A method for calculating and correcting the average sensitivity of overall power failure of distribution network station area

Abstract

The application discloses a kind of distribution network station area overall power failure sensitivity mean measurement and correction method, belong to distribution network operation and maintenance technical field.The method first dynamically identifies the user set actually affected by power failure event, based on the importance weight of user electricity category, calculate the weighted sensitivity mean of affected user;Secondly, introduce real-time context factor, from time period, season, holiday, four dimensions of current power failure duration to the mean coupling correction;Finally, construct the dual-dimension evaluation system of mean and extreme value risk, obtain the overall power failure sensitivity of station area by dynamic weighted fusion, and implement grading correction according to sensitive grade.The application solves the technical problems in traditional arithmetic mean method, such as non-power failure user lowering mean, static sensitivity and real-time context disconnection, single mean shielding extreme risk, etc., can accurately reflect the overall power failure influence degree of station area, and provide scientific decision-making basis for distribution network operation and maintenance priority sequencing and resource optimization configuration.

Description

Technical Field

[0001] This invention relates to the field of power distribution network operation and maintenance management technology. Specifically, it relates to a method for calculating and correcting the average value of power outage sensitivity of the entire distribution network area, which is applicable to power outage impact assessment and maintenance priority decision-making at the distribution network area level. Background Technology

[0002] With the continuous advancement of power distribution network construction and the increasing demands of users for power supply reliability, scientifically assessing the impact of power outages on users and rationally allocating operation and maintenance resources have become key issues for power supply companies to improve service quality. Power outage sensitivity, as a core indicator for measuring users' tolerance and response intensity to power outages, directly impacts the scientific nature of operation and maintenance decisions and customer satisfaction.

[0003] Currently, relevant technologies have been studied to assess user sensitivity to power outages. For example, Chinese invention patent CN121114631B discloses a "Method for Judging Low-Voltage Instantaneous Interruption Hazards and User Sensitivity Based on Power Outage Information." This method, through comparison of historical data from multiple systems, can accurately identify the actual causes of power outages in massive amounts of meter data. It also uses a hierarchical analysis algorithm to quantify user sensitivity to power outages from dimensions such as total number of outages and total outage duration, and combines this sensitivity to determine the urgency ranking of power outage handling. This technology has achieved certain results in sensitivity assessment at the single-user level.

[0004] However, in practical applications, the inventors discovered that the existing technology still has the following unaddressed and unresolved technical problems: First, existing technologies primarily focus on sensitivity calculation at the single-user level, lacking a method for aggregated sensitivity measurement from the perspective of the entire distribution area. In actual distribution network operation and maintenance, the allocation of emergency repair resources and the arrangement of power outage plans are often decided on a distribution area basis. Single-user sensitivity cannot directly support operation and maintenance decisions at the distribution area level. If the arithmetic mean of the sensitivity of all users within a distribution area is simply used as the overall sensitivity of the distribution area, there is a serious "zero-value lowering effect for non-outage users"—when a partial power outage occurs in a distribution area, unaffected users have lower sensitivity values ​​because there is no record of the current outage duration, directly lowering the overall average and causing the overall impact of the power outage in the distribution area to be severely underestimated.

[0005] Second, existing user sensitivity calculations are mostly static scores based on historical data, failing to consider real-time contextual factors at the time of power outages. In reality, users' feelings and tolerance for power outages are significantly time- and context-dependent: the same power outage event can elicit vastly different user experiences and willingness to complain depending on whether it occurs during the hottest part of summer versus early morning in winter, on a weekday versus a holiday, or during the day versus late at night. Static sensitivity cannot reflect this dynamic change, leading to discrepancies between the calculated results and the actual user experience.

[0006] Third, a single mean metric suffers from "mean reversion bias" and "extreme user masking effect." When a distribution area contains highly sensitive users such as hospitals, data centers, and important government agencies, the arithmetic mean will be diluted by a large number of ordinary users, failing to reflect the extreme risks of these critical users. Conversely, when a small distribution area contains a small number of highly sensitive users, the mean may be excessively inflated, resulting in an artificially high sensitivity level. A single mean cannot simultaneously reflect both the overall average level and extreme risk points of a distribution area, easily leading to misjudgments in operational decisions.

[0007] Fourth, existing technologies lack differentiated correction mechanisms for different sensitivity levels, and adopt a one-size-fits-all calculation method, which cannot classify and calibrate according to the sensitivity of the transformer area, thus affecting the applicability of the calculation results in different scenarios.

[0008] Therefore, there is an urgent need to develop a method that can accurately measure the overall power outage sensitivity of the distribution network area and make scientific corrections to solve the above-mentioned technical problems and provide a more accurate and reliable basis for distribution network operation and maintenance decisions. Summary of the Invention

[0009] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for calculating and correcting the average sensitivity of the overall power outage of a distribution network area. This method solves the technical problems in the traditional calculation of the overall sensitivity of a distribution network area, such as the lowering of the average value by non-power outage users, the disconnect between static sensitivity and real-time situation, and the obscuring of extreme risks by a single average value. This method enables an accurate quantitative assessment of the impact of the overall power outage on the distribution network area.

[0010] To achieve the above objectives, the present invention adopts the following technical solution: A method for calculating and correcting the average power outage sensitivity of a distribution network area includes the following steps: S1. Dynamic identification of affected users: Based on the power outage event record, dynamically identify the set of users actually affected by this power outage and exclude unaffected users; S2. Basic Calculation of User Sensitivity: Based on users' historical power outage data, electricity consumption characteristics and request records, calculate the basic value of individual power outage sensitivity for each affected user; S3. Weighted average calculation: Based on the user's electricity consumption category, assign corresponding importance weight coefficients, calculate the weighted average of the individual sensitivities of the affected users, obtain the weighted average sensitivity of the transformer area, and introduce a transformer area size correction coefficient for calibration. S4. Real-time Context Correction: Construct a real-time context correction system that includes time period factors, seasonal factors, holiday factors and current power outage duration factors. Context correction coefficients are obtained through multi-factor coupling calculation, and the weighted sensitivity mean is dynamically corrected. S5. Extreme Value Risk Calculation: Extract the top N most sensitive users in the transformer area, calculate the extreme value risk contribution, and reflect the level of extreme power outage risk in the transformer area. S6. Dual-dimensional fusion: Based on the user composition characteristics and outage type of the transformer area, the fusion weight is dynamically adjusted, and the corrected weighted average and extreme value risk contribution are weighted and fused to obtain the initial value of the overall outage sensitivity of the transformer area. S7. Graded Correction Output: Based on the initial value of overall sensitivity, the transformer area is divided into different sensitivity levels. Different graded correction strategies are adopted for different levels, and the final value of the overall power outage sensitivity of the transformer area is finally output.

[0011] Furthermore, the specific method for dynamically identifying affected users in step S1 is as follows: Based on the power outage records of the meter in the electricity information collection system, the users with valid power outage records within the same transformer area are grouped into an affected user set, with a preset time window before and after the start time of the power outage event as the judgment interval. For record misalignment caused by clock deviation, the average power outage duration sequence of the transformer area is used for clock correction and user completion processing.

[0012] Furthermore, in step S3, the user importance weighting coefficient is set according to the electricity consumption category: The weighting coefficients for large industrial users are 1.5-2.0, for commercial users 1.2-1.5, for general industrial users 1.0-1.2, for residential users 0.8-1.0, and for agricultural production users 0.6-0.8. The formula for calculating the transformer area size correction factor is: K_scale=1-e (-N / 50) , where N is the total number of users in the region.

[0013] Furthermore, the specific composition of the real-time context factor in step S4 includes: Time-of-use factor: Based on the time of power outage, it is divided into three time-of-use levels: early morning off-peak, daytime flat period, and evening peak, each corresponding to a different correction coefficient; Seasonal factors: Based on the months, the seasons are divided into three levels: winter heating season, summer cooling season, and spring and autumn transition season. Differential correction coefficients are set in combination with electricity consumption categories. Holiday factor: Differentiate between weekdays, weekends, and statutory holidays, and set differentiated correction coefficients for different electricity consumption categories; Current power outage duration factor: Divided into multiple intervals according to the duration of the power outage, the longer the power outage duration, the larger the correction coefficient.

[0014] Furthermore, the multi-factor coupling calculation in step S4 adopts a nonlinear coupling method: The context correction coefficient K_context = 1 + α × (K_time-1) + β × (K_season-1) + γ × (K_holiday-1) + δ × (K_duration-1) + ε × (K_time-1) × (K_season-1); where K_time is the time period factor, K_season is the season factor, K_holiday is the holiday factor, and K_duration is the power outage duration factor; α, β, γ, and δ are the main effect weights of each factor, and ε is the interaction effect weight between time period and season.

[0015] Furthermore, the method for calculating the extreme value risk contribution in step S5 is as follows: Extract the top 5% of highly sensitive users within the transformer area, with no fewer than 3 users, and calculate their average sensitivity S_top; extreme value risk contribution S_extreme = S_top × K_concentration, where K_concentration is the concentration coefficient of highly sensitive users, calculated as K_concentration = 1 + (n_top / N_total), where n_top is the number of highly sensitive users and N_total is the total number of affected users.

[0016] Furthermore, the dynamic weight adjustment mechanism for the two-dimensional fusion in step S6 is as follows: When the power outage type is planned power outage, the mean weight β is taken as 0.6-0.7, and the extreme value weight is taken as 0.3-0.4; When the power outage type is fault outage, the mean weight β is taken as 0.4-0.5, and the extreme value weight is taken as 0.5-0.6; When a distribution area contains important power users, the extreme value weight is automatically increased by 0.1-0.2, while the mean weight is reduced accordingly.

[0017] Furthermore, the hierarchical correction strategy in step S7 specifically includes: The overall sensitivity of the transformer area is divided into four levels: low sensitivity, moderate sensitivity, high sensitivity, and extreme sensitivity. Low-sensitivity areas are corrected using a baseline correction factor of 1.0. For moderately sensitive areas, a slight upward adjustment is applied, with a correction factor of 1.05-1.1. For highly sensitive areas, a moderate upward adjustment is applied, with a correction factor of 1.1-1.2. Extremely sensitive areas will be subject to a significant upward adjustment, with a correction factor of 1.2-1.3. The thresholds for each level are dynamically adjusted based on historical complaint data.

[0018] Furthermore, the formula for calculating the baseline value of individual user power outage sensitivity is as follows: S_individual=α×(a×N×b+C+C9×D+H+F); Wherein, α is the electricity consumption coefficient, reflecting the matching degree between the user's electricity consumption and the contracted capacity; a is the scaling parameter, set according to the electricity consumption category; N is the number of power outages in the past 60 days; b is the power outage impact adjustment parameter, which is positively correlated with the number of power outages; C is the comprehensive time period factor, including seasonal factors, intraday time period factors, holiday factors, major event factors, and individual difference characteristic factors; C9 is the power outage type coefficient; D is the current power outage duration impact factor; H is the power outage request frequency parameter; and F is the non-power outage request frequency parameter.

[0019] Furthermore, this method is implemented through a power distribution network service command system, which includes a data acquisition module, a user sensitivity calculation module, a transformer area average value calculation module, a scenario correction module, an extreme value risk assessment module, a fusion output module, and a hierarchical correction module. The data acquisition module is connected to the electricity consumption information acquisition system, the marketing business system, the production management system, and the 95598 work order system to achieve unified access and preprocessing of multi-source data.

[0020] Compared with the prior art, the present invention has the following beneficial effects: (1) Dynamic identification and weighted average calculation of affected users solve the problem of the zero value of non-power outage users being lowered in the traditional arithmetic average method. By accurately locating the actual set of affected users and assigning differentiated weights according to the user's electricity consumption category, the overall sensitivity of the transformer area can truly reflect the actual impact of the power outage event, avoiding the dilution effect of unaffected users on the mean, and the calculation results are more accurate.

[0021] (2) The real-time contextual factor coupling correction mechanism realizes the transformation from static sensitivity to dynamic sensitivity. By introducing four-dimensional contextual factors such as time period, season, holidays, and current power outage duration, and considering the interactive coupling effect between factors, the calculation results can be dynamically adjusted according to the specific context of the power outage, which is more in line with the actual feelings and complaint intentions of users, and improves the timeliness and accuracy of sensitivity assessment.

[0022] (3) The dual-dimensional evaluation system of mean and extreme values ​​solves the problem of the masking effect of a single mean indicator. By simultaneously considering the overall average sensitivity level of the transformer area and the extreme risk level of highly sensitive users, and dynamically adjusting the fusion weight according to the power outage type and user composition, it can reflect the overall situation of the transformer area without ignoring the important impact of key users, providing more comprehensive information support for operation and maintenance decisions.

[0023] (4) The graded correction strategy enables differentiated and precise calibration. Different correction coefficients are used for transformer areas with different sensitivity levels, avoiding a one-size-fits-all calculation method, so that the measurement results can maintain good accuracy and applicability for transformer areas with different sensitivity levels.

[0024] (5) All data in this invention can be obtained from existing power information systems without the need for additional hardware monitoring equipment. The algorithm logic is clear, the calculation efficiency is high, and it is easy to integrate and deploy in existing power supply service command systems. It has strong practicality and promotion value. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the overall process of the method of the present invention; Figure 2 This is a schematic diagram of the system architecture of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0027] Example 1 Reference Figure 1 As shown in the figure, this embodiment provides a method for calculating and correcting the average power outage sensitivity of a distribution network area. The overall process includes seven main steps, and the specific implementation process is as follows: S1. Dynamic identification of affected users.

[0028] During the operation of a power distribution network, after a power outage occurs, the first step is to determine which users are actually affected. Traditionally, it's assumed that all users within the affected area are impacted. However, in reality, especially in low-voltage areas, branch line faults may cause localized outages, affecting not all users. Including unaffected users in the average calculation would significantly lower the overall sensitivity average since their outage duration is zero, leading to assessment bias.

[0029] This step identifies users based on the power outage records of the electricity information collection system. Specifically, a 15-minute time window is set before and after the start time of the power outage event. Within this time window, the power outage records of all users in the same area are retrieved. Users with valid power outage records (i.e., the power outage start time falls within the time window and the power outage duration is greater than 1 minute) are included in the set of affected users for this power outage event.

[0030] In practical applications, clock deviations in some users' meters can cause misalignments between the timestamps of power outage records and the actual outage times, affecting identification accuracy. To address this issue, this invention uses an average outage duration sequence for the transformer area for correction. First, the outage duration sequence of core users with accurate clocks within the transformer area (typically accounting for about 30%, calibrated using historical data) is selected as the baseline sequence. Then, the outage records of other users are time-axis shifted and matched. When a user's outage duration sequence is highly correlated with the baseline sequence in terms of morphology (correlation coefficient greater than 0.85) but has a time offset, the user's clock is corrected accordingly, and the user is added to the set of affected users.

[0031] The above dynamic identification method can accurately identify the user groups actually affected by the power outage, eliminate the interference of unaffected users on the subsequent average calculation, and lay the foundation for the accurate measurement of the overall sensitivity of the transformer area.

[0032] S2. Basic calculation of user sensitivity.

[0033] After determining the set of affected users, it is necessary to calculate the basic value of each user's individual power outage sensitivity. This invention adopts a multi-dimensional comprehensive scoring model to comprehensively measure the user's power outage sensitivity from multiple dimensions such as electricity consumption characteristics, historical power outages, and complaint records.

[0034] The formula for calculating the baseline value of individual sensitivity is: S_individual = α × (a × N × b + C + C9 × D + H + F); where: The electricity consumption factor α reflects the degree of matching between a user's actual electricity consumption and their contracted capacity. Higher electricity consumption indicates a stronger dependence on electricity and a correspondingly higher sensitivity to power outages. The calculation method is as follows: α = (Average daily electricity consumption / 24) / (Contracted capacity × Baseline power factor); where, the average daily electricity consumption is the average of the past 30 days, divided by 24 to obtain the hourly average electricity consumption; the contracted capacity is the capacity of the contract signed between the user and the power supply company; the baseline power factor is determined according to the user's electricity consumption category, which is 0.85 for general industrial, residential, and commercial electricity consumption, 0.9 for large industrial electricity consumption, and 0.8 for agricultural irrigation and drainage, agricultural production, etc.

[0035] The scaling parameter 'a' reflects the inherent differences in the sensitivity of different industry users to power outages and is set according to the user's electricity consumption category. For example, large industrial users, due to high requirements for production continuity, suffer significant losses from power outages, so the 'a' value is set to 0.7-0.8; commercial users, especially shopping malls and restaurants, are directly affected by power outages, so the 'a' value is set to 0.6; general industrial users have an 'a' value of 0.6; residential users have an 'a' value of 0.5; agricultural production users have an 'a' value of 0.4; and power generation users such as those using solar power have an 'a' value of 0.2. For users who do not match the standard category, the default value of 'a' is set to 0.48.

[0036] The number of power outages (N) in the past 60 days represents the total number of power outage events that occurred to the user within the past 60 days, reflecting the frequency of power outages experienced by the user. Users who experience more frequent power outages typically have lower tolerance for power outages and are more sensitive to them.

[0037] The power outage impact adjustment parameter 'b' is correlated with the number of power outages (N) in the past 60 days, reflecting a non-linear amplification effect of "the more frequent the power outages, the greater the sensitivity." Specifically, the values ​​are: b=1 for 1-2 power outages; b=2 for 3-4 outages; b=3 for 5-7 outages; and b=4 for 8 or more outages. Through the combined calculation of a×N×b, the sensitivity value for users experiencing high-frequency power outages can be effectively amplified.

[0038] The comprehensive time period factor C is composed of five sub-factors, namely C=C1+C2+C3+C4+C5, which correspond to seasonal factors, intraday time period factors, holiday factors, major event factors, and individual difference characteristics factors, respectively.

[0039] C1 Seasonal Factor: Different scores are set based on the month in which the power outage occurs and the user's electricity consumption category. For example, during the high-temperature summer months of July and August, residential and commercial users have high air conditioning loads, and power outages have a significant impact, so C1 is assigned a higher value; agricultural irrigation and drainage have a high C1 value during the summer irrigation season, while it is 0 in other seasons.

[0040] C2 Intraday Time Factor: Set according to the time period of the power outage. For example, commercial users are most affected by power outages during the peak business hours of 7-10 pm, so C2 is set to 4; residential users also have peak electricity consumption in the evening, so C2 is set to 2; most users have low electricity load from 0-7 am, so C2 is set to a negative value.

[0041] C3 Holiday Factor: Distinguish between holidays and non-holidays. Residential and commercial users tend to stay home or consume more during holidays, increasing the impact of power outages, so C3 takes a positive value; industrial users may cease production during holidays, thus reducing the impact of power outages, so C3 takes a negative value or zero.

[0042] C4 Major Event Factors: During special periods such as the college entrance examination and the high school entrance examination, the sensitivity of education users is significantly increased, so C4 is set at 2; the sensitivity of residential users is also increased to a certain extent, so C4 is set at 0.5.

[0043] C5 Individual Difference Factors: These are set based on users' historical complaints. Users with a history of complaints have a correspondingly higher sensitivity. The specific calculation method is (score for high-sensitivity channel work orders + score for local work orders + emotional bonus) multiplied by a constant. High-sensitivity channels include 12345, 95598, 12398, etc., with each complaint work order worth 5 points, each suggestion work order worth 3 points, and each other work order worth 1.5 points. Each local work order is worth 3 points for a complaint, 2 points for a suggestion, and 1 point for other work orders. Emotional bonuses are achieved through keyword matching: work orders containing angry words such as "must be resolved" or "unacceptable" receive 2 points, and work orders containing disappointed words such as "occurred multiple times" or "has a significant impact" receive 1 point.

[0044] The current power outage duration impact factor C9×D reflects the influence of the duration of the current power outage on sensitivity. D is the power outage duration factor, categorized by duration: 0-10 minutes D=0.5; 10-30 minutes D=1.0; 30 minutes-1 hour D=2.0; 1-3 hours D=4.0; over 3 hours D=6.0. C9 is the power outage type coefficient, with values ​​of 0.25, 0.5, 0.75, and 1.0 for "low-voltage planned," "medium-voltage planned," "low-voltage fault," and "medium-voltage fault," respectively. The impact of fault outages is greater than that of planned outages.

[0045] The power outage request frequency parameter H is set to H=4n for customers with a history of power outage requests, where n is the number of requests. This parameter, together with C5, forms a double weighting, further enhancing the sensitivity of users with a history of power outage complaints.

[0046] The parameter F, representing the frequency of non-power outage requests, is set to F=2.5n for customers with a history of non-power outage requests, where n is the number of requests. This reflects the overall user's awareness of their rights and their sensitivity to service; users with more non-power outage requests are generally more sensitive to power outages.

[0047] By comprehensively calculating the above multi-dimensional parameters, the power outage sensitivity of each user can be fully and objectively reflected, providing reliable basic data for subsequent overall aggregation calculation of the transformer area.

[0048] S3. Calculation of weighted average.

[0049] After obtaining the individual sensitivity of all affected users, they need to be aggregated into the overall sensitivity mean of the entire network area. This invention does not use a simple arithmetic average, but instead introduces user importance weights and network area size corrections to make the mean more scientific and reasonable.

[0050] First, different importance weight coefficients are assigned to users based on their electricity consumption categories. This is because the social impact and economic losses from power outages differ for different types of users. For example, a power outage for large industrial users may lead to production line shutdowns and raw material waste, resulting in huge economic losses; a power outage for critical users such as hospitals and data centers may endanger lives or cause significant social impact; while the impact of a power outage for ordinary residential users is relatively small. Therefore, assigning different weights reflects this difference.

[0051] The specific weighting settings are as follows: Large industrial users have a weighting coefficient of 1.8, commercial users 1.3, general industrial users 1.1, urban residential users 1.0, rural residential users 0.9, and agricultural production users 0.7. Important electricity users (such as hospitals, government agencies, transportation hubs, etc.) are further multiplied by an important user coefficient of 1.5 based on their corresponding category.

[0052] The formula for calculating the weighted mean is: S_weighted_mean = (ΣS_i × w_i) / Σw_i; Where S_i is the individual sensitivity of the i-th user, w_i is the weight coefficient of the i-th user, and the summation range is all affected users.

[0053] In addition to the weighted mean, a district size correction coefficient K_scale needs to be introduced. This is because for small districts with few users, a small number of highly sensitive users may cause the mean to be excessively inflated, making it incomparable to the mean of larger districts. The district size correction uses an exponential convergence function: K_scale=1-e (-N / 50) Where N is the total number of users in the transformer area. As N approaches infinity, K_scale approaches 1; when N=50, K_scale is approximately 0.63; and when N=100, K_scale is approximately 0.86. This adjustment appropriately lowers the average sensitivity of smaller transformer areas, making them more comparable across transformer areas of different sizes.

[0054] The adjusted weighted mean is: S_mean_adjusted = S_weighted_mean × K_scale; By using weighted calculations and size corrections, the resulting average sensitivity of the transformer area takes into account the differences in importance among different users and eliminates the bias caused by differences in transformer area size, thus more objectively reflecting the overall sensitivity level of the transformer area.

[0055] S4. Real-time scenario correction.

[0056] The weighted average calculated above is essentially a static sensitivity based on historical data, reflecting the "average" sensitivity level of users. However, users' actual feelings about power outages are closely related to the specific context in which the outage occurs. For the same user and the same outage duration, the user's tolerance and willingness to complain vary significantly depending on the time context. Therefore, it is necessary to introduce a real-time contextual factor to dynamically correct the static average.

[0057] This invention constructs a four-dimensional real-time context factor system, including time period factor, seasonal factor, holiday factor, and current power outage duration factor.

[0058] The time period factor K_time divides the day into three time period levels based on the time of the power outage: During the off-peak hours in the early morning (0:00-7:00): most users are asleep, the power load is low, and the perception of power outages is weak, so K_time is set to 0.8; Daytime normal hours (8:00-18:00): Normal working and living hours, power outage impact is moderate, K_time is set to 1.0; Evening peak (19:00-22:00): Peak hours for residential and commercial operations, with high electricity load and strong perception of power outages. K_time is set to 1.3.

[0059] The period from 23:00 to 24:00 falls between the low point and the normal period, so K_time is set to 0.9.

[0060] The seasonal factor K_season is divided into three seasonal levels based on the month: Summer cooling season (July-August): Air conditioning load is high under high temperatures, and power outages severely affect comfort. Residents and commercial users are significantly more sensitive to heat. K_season is set to 1.25. Winter heating season (December-February): Electricity demand for heating is high in cold weather, so K_season is set to 1.1; Spring and autumn transition period (March-June, September-November): The climate is mild and the electricity load is relatively stable. K_season is set to 1.0.

[0061] For agricultural irrigation and drainage users, K_season is set to 1.5 during the summer irrigation season and 0.5 during other seasons; for educational users, K_season is appropriately reduced during winter and summer vacations.

[0062] The holiday factor K_holiday distinguishes three date types: Statutory holidays: Residents spend more time at home, and businesses have high customer traffic, so power outages have a significant impact. For residential and commercial users, K_holiday is set at 1.15; for industrial users, K_holiday is set at 0.9 because production may be suspended. Weekends: The impact is between that of weekdays and public holidays; for residential and commercial users, K_holiday is set to 1.08. Weekday: Baseline state, K_holiday is 1.0.

[0063] The current power outage duration factor K_duration shows that as the power outage time increases, user dissatisfaction accumulates and sensitivity rises. The tiered settings are as follows: 0-10 minutes: Short power outage, most users are unaware or the impact is minimal, K_duration is set to 0.7; 10-30 minutes: Users begin to perceive the power outage, and the impact gradually becomes apparent; K_duration is set to 0.9. 30 minutes - 1 hour: The power outage has a significant impact, and users begin to feel dissatisfied. K_duration is set to 1.1. 1-3 hours: Prolonged power outages significantly increase user dissatisfaction and the probability of complaints; K_duration is set to 1.4. More than 3 hours: Severe and prolonged power outage, causing strong user dissatisfaction, K_duration is set to 1.8.

[0064] It is important to note that the four factors mentioned above are not simply linearly multiplied, but rather exhibit interactive coupling effects. For example, the impact of a power outage during peak evening hours in summer is greater than the simple sum of the "summer impact" and the "evening impact," because the combination of high temperatures and peak evening electricity consumption creates a magnifying effect.

[0065] Therefore, this invention employs a nonlinear coupling calculation method, and the formula for calculating the scenario correction coefficient is as follows: K_context=1+α×(K_time-1)+β×(K_season-1)+γ×(K_holiday-1)+δ×(K_duration-1)+ε×(K_time-1)×(K_season-1); In this context, α, β, γ, and δ represent the main effect weights of the four factors: time period, season, holiday, and power outage duration, with typical values ​​of 0.3, 0.25, 0.15, and 0.3, respectively; ε represents the interaction effect weight between time period and season, with a typical value of 0.2. By setting the interaction terms, the amplification effect when different contextual factors are superimposed can be reflected.

[0066] The weighted average of the transformer substations after context correction is: S_context_mean=S_mean_adjusted×K_context; By using real-time context correction, the overall sensitivity of the transformer area can be dynamically adjusted according to the specific time context of the power outage, which is closer to the actual experience of users and improves the accuracy and timeliness of the assessment.

[0067] S5. Extreme value risk assessment.

[0068] Relying solely on average metrics has a significant drawback: when there are a few extremely sensitive users within a distribution area, the extreme risks posed by these users are diluted by the large number of ordinary users, and the average cannot reflect the importance of these critical users. In actual operation and maintenance, it is precisely these highly sensitive users (such as hospitals, important government agencies, and large enterprises) whose power outages are most likely to cause serious consequences and complaints. Therefore, it is necessary to separately calculate extreme risk as a supplement to the average dimension.

[0069] The specific steps for extreme value risk assessment are as follows: First, affected users are sorted from highest to lowest individual sensitivity, and the top 5% are selected as the highly sensitive user group. A minimum number of highly sensitive users is set; if the top 5% has fewer than three users, then the top three users are selected. This is to avoid excessive influence of individual outliers on extreme value calculations.

[0070] Then, calculate the average sensitivity S_top of highly sensitive users: S_top=(ΣS_top_i) / n_top; where S_top_i is the individual sensitivity of the i-th highly sensitive user, and n_top is the number of highly sensitive users.

[0071] Next, we introduce the concentration coefficient K_concentration of highly sensitive users. This is because if highly sensitive users account for a high proportion of users in a distribution area, it indicates that the overall user composition of that distribution area is more sensitive, and the impact of extreme risks is more widespread; conversely, if the proportion of highly sensitive users is very low, it indicates that only a few users are particularly sensitive, and the overall extreme risks are relatively controllable.

[0072] The concentration factor is calculated as: K_concentration = 1 + (n_top / N_total); Where N_total represents the total number of affected users.

[0073] The final extreme value risk contribution is: S_extreme = S_top × K_concentration; For example, if a certain area has 100 affected users, including 5 highly sensitive users with an average sensitivity score of 80, then the concentration coefficient is 1 + 5 / 100 = 1.05, and the extreme value risk contribution is 80 × 1.05 = 84.

[0074] Extreme value risk calculation can independently quantify the extreme power outage risk brought by highly sensitive users in the distribution area, making up for the shortcomings of a single average indicator and providing a more comprehensive risk perspective for operation and maintenance decisions.

[0075] S6. Dual-dimensional integration.

[0076] After obtaining the two dimensions of indicators—the weighted mean and the contribution of extreme value risk—after scenario correction, they need to be integrated into a comprehensive initial value for the overall power outage sensitivity of the transformer area. This integration is not a simple fixed-weighted sum, but rather involves dynamically adjusting the weights based on the specific scenario.

[0077] The basic formula for two-dimensional fusion is: S_initial=β×S_context_mean+(1-β)×S_extreme; where β is the weight of the mean dimension and (1-β) is the weight of the extreme value dimension.

[0078] The value of the weight β is not fixed, but is dynamically adjusted according to the power outage type and the composition of users in the transformer area: Impact of different types of power outages: When the power outage type is a planned power outage, since the power outage time is notified in advance, users have some preparation and the overall impact is relatively controllable. At this time, more attention is paid to the average impact level of the transformer area, with the mean weight β set to 0.65 and the extreme value weight set to 0.35.

[0079] When the power outage type is fault outage, the power outage is sudden and users are not prepared. Sudden power outages of highly sensitive users are more likely to trigger emergencies and strong complaints. At this time, more attention needs to be paid to extreme risks. The mean weight β is 0.45 and the extreme value weight is 0.55.

[0080] Impact on key users: If the distribution area includes first- or second-level critical power users (such as hospitals, emergency command centers, transportation hubs, etc.), the extreme value weight will automatically increase by 0.15, and the mean value weight will decrease by 0.15 accordingly. This is because power outages to critical users can have significant social impacts and must be given higher priority.

[0081] The impact of user composition uniformity: If the coefficient of variation (standard deviation / mean) of user sensitivity within a station area is greater than 0.5, it indicates that there is a large difference in sensitivity among users. In this case, the extreme value weight should be increased by 0.05 to better reflect the differentiation. If the coefficient of variation is less than 0.2, it indicates that the user sensitivity is relatively close and the mean is highly representative. In this case, the mean weight can be increased by 0.05.

[0082] Through the aforementioned dynamic weight adjustment mechanism, the dual-dimensional fusion can adapt to different power outage scenarios and transformer area characteristics, making the comprehensive sensitivity value reflect both the overall average level and extreme risks, thus making it more scientific and reasonable.

[0083] Taking a typical scenario as an example: A commercial transformer area experiences a power outage due to a fault. After scenario correction, the mean score is 60, the extreme risk score is 85, there are no important users, and the user dispersion is moderate. Since it is a power outage due to a fault, β is taken as 0.45, then: S_initial=0.45×60+0.55×85=27+46.75=73.75 points; As can be seen, since the power outage fault places more emphasis on extreme risk, the final score of 73.75 is closer to the extreme value of 85, which is consistent with the risk preference under the fault scenario.

[0084] S7. Hierarchical correction output.

[0085] After obtaining the initial value of the overall power outage sensitivity of the transformer substation, a graded correction is required to finally output the final sensitivity value. This is because the slope of the probability curve for user complaints differs for substations with different sensitivity levels—the higher the sensitivity of the substation, the greater the increase in the probability of complaints for each 1-point increase in sensitivity. Therefore, a certain degree of upward adjustment is needed for highly sensitive substations to match the actual complaint risk curve.

[0086] First, the overall sensitivity of the transformer area is divided into four levels: Low sensitivity: S_initial < 40 points; Moderate sensitivity: 40 points ≤ S_initial < 60 points; Highly sensitive: 60 points ≤ S_initial < 80 points; Extremely sensitive: S_initial ≥ 80 points; The thresholds for each level can be dynamically adjusted based on local historical complaint data. For example, by statistically analyzing the actual complaint rates corresponding to different sensitivity ranges in history, key points of sudden changes in complaint rates can be identified as level thresholds, making the classification more closely aligned with local realities.

[0087] Then, differentiated correction coefficients are applied to different levels: Low-sensitivity transformer areas: Baseline correction is used, with a correction factor K_grade=1.0. Users in this level of transformer area generally have high tolerance and low complaint risk, requiring no additional correction.

[0088] Moderately sensitive areas: A slight upward adjustment is adopted, with an adjustment factor K_grade=1.08. Areas at this level begin to show some complaint risk, and the adjustment is appropriately increased to reflect the cumulative effect of risk.

[0089] Highly sensitive areas: A moderate upward adjustment is adopted, with an adjustment factor K_grade=1.15. Areas in this category have a higher probability of complaints and require more attention; therefore, the adjustment magnitude is increased.

[0090] Extremely sensitive areas: A significant upward adjustment is applied, with an adjustment factor K_grade=1.25. Areas at this level have an extremely high risk of complaints and are likely to trigger a surge in complaints or major public opinion incidents; the substantial upward adjustment highlights the urgency of the situation.

[0091] The final overall power outage sensitivity value for the transformer substation is: S_final = S_initial × K_grade; By implementing a tiered adjustment system, a better linear correlation is established between sensitivity scores and actual complaint risks, allowing operations and maintenance personnel to intuitively assess risk levels and urgency based on the scores. Simultaneously, the tiered adjustment system prioritizes highly sensitive areas in the order of priority allocation, ensuring they receive operational resources first.

[0092] This completes the entire process of calculating and correcting the overall power outage sensitivity of the distribution network area. This method, from user identification, basic calculation, weighted aggregation, scenario correction, extreme value supplementation, two-dimensional fusion to hierarchical output, forms a complete and scientific evaluation system that can accurately and comprehensively reflect the overall power outage sensitivity of the distribution network area.

[0093] Example 2 This embodiment, based on the method described in Embodiment 1, provides a distribution network power supply service command system for implementing the method, referring to... Figure 2 As shown.

[0094] The system adopts a modular architecture design, mainly including the following seven functional modules: (1) Data acquisition module: This module is the system's data input layer, responsible for interfacing with multiple external business systems to achieve unified access and preprocessing of multi-source data. The interfacing systems include: Electricity Information Acquisition System: Acquires real-time measurement data such as power outage records, electricity consumption data, voltage and current from user meters; Marketing business system: Obtain user profile information, including electricity consumption type, contracted capacity, user address, and key user identifiers; Production Management System (PMS): Acquires production data such as equipment ledgers, power outage plans, and fault records; 95598 Work Order System: Obtains work order records such as user complaints, opinions, and inquiries, including work order content, work order type, and processing time; Distribution automation master station system: acquires information such as switch tripping and medium-voltage line power outages.

[0095] The data acquisition module performs preprocessing operations such as cleaning, deduplication, and format conversion on the incoming raw data to form a standardized data resource pool, providing a unified data foundation for subsequent calculations.

[0096] (2) User sensitivity calculation module: This module corresponds to the second step in the method and is responsible for calculating the basic individual power outage sensitivity value for each user. The module has a built-in complete parameter configuration table, including scaling parameters 'a' for various user types, power factor baselines, seasonal factor tables, time-of-day factor tables, and holiday factor tables. Based on user profiles and historical data, the module automatically calculates the individual sensitivity score for each user according to a preset formula and supports flexible parameter configuration and adjustment.

[0097] (3) Substation area mean value calculation module: This module corresponds to steps one and three in the method, and is responsible for the dynamic identification of affected users and the calculation of the weighted average. First, based on the power outage event information, the module matches the set of affected users from the data resource pool; then, based on the user category, it calls the weight configuration to calculate the weighted sensitivity average; finally, it applies a transformer area size correction coefficient for calibration and outputs the corrected weighted average.

[0098] (4) Context Correction Module: This module corresponds to step four in the method and is responsible for the calculation and correction of real-time situational factors. Based on the time information of the power outage, the module automatically determines the time period, season, and holiday type, and calculates each situational factor in conjunction with the current duration of the power outage. Then, it calculates the comprehensive situational correction coefficient according to the coupled formula, dynamically correcting the weighted average of the transformer area. The module has a built-in clock service to ensure the accuracy of time determination.

[0099] (5) Extreme Value Risk Assessment Module: This module corresponds to step five in the method and is responsible for calculating the contribution of extreme value risk. The module sorts affected users by sensitivity, extracts a set of highly sensitive users, then calculates the average sensitivity and concentration coefficient of these highly sensitive users, and finally outputs the contribution of extreme value risk. The module supports parameter configuration for the proportion threshold and minimum number of highly sensitive users.

[0100] (6) Fusion Output Module: This module corresponds to step six in the method and is responsible for the weighted fusion of two dimensions. The module dynamically calculates the fusion weights of the mean and extreme value dimensions based on factors such as the power outage type (planned / faulty), whether the transformer area contains important users, and the dispersion of user sensitivity. Then, it performs a weighted fusion of the scenario-corrected mean and the extreme value risk contribution, and outputs the initial value of the overall sensitivity of the transformer area.

[0101] (7) Hierarchical correction module: (8) This module corresponds to step seven in the method and is responsible for graded correction and final output. The module determines the sensitivity level of the transformer area based on the initial sensitivity value and calls the corresponding grade correction coefficient; after calculating the final sensitivity value, it outputs it to the display interface of the power supply service command system to provide decision-making reference for operation and maintenance dispatchers. At the same time, the module supports extended functions such as sensitivity sorting, graded alarms, and statistical reports.

[0102] The seven modules described above work together to fully realize the calculation and correction function of the overall power outage sensitivity of the distribution network area. The system adopts a B / S architecture, supports web access, and is easy for maintenance personnel at all levels to use. All calculations are completed on the server side, and the client is only responsible for display, so the requirements for terminal performance are low.

[0103] In practical deployment, this system can be integrated as a functional module of the existing power supply service command system, or it can be deployed and operated independently. All input data of the system comes from the existing business system, requiring no additional hardware monitoring equipment, resulting in low construction costs, short deployment cycles, and quick results.

[0104] By applying this system, power supply companies can monitor the overall outage sensitivity of each distribution area in real time, quickly assess the scope and severity of the impact when an outage occurs, rationally allocate the priority of emergency repair and maintenance resources, and prioritize the handling of highly sensitive distribution areas, thereby effectively reducing the complaint rate and improving customer satisfaction and power supply service quality.

Claims

1. A method for calculating and correcting the average sensitivity of a distribution network area to overall power outages, characterized in that, Includes the following steps: S1. Dynamic identification of affected users: Based on the power outage event record, dynamically identify the set of users actually affected by this power outage and exclude unaffected users; S2. Basic Calculation of User Sensitivity: Based on users' historical power outage data, electricity consumption characteristics and request records, calculate the basic value of individual power outage sensitivity for each affected user; S3. Weighted average calculation: Based on the user's electricity consumption category, assign corresponding importance weight coefficients, calculate the weighted average of the individual sensitivities of the affected users, obtain the weighted average sensitivity of the transformer area, and introduce a transformer area size correction coefficient for calibration. S4. Real-time Context Correction: Construct a real-time context correction system that includes time period factors, seasonal factors, holiday factors and current power outage duration factors. Context correction coefficients are obtained through multi-factor coupling calculation, and the weighted sensitivity mean is dynamically corrected. S5. Extreme Value Risk Calculation: Extract the top N most sensitive users in the transformer area, calculate the extreme value risk contribution, and reflect the level of extreme power outage risk in the transformer area. S6. Dual-dimensional fusion: Based on the user composition characteristics and outage type of the transformer area, the fusion weight is dynamically adjusted, and the corrected weighted average and extreme value risk contribution are weighted and fused to obtain the initial value of the overall outage sensitivity of the transformer area. S7. Graded Correction Output: Based on the initial value of overall sensitivity, the transformer area is divided into different sensitivity levels. Different graded correction strategies are adopted for different levels, and the final value of the overall power outage sensitivity of the transformer area is finally output.

2. The method for calculating and correcting the average sensitivity of the overall power outage of a distribution network area according to claim 1, characterized in that, The specific method for dynamically identifying affected users in step S1 is as follows: Based on the power outage records of the meter in the electricity information collection system, the users with valid power outage records within the same transformer area are grouped into an affected user set, with a preset time window before and after the start time of the power outage event as the judgment interval. For record misalignment caused by clock deviation, the average power outage duration sequence of the transformer area is used for clock correction and user completion processing.

3. The method for calculating and correcting the average sensitivity of the overall power outage of a distribution network area according to claim 1, characterized in that, In step S3, the user importance weighting coefficient is set according to the electricity consumption category: The weighting coefficients for large industrial users are 1.5-2.0, for commercial users 1.2-1.5, for general industrial users 1.0-1.2, for residential users 0.8-1.0, and for agricultural production users 0.6-0.

8. The formula for calculating the transformer area size correction factor is: K_scale=1-e (-N / 50) , where N is the total number of users in the region.

4. The method for calculating and correcting the average sensitivity of the overall power outage of a distribution network area according to claim 1, characterized in that, The specific composition of the real-time context factor in step S4 includes: Time-of-use factor: Based on the time of power outage, it is divided into three time-of-use levels: early morning off-peak, daytime flat period, and evening peak, each corresponding to a different correction coefficient; Seasonal factors: Based on the months, the seasons are divided into three levels: winter heating season, summer cooling season, and spring and autumn transition season. Differential correction coefficients are set in combination with electricity consumption categories. Holiday factor: Differentiate between three types of electricity consumption: weekdays, weekends, and statutory holidays, and set differentiated correction coefficients for different types of users; Current power outage duration factor: Divided into multiple intervals according to the duration of the power outage, the longer the power outage duration, the larger the correction coefficient.

5. The method for calculating and correcting the average sensitivity of the overall power outage of a distribution network area according to claim 4, characterized in that, The multi-factor coupling calculation in step S4 adopts a nonlinear coupling method: The context correction coefficient K_context = 1 + α × (K_time-1) + β × (K_season-1) + γ × (K_holiday-1) + δ × (K_duration-1) + ε × (K_time-1) × (K_season-1); where K_time is the time period factor, K_season is the season factor, K_holiday is the holiday factor, and K_duration is the power outage duration factor; α, β, γ, and δ are the main effect weights of each factor, and ε is the interaction effect weight between time period and season.

6. The method for calculating and correcting the average value of overall power outage sensitivity of distribution network areas according to claim 1, characterized in that, The method for calculating the extreme value risk contribution in step S5 is as follows: Extract the top 5% of highly sensitive users within the transformer area, with no fewer than 3 users, and calculate their average sensitivity S_top; extreme value risk contribution S_extreme = S_top × K_concentration, where K_concentration is the concentration coefficient of highly sensitive users, calculated as K_concentration = 1 + (n_top / N_total), where n_top is the number of highly sensitive users and N_total is the total number of affected users.

7. The method for calculating and correcting the average sensitivity of the overall power outage of a distribution network area according to claim 1, characterized in that, The dynamic weight adjustment mechanism for the two-dimensional fusion in step S6 is as follows: When the power outage type is planned power outage, the mean weight β is taken as 0.6-0.7, and the extreme value weight is taken as 0.3-0.4; When the power outage type is fault outage, the mean weight β is taken as 0.4-0.5, and the extreme value weight is taken as 0.5-0.6; When a distribution area contains important power users, the extreme value weight is automatically increased by 0.1-0.2, while the mean weight is reduced accordingly.

8. The method for calculating and correcting the average value of overall power outage sensitivity of distribution network areas according to claim 1, characterized in that, The hierarchical correction strategy in step S7 is specifically as follows: The overall sensitivity of the transformer area is divided into four levels: low sensitivity, moderate sensitivity, high sensitivity, and extreme sensitivity. Low-sensitivity areas are corrected using a baseline correction factor of 1.

0. For moderately sensitive areas, a slight upward adjustment is applied, with a correction factor of 1.05-1.

1. For highly sensitive areas, a moderate upward adjustment is applied, with a correction factor of 1.1-1.

2. Extremely sensitive areas will be subject to a significant upward adjustment, with a correction factor of 1.2-1.

3. The thresholds for each level are dynamically adjusted based on historical complaint data.

9. The method for calculating and correcting the average sensitivity of the overall power outage of a distribution network area according to claim 1, characterized in that, The formula for calculating the baseline value of individual user power outage sensitivity is as follows: S_individual=α×(a×N×b+C+C9×D+H+F); Wherein, α is the electricity consumption coefficient, reflecting the matching degree between the user's electricity consumption and the contracted capacity; a is the scaling parameter, set according to the electricity consumption category; N is the number of power outages in the past 60 days; b is the power outage impact adjustment parameter, which is positively correlated with the number of power outages; C is the comprehensive time period factor, including seasonal factors, intraday time period factors, holiday factors, major event factors, and individual difference characteristic factors; C9 is the power outage type coefficient; D is the current power outage duration impact factor; H is the power outage request frequency parameter; and F is the non-power outage request frequency parameter.

10. The method for calculating and correcting the average sensitivity of the overall power outage of a distribution network area according to claim 1, characterized in that, This method is implemented through a power distribution network service command system, which includes a data acquisition module, a user sensitivity calculation module, a transformer area average value calculation module, a scenario correction module, an extreme value risk assessment module, a fusion output module, and a hierarchical correction module. The data acquisition module is connected to the electricity consumption information acquisition system, the marketing business system, the production management system, and the 95598 work order system to achieve unified access and preprocessing of multi-source data.

Images