The calibration records provide an accuracy benchmark for the metering equipment at a specific point in time, while the historical sensor drift data records the potential performance degradation or measurement deviation patterns that may occur in the sensors during long-term operation.

[0067] Furthermore, based on the obtained calibration records, the time point of the most recent calibration of the measuring equipment can be identified. Using this calibration time point as a boundary, the line loss data is segmented to isolate and analyze data under different calibration conditions, thereby avoiding interference from inconsistencies in the data before and after calibration.

[0068] Based on this, for the line loss data within each segment, combined with historical sensor drift data, the cumulative deviation or step change at different time scales is calculated. Cumulative deviation refers to the error in sensor performance that gradually accumulates over time, typically manifesting as a slow deviation in the measured value; step change may correspond to a sudden change in the measurement reference caused by events such as calibration, maintenance, or sensor replacement. By calculating these deviations or changes, the measurement error that the measuring equipment itself may introduce can be quantified.

[0069] Finally, the cause of the line loss anomaly is attributed by comparing the degree of temporal agreement between the sustained high level and consistent trend of the line loss data and the calculated cumulative deviation or step change. Specifically, when the observed sustained high level and consistent trend of the line loss data highly coincides temporally with the calculation deviation caused by sensor drift or calibration error of the metering equipment—for example, the increasing trend of the line loss rate is consistent with the cumulative trend of sensor drift, or the abrupt change in the line loss rate coincides with the time point of equipment calibration or replacement—then the cause of the line loss anomaly is attributed to the calculation deviation caused by sensor drift or calibration error of the metering equipment. Conversely, if the sustained high level and consistent trend of the line loss data does not coincide with a problem with the metering equipment, the cause of the line loss anomaly is attributed to actual physical loss. Thus, the line segment attributable to actual physical loss can be identified as the line segment with sustained high line loss.

[0070] Through the above technical solution, this application can significantly improve the accuracy and reliability of screening for line loss nodes in distribution networks. By systematically considering the calibration status of metering equipment and sensor drift effects, this solution effectively avoids misjudging calculation deviations caused by problems with the metering equipment itself (such as sensor drift or calibration errors) as actual physical losses. This reduces ineffective investigations of non-physical losses, saves human and material resources, and ensures that the investigation work can more accurately focus on line sections with actual physical losses. This not only improves the accuracy of attributing abnormal line losses but also provides a more solid data foundation for the refined management and fault location of distribution networks, thereby optimizing the overall efficiency of line loss management.

[0071] In some preferred embodiments, a specific example is given below. Suppose that line segment A in a distribution network shows a line loss rate higher than normal on multiple time scales such as daily, weekly, and monthly for several consecutive weeks, and the trend is consistent, and it is initially marked as a line segment with persistently high line loss.

[0072] To further refine the attribution, the system first retrieved the calibration records of the metering equipment associated with line segment A. The records showed that the metering equipment had been calibrated six months prior. Based on this, the system segmented the line loss data before and after six months. Simultaneously, the system retrieved the historical drift data of the metering equipment's sensors, which showed that its current sensor exhibited a cumulative positive drift of approximately 0.1% per month.

[0073] The system analyzes the calibrated line loss data in segments and, combined with historical sensor drift data, calculates that the cumulative deviation of the line loss calculation due to sensor drift over the past six months is approximately 0.6%. Comparison reveals that the consistently high line loss rate observed in line segment A closely matches this 0.6% cumulative deviation in time; that is, the increase in the line loss rate is essentially consistent with the magnitude of the calculation deviation caused by sensor drift.

[0074] Based on this analysis, the system attributes the abnormal line loss in line segment A to calculation errors caused by sensor drift in the metering equipment, rather than actual physical losses. Therefore, line segment A will not be identified as a line segment with consistently high line loss, thus avoiding unnecessary physical investigation of this line segment and concentrating resources on other line segments that are more likely to have actual physical losses.

[0075] This application also discloses a distribution network line loss node screening system, which includes: a data receiving module for receiving power data from various metering terminals in the distribution network and calculating the initial line loss of each line segment based on the power data and the energy balance principle; a multi-time-scale analysis module for extracting line loss data of the line segments initially marked as having fluctuating values ​​at multiple time scales and determining line segments with persistently high line losses based on the line loss data at the multiple time scales; the line loss data of the line segments with persistently high line losses at all time scales showing a level higher than normal and a consistent trend; an anomaly analysis module for performing line loss anomaly analysis on the line segments with persistently high line losses and obtaining line loss anomaly results; and an anomaly classification and suggestion generation module for classifying the line loss anomalies of each line segment and generating different types of investigation suggestions.

[0076] The distribution network line loss node screening system proposed in this application aims to systematically solve the "computational noise" problem in distribution network line loss screening through modular design. The system first acquires and processes raw power data through a data receiving module to calculate initial line losses. Subsequently, a multi-timescale analysis module conducts in-depth analysis of line segments initially marked as having fluctuating calculations to identify line segments with persistently high line losses. Next, an anomaly analysis module performs detailed anomaly attribution for these persistently high line loss segments. Finally, an anomaly classification and suggestion generation module provides targeted troubleshooting suggestions based on the analysis results. Through the collaborative work of these modules, this system can effectively distinguish between calculation artifacts caused by non-physical factors such as metering equipment timing deviations and data quality issues, and actual physical losses, thereby improving the accuracy and efficiency of line loss screening.

[0077] To better understand the distribution network line loss node screening system proposed in this application, some key modules involved are described in detail below.

[0078] The data receiving module's main function is to receive electricity data from various metering terminals in the distribution network and calculate the initial line loss of each line segment based on the electricity data and the energy balance principle. Specifically, the data receiving module can be configured to acquire real-time voltage, current, power, and electricity data from metering terminals such as smart meters, remote terminal units (RTUs), or distribution automation terminals (FTUs) via various communication interfaces, such as fiber optics, wireless networks (e.g., GPRS / 4G / 5G), or power line carrier communication (PLC). This data can be received in batch file, streaming data, or database records. After receiving the electricity data, the data receiving module compares the input and output electricity of each line segment based on the energy balance principle to calculate the initial line loss of that line segment within a specific time period. For example, the data source and data format can be manually configured, or data can be interfaced through a preset API interface. In some implementations, the data receiving module can be set to periodically extract historical electricity data from a data warehouse for batch processing.

[0079] The multi-timescale analysis module is used to extract line loss data for line segments initially marked for fluctuation calculation at multiple time scales, and to identify line segments with persistently high line loss based on the line loss data at these multiple time scales. The analysis process and specific implementation of this module have been described in the above embodiments and will not be repeated here. It is important to emphasize that, as a core component of this system, this multi-timescale analysis module observes line loss data at different time granularities such as minute, hour, day, week, or month, and determines whether it shows above-normal levels and a consistent trend across all time scales. This effectively filters out occasional or short-term data anomalies, focusing on persistent problems.

[0080] The anomaly analysis module is used to perform line loss anomaly analysis on the persistently high line loss section and obtain the line loss anomaly results. The analysis process and specific implementation method performed by this module have been described in the above embodiments and will not be repeated here. It should be emphasized that, as a key functional module of this system, this anomaly analysis module, by comprehensively considering various potential causes of anomalies, such as timing drift of metering equipment, data quality problems, topology changes, distributed power supply access, etc., attributes the line loss anomalies to their causes. This allows it to accurately isolate the calculation deviations caused by various non-physical factors from complex line loss data and focus on the actual physical losses.

[0081] The anomaly classification and suggestion generation module is used to classify the abnormal line losses of each line segment and generate different types of troubleshooting suggestions. The classification and suggestion generation process and specific implementation methods of this module have been described in the above embodiments and will not be repeated here. It is important to emphasize that, as the final output of this system, this anomaly classification and suggestion generation module transforms abstract analysis results into actionable guidance schemes. For example, for line segments with metering equipment timing drift, it suggests equipment calibration; for line segments with data quality problems, it suggests checking the communication link and data acquisition system; and for line segments with actual physical losses, it suggests on-site inspection and maintenance. Through this classification and suggestion generation, this system can provide distribution network operation and maintenance personnel with clear and targeted action guidelines, greatly improving the efficiency and accuracy of line loss troubleshooting.

[0082] The distribution network line loss node screening system proposed in this application demonstrates significant technological progress and innovation in solving the "computational noise" problem in distribution network line loss screening compared to existing technologies. Traditional methods often struggle to distinguish between computational artifacts caused by non-physical factors such as timing deviations of metering equipment and data quality issues, and actual physical losses. This leads to low accuracy in screening results and may even incorrectly label areas with the most significant data synchronization differences as high-loss nodes.

[0083] The core innovation of this application lies in its modular system architecture and refined line loss anomaly attribution mechanism. The data receiving module performs preliminary processing and line loss calculation on raw power data, laying the foundation for subsequent analysis. More importantly, the collaborative work of the multi-timescale analysis module, anomaly analysis module, and anomaly classification and suggestion generation module enables the system to effectively distinguish between "computational noise" caused by various non-physical factors and actual physical losses. For example, through multi-timescale analysis, the system can filter out occasional or short-term line loss fluctuations, focusing on line segments that exhibit consistently high line loss characteristics at different time granularities. Subsequently, the anomaly analysis module can perform in-depth anomaly attribution for these consistently high line loss line segments, identifying specific causes such as timing drift, data quality issues, topology changes, or distributed power supply integration. This refined attribution capability is lacking in existing technologies. Finally, the anomaly classification and suggestion generation module transforms complex analysis results into specific and actionable operation and maintenance guidance, greatly improving the efficiency and targeting of line loss investigation and avoiding the blind investigation and resource waste that may occur in traditional methods.

[0084] In some embodiments described above in this application, the multi-timescale analysis module primarily relies on the fact that line loss data shows above-normal levels and a consistent trend across all analyzed timescales when identifying line loss sections with persistently high losses. However, in practical applications, line loss data for certain sections may experience timing drift in metering equipment due to environmental factors (such as high temperatures), leading to calculation errors and artificially inflated line loss rates that do not reflect actual physical losses. Failure to differentiate between these factors could result in resources being incorrectly allocated to investigating non-physical loss issues, reducing screening efficiency and accuracy.

[0085] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A power distribution network line loss node screening method, characterized in that, The method comprises the following steps: monitoring the change of line loss rate of each line section in the power distribution network in a continuous time period, and preliminarily marking the line section as a calculation fluctuation when the change amplitude of the line loss rate exceeds a preset threshold and the change trend does not conform to the actual load change trend; extracting line loss data of the line section preliminarily marked as the calculation fluctuation in multiple time scales, and determining a continuous high line loss line section according to the line loss data in the multiple time scales; the continuous high line loss line section has line loss data in all time scales that are all higher than a normal level and have consistent trends; continuously monitoring real-time current and power data of the line section with continuous high line loss; marking as a significant load change event when the average change rate of the real-time current or the real-time power exceeds a preset change rate threshold; extracting line loss calculation values, average current values and line loss power values before, during and after the significant load change event; performing line loss positive value verification to obtain a line loss positive value verification result; the line loss positive value verification is used to check whether the extracted line loss calculation values are always positive values; performing square relationship response verification to obtain a square relationship response verification result; the square relationship response verification is used to check whether the change of the extracted line loss calculation values is consistent with the square change of the average current values; performing asymmetric response verification to obtain an asymmetric response verification result; the asymmetric response verification is used to identify a symmetric load change event in which a pair of loads increases first and then decreases or decreases first and then increases in a short time, and to check whether the increase amplitude and the decrease amplitude of the line loss power values in the symmetric load change event both show significant asymmetry; determining a line loss abnormality result according to the line loss positive value verification result, the square relationship response verification result and the asymmetric response verification result; classifying line loss abnormality conditions of each line section according to the line loss abnormality result, and generating different types of troubleshooting suggestions.

2. The power distribution network line loss node screening method of claim 1, wherein, The method of extracting line loss data of the line section preliminarily marked as the calculation fluctuation in multiple time scales, and determining a continuous high line loss line section according to the line loss data in the multiple time scales, comprises the following steps: extracting line loss data of the line section preliminarily marked as the calculation fluctuation in multiple time scales; when the line loss data in all analyzed time scales are all higher than a normal level and the trends are consistent, further extracting line loss data of the line section in a past continuous plurality of days to obtain line loss rates in a time window of high loss phenomenon; calling environmental temperature data of a region where the line section is located, and calculating the correlation between the line loss rates in the time window of high loss phenomenon and the environmental temperature data; attributing the line loss abnormality cause to a calculation deviation caused by a time drift of environmental thermal stress when the line loss rate in the time window of the high loss phenomenon reaches a line loss rate threshold, the correlation is higher than a correlation threshold, and the environmental temperature data continuously is higher than a specific high temperature threshold; otherwise, attributing the line loss abnormality cause to a real physical loss; attributing the line loss abnormality cause to a real physical loss to determine the line section as the continuously high line loss section.

3. The method for power distribution network line loss node screening according to claim 1, characterized in that, The method for determining the continuously high line loss section by extracting line loss data of the line section at multiple time scales and determining the continuously high line loss section according to the line loss data at the multiple time scales, comprises: extracting line loss data of the line section preliminarily marked as a calculation fluctuation at multiple time scales, and analyzing average value, standard deviation and variation trend of the line loss data; when the line section shows higher than normal level at all analyzed time scales and the trend remains consistent, analyzing whether there is intermittent, non-random data loss or abnormal mutation in original metering data of the line section; performing pattern analysis on the identified data loss or abnormal mutation in the original metering data to obtain a pattern analysis result; the pattern analysis is used to judge whether the data loss or abnormal mutation has a time regularity or a correlation with a specific event; determining a pseudo-continuity feature of a line loss trend according to the average value, the standard deviation and the variation trend of the line loss data; attributing the line loss abnormality cause according to the pattern analysis result and the pseudo-continuity feature of the line loss trend, wherein when the data loss or abnormal mutation has a specific regularity and is highly consistent with the pseudo-continuity trend, attributing the line loss abnormality cause to a calculation deviation caused by a data quality problem; otherwise, attributing the line loss abnormality cause to a real physical loss; attributing the line loss abnormality cause to a real physical loss to determine the line section as the continuously high line loss section.

4. The power distribution network line loss node screening method of claim 1, wherein, The method for determining the continuously high line loss section by extracting line loss data of the line section at multiple time scales and determining the continuously high line loss section according to the line loss data at the multiple time scales, comprises: extracting line loss data of the line section preliminarily marked as a calculation fluctuation at multiple time scales, and analyzing average value, standard deviation and variation trend of the line loss data; when the line section shows higher than normal level at all analyzed time scales and the trend remains consistent, obtaining real-time distributed power output data and real-time load data of the line section; calculating fluctuation rates of the distributed power output data and the load data of the line section at different time scales; when the fluctuation rate of the distributed power output data and the fluctuation rate of the load data are highly correlated with significant fluctuation trends at the same time scale, attributing the line loss abnormality cause to a power flow fluctuation caused by distributed power access; otherwise, attributing the line loss abnormality cause to a real physical loss; attributing the line loss anomaly to a line section with real physical loss, and determining the line section with persistent high line loss as the line section with persistent high line loss.

5. The method for distribution network line loss node screening according to claim 1, characterized in that, The line section preliminarily marked as a line section with calculation fluctuation is extracted for line loss data at multiple time scales, and the line section with persistent high line loss is determined according to the line loss data at the multiple time scales, including: The line loss data of the line section preliminarily marked as a line section with calculation fluctuation is extracted at multiple time scales, and the average value, standard deviation and change trend of the line loss data are analyzed; When the line section shows higher than normal level at all analyzed time scales and the trend remains consistent, historical topology structure data and real-time topology structure change event records of the line section are obtained; According to the historical topology structure data, the topology structure configuration of the line section in different time periods is identified; According to the real-time topology structure change event records, the line loss data is processed in segments to ensure that the line topology structure remains consistent within each segment; The average value, standard deviation and change trend of the line loss data of each segment with consistent topology structure are analyzed at multiple time scales to determine whether the line loss data at the topology structure change point shows significant discontinuity mutation; When the line loss data at the topology structure change point shows significant discontinuity mutation, the line loss anomaly is attributed to calculation deviation caused by line topology structure change; otherwise, the line loss anomaly is attributed to real physical loss; attributing the line loss anomaly to a line section with real physical loss, and determining the line section with persistent high line loss as the line section with persistent high line loss.

6. The method for power distribution network line loss node screening according to claim 1, characterized in that, The line section preliminarily marked as a line section with calculation fluctuation is extracted for line loss data at multiple time scales, and the line section with persistent high line loss is determined according to the line loss data at the multiple time scales, including: The line loss data of the line section preliminarily marked as a line section with calculation fluctuation is extracted at multiple time scales, and the average value, standard deviation and change trend of the line loss data are analyzed; When the line section shows higher than normal level at all analyzed time scales and the trend remains consistent, historical topology structure data and real-time topology structure change event records of the line section are obtained; According to the historical topology structure data, the topology structure configuration of the line section in different time periods is identified; According to the real-time topology structure change event records, the line loss data is processed in segments to ensure that the line topology structure remains consistent within each segment; The average value, standard deviation and change trend of the line loss data of each segment with consistent topology structure are analyzed at multiple time scales to determine whether the line loss data at the topology structure change point shows significant discontinuity mutation; When the line loss data at the topology structure change point shows significant discontinuity mutation, the line loss anomaly is attributed to calculation deviation caused by line topology structure change; otherwise, the line loss anomaly is attributed to real physical loss; attributing the line loss abnormality reason to a real physical loss when the corrected line loss data still shows higher than normal level and the trend remains consistent in all analyzed time scales; otherwise, attributing the line loss abnormality reason to a calculation deviation caused by seasonal load change or holiday effect; attributing the line loss abnormality reason to a real physical loss when the corrected line loss data still shows higher than normal level and the trend remains consistent in all analyzed time scales; otherwise, attributing the line loss abnormality reason to a calculation deviation caused by seasonal load change or holiday effect; 7. The method for power distribution network line loss node screening according to claim 1, characterized in that, attributing the line loss abnormality reason to a real physical loss when the corrected line loss data still shows higher than normal level and the trend remains consistent in all analyzed time scales; otherwise, attributing the line loss abnormality reason to a calculation deviation caused by seasonal load change or holiday effect; attributing the line loss abnormality reason to a real physical loss when the corrected line loss data still shows higher than normal level and the trend remains consistent in all analyzed time scales; otherwise, attributing the line loss abnormality reason to a calculation deviation caused by seasonal load change or holiday effect; attributing the line loss abnormality reason to a real physical loss when the corrected line loss data still shows higher than normal level and the trend remains consistent in all analyzed time scales; otherwise, attributing the line loss abnormality reason to a calculation deviation caused by seasonal load change or holiday effect; attributing the line loss abnormality reason to a real physical loss when the corrected line loss data still shows higher than normal level and the trend remains consistent in all analyzed time scales; otherwise, attributing the line loss abnormality reason to a calculation deviation caused by seasonal load change or holiday effect; attributing the line loss abnormality reason to a real physical loss when the corrected line loss data still shows higher than normal level and the trend remains consistent in all analyzed time scales; otherwise, attributing the line loss abnormality reason to a calculation deviation caused by seasonal load change or holiday effect; the system comprises: a data receiving module configured to receive power data from each metering terminal in the power distribution network and calculate initial line loss of each line section according to the power data and energy balance principle; 8. A power distribution network line loss node screening system characterized by, a multi-time scale analysis module configured to extract line loss data of a line section preliminarily marked as calculation fluctuation in multiple time scales and determine a continuously high line loss line section according to the line loss data in the multiple time scales; the continuously high line loss line section has line loss data higher than normal level and consistent trend in all time scales; ​ ​ The abnormality analysis module is configured to continuously monitor real-time current and power data of the line section with continuously high line loss; when an average change rate of the real-time current or the real-time power exceeds a preset change rate threshold, mark a significant load change event; extract line loss calculation values, average current values and line loss power values before, during and after the significant load change event; perform line loss positive value checking to obtain a line loss positive value checking result; the line loss positive value checking is configured to check whether the extracted line loss calculation values are always positive values; perform square relationship response checking to obtain a square relationship response checking result; the square relationship response checking is configured to check whether changes in the extracted line loss calculation values are consistent with square changes in the average current values; perform asymmetric response checking to obtain an asymmetric response checking result; the asymmetric response checking is configured to identify a pair of symmetric load change events with load first increasing and then decreasing or load first decreasing and then increasing in a short time, and check whether an increase amplitude and a decrease amplitude of the line loss power values in the symmetric load change events both show significant asymmetry; and determine a line loss abnormality result according to the line loss positive value checking result, the square relationship response checking result and the asymmetric response checking result. The abnormality classification and suggestion generation module is configured to classify line loss abnormality conditions of each line section and generate different types of troubleshooting suggestions.

9. The power distribution network line loss node screening system of claim 8, wherein, The multi-time scale analysis module is specifically configured to: extract line loss data of a line section preliminarily marked as a calculation fluctuation at multiple time scales, when the line loss data at all analyzed time scales all show higher than normal levels and the trends are consistent, further extract line loss data of the line section in the past continuous days to obtain a line loss rate in a time window of a high loss phenomenon; retrieve environmental temperature data of a region where the line section is located, and calculate a correlation between the line loss rate in the time window of the high loss phenomenon and the environmental temperature data; attribute a line loss abnormality cause according to the line loss rate in the time window of the high loss phenomenon, the correlation and the environmental temperature data, wherein when the line loss rate in the time window of the high loss phenomenon reaches a line loss rate threshold, the correlation is higher than a correlation threshold, and the environmental temperature data continuously exceed a specific high temperature threshold, attribute the line loss abnormality cause to a time drift calculation deviation caused by environmental thermal stress; otherwise, attribute the line loss abnormality cause to a real physical loss; determine the line section with the line loss abnormality cause attributed to the real physical loss as the line section with continuously high line loss.

Images