Power distribution automation defect dynamic grading system and method

Abstract

The application provides a power distribution automation defect dynamic grading system and method, comprising: collecting basic information of a power distribution automation defect, real-time operation data of a power grid and operation parameters of a defect-related device; based on the basic information of the defect, calculating an initial risk score of the defect through a weighted scoring model, and the weight of each scoring dimension in the weighted scoring model is dynamically adjusted according to the real-time operation data of the power grid; normalizing the operation parameters of the defect-related device, calculating a risk growth factor, combining a risk sensitivity coefficient associated with the initial grade of the defect to adjust the initial risk score with a preset constraint condition, and obtaining a real-time risk index of the defect; identifying other defects that have an electrical association with the current defect, determining a defect coupling influence factor according to the number and grade of the associated defects, and calculating a coupling correction score; and dynamically determining the final grade of the defect according to the real-time risk index and the coupling correction score.

Description

Technical Field

[0001] This invention belongs to the field of power distribution automation technology, specifically relating to a dynamic classification system and method for power distribution automation defects. Background Technology

[0002] Distribution automation, as a core component of smart grid construction, is a key support for achieving lean operation and maintenance of distribution networks and improving power supply reliability. In recent years, with the accelerated digital and intelligent upgrading of the power grid, the number of various field terminal devices such as feeder terminals (FTU), distribution terminals (DTU), and transformer terminals (TTU) in the distribution network has exploded. The monitoring scope of the master station system has also gradually extended from the core urban areas to the urban-rural fringe and rural distribution network areas. During long-term operation, the massive number of terminal devices and the master station system have generated increasingly diverse types of defects and a continuous increase in the frequency of defects due to various factors such as equipment aging, environmental interference, communication anomalies, and changes in power grid operating conditions. As a result, defect management has become a core challenge in the daily operation and maintenance of distribution networks. The traditional distribution automation defect management model still relies on human experience as the core basis for judgment. Under the control of massive defects, its inherent management shortcomings are becoming increasingly prominent, and it can no longer meet the requirements of smart grids for efficient and accurate defect control, urgently requiring upgrading and optimization.

[0003] The core problem with traditional defect management models lies in their overly broad classification standards, which lack relevance to the actual operating scenarios of distribution networks. Currently, most traditional classification methods in the industry simply divide defects into two levels: "general defects" and "serious defects," without differentiating based on the core importance of the defective equipment in the power grid network topology. For critical equipment such as tie switches and main switches, which play the role of power grid load transfer and network hubs, the defects they generate are not distinguished from those of ordinary branch switches. Furthermore, key factors such as the equipment's service life, the frequency of historical defects, and the power grid operating characteristics of the region are not considered. This results in a mismatch between the risk level of the defect and its actual impact on power grid safety and power supply reliability. Valuable operation and maintenance resources cannot be accurately focused on high-risk defects, leading to both lagging management of core equipment defects and excessive resource investment in non-core equipment defects, which violates the principle of precise risk management.

[0004] Traditional defect management also suffers from rigid response strategies. Once a defect is manually classified, it enters a static control process, lacking dynamic assessment and trend analysis of defect risks. On the one hand, it fails to continuously monitor the real-time load rate, voltage deviation rate, environmental meteorological conditions, and other operating parameters of defect-related equipment, making it impossible to promptly perceive the risk escalation caused by changes in power grid conditions. On the other hand, it fails to identify the electrical coupling effect of multiple defects within the same line and protection zone in the distribution network, lacking effective assessment of the linkage risks formed by the superposition of multiple defects. As a result, some defects initially judged as general gradually evolve into serious defects or even faults due to factors such as increased power grid load and the linkage effect of defects. In some cases, it may even trigger a chain of equipment anomalies, further expanding the scope and prolonging the power outage time, seriously affecting the power supply stability of the distribution network.

[0005] At the level of operation and maintenance resource allocation, the traditional resource scheduling logic also lacks scientificity and rationality. It often adopts the simple approach of "assigning orders based on proximity" or "assigning orders according to the time of defect discovery," without coordinating and matching the actual risk level of the defect with the defect elimination time limit, the professional skills and qualifications of the operation and maintenance team, the on-site spare parts inventory, and the current load capacity of the operation and maintenance team. This often results in delays in the defect elimination response of critical defects because high-quality operation and maintenance resources are occupied by non-urgent defects, or waste of resources by assigning general defects to teams with higher professional capabilities. At the same time, traditional defect elimination work orders do not set differentiated priorities, and the operation and maintenance team's work order lacks scientific guidance, which further reduces the overall defect elimination efficiency and keeps the utilization efficiency of operation and maintenance resources at a low level.

[0006] Furthermore, traditional defect management models lack full lifecycle control. Management processes often stop at the on-site defect elimination, lacking quantitative verification and data feedback on the effectiveness of defect elimination. Key data such as defect recurrence rate, actual elimination time, and unit consumption of maintenance resources are not effectively collected and analyzed. This prevents the optimization and iteration of existing defect classification standards and resource scheduling strategies based on actual maintenance data, resulting in a one-way management process rather than a closed-loop management system. This feedback-less management model leads to the long-term rigidity of defect classification and scheduling strategies, failing to adapt to changes in distribution network topology, equipment upgrades, and operating conditions. This significantly contradicts the current smart grid development requirements for lean, intelligent, and continuous optimization of maintenance work. Therefore, constructing a scientific, precise, and adaptive distribution automation defect classification and control technology to achieve a shift from "passive emergency repair" to "proactive prevention and precise control" has become an urgent need for distribution network development. Summary of the Invention

[0007] To address the shortcomings and deficiencies of existing technologies, this invention provides a dynamic defect classification system and method for distribution automation, aiming to achieve refined, dynamic, and intelligent management of defects throughout their entire lifecycle, from discovery and classification to handling and verification. The core of this solution lies in constructing a multi-module collaborative dynamic defect classification and control system. First, a defect acquisition module collects defect information, power grid operation data, and related equipment parameters in real time and performs standardized preprocessing. Then, an intelligent classification module calculates an initial risk score for the defect based on the equipment's critical location in the network topology (such as tie switches and main switches) and the defect type, using a weighted scoring model. The module then dynamically adjusts the weights of each dimension in the scoring model according to real-time power grid conditions (such as peak load periods and equipment service life) to achieve preliminary classification. To overcome the limitations of static grading, the solution further introduces a dynamic grading mechanism: by normalizing the operating parameters of associated equipment (such as load rate, ambient temperature, and voltage deviation), a risk growth factor is calculated. Combined with a risk sensitivity coefficient negatively correlated with the initial defect level, the initial score is adjusted under preset upper and lower limits to obtain an index reflecting real-time risk. Simultaneously, other electrically related defects within the same line or protection domain are identified, and coupling influence factors are determined based on their level and quantity, calculating a coupling correction score. The final defect level is dynamically determined based on the real-time risk index and the coupling correction score. The defect level is upgraded when any of the following conditions are met: real-time risk exceeds a threshold, the coupled score reaches a higher level, the defect times out, or the equipment is in a state where grid safety verification is not met. Based on the final level, the solution matches differentiated defect elimination time limits and uses a lightweight intelligent scheduling model adapted to the real-time requirements of the distribution network to optimize the allocation of operation and maintenance resources. Furthermore, the solution includes short-term risk trend prediction of the operating status of associated equipment, as well as monitoring, effect verification, and data closed-loop feedback throughout the defect elimination process, thereby continuously optimizing the grading model and scheduling strategy. This invention solves the problems of the crudeness and rigidity of traditional defect management in terms of classification standards, dynamic evaluation and resource allocation, and realizes the transformation of operation and maintenance mode from "passive repair" to "proactive prevention".

[0008] The specific technical solution adopted by this invention to solve its technical problem is as follows:

[0009] A dynamic classification method for defects in power distribution automation includes:

[0010] Collect basic information on distribution automation defects, real-time power grid operation data, and operating parameters of defect-related equipment. The basic information on defects includes the location information of the defective equipment in the network topology and the defect type information.

[0011] Based on the basic information of the defect, an initial risk score of the defect is calculated through a weighted scoring model. The weights of each scoring dimension in the weighted scoring model are dynamically adjusted according to real-time power grid operation data.

[0012] The operating parameters of the defect-associated equipment are normalized, and the risk growth factor composed of the load rate, ambient temperature, and voltage deviation rate of the associated equipment is calculated. The initial risk score is adjusted with preset constraints in combination with the risk sensitivity coefficient associated with the initial level of the defect to obtain the real-time risk index of the defect.

[0013] Identify other defects that are electrically associated with the current defect, determine the defect coupling impact factor based on the number and level of associated defects, and calculate the coupling correction score, which is the product of the initial risk score and the defect coupling impact factor;

[0014] The final level of the defect is dynamically determined based on the real-time risk index and the coupled correction score.

[0015] Furthermore, the weighted scoring model includes three scoring dimensions: equipment criticality, defect severity, and network impact. The initial risk score is a weighted sum of the three scoring dimensions, and the sum of the weight coefficients of each dimension is 1. The weights of each dimension are adjusted according to the priority order of the power supply level of important users, the operating characteristics of defective equipment, and the peak and valley loads of the power grid. After adjustment, the sum of the weight coefficients of each dimension is still 1.

[0016] Furthermore, the preset constraints include an upper threshold constraint and a lower protection constraint. The upper threshold is the maximum value between the multiple of the initial risk score and the additional fixed value of the critical defect benchmark score threshold. The lower protection is that the real-time risk index is not lower than the initial risk score. The risk sensitivity coefficient is calibrated differently according to the initial defect level. The higher the initial defect level, the smaller the value of the risk sensitivity coefficient.

[0017] Furthermore, the other defects that are electrically associated with the current defect are defects within the same line or the same protection domain; the defect coupling influence factor is set in a step-by-step quantitative manner according to the level and quantity of associated defects, and the coupling influence factor is 1 when there are no electrically associated defects.

[0018] Furthermore, the final level of the defect is dynamically determined based on the real-time risk index and the coupling correction score. The defect level is raised by one level if any of the following conditions are met: the real-time risk index is greater than or equal to the upgrade threshold of the corresponding defect level; there is an electrical associated defect and the coupling correction score is higher than the score threshold of the higher level; the duration of the defect exceeds the preset tolerance time limit of its initial level; the line where the defective equipment is located is in a state where the N-1 check of the power grid is not met or the area where it is located is a historical high-incidence area of ​​faults.

[0019] Furthermore, the initial risk score is divided into four defect levels, and the four defect levels correspond to a preset scoring threshold range. The scoring threshold range is dynamically calibrated according to the peak and valley load conditions of the power grid. During the peak load period of the power grid, the scoring threshold for high-level defects is lowered, while the scoring threshold for low-level defects remains unchanged.

[0020] Furthermore, based on the final level of the defect, a differentiated defect elimination time limit strategy is matched, and combined with the geographical location of the defect and the characteristics of operation and maintenance resources, a lightweight intelligent scheduling model adapted to the real-time requirements of power distribution automation operation and maintenance is adopted to schedule operation and maintenance resources.

[0021] Furthermore, it also includes short-term risk trend prediction of the operating status of defect-related equipment, using a lightweight time series prediction model to make short-term predictions of the load rate and voltage qualification rate of the related equipment, and generating an early warning signal when the prediction results reach a preset high-risk threshold and directly triggering the upgrade and review of the defect level.

[0022] Furthermore, it also includes full-process supervision of defect elimination, verification of defect elimination effect and closed-loop optimization. The defect elimination process data is classified and fed back to the relevant links of weighted scoring model, real-time risk index calculation and operation and maintenance resource scheduling, and the model parameters and quantification coefficients of each link are iteratively optimized.

[0023] Furthermore, a dynamic classification system for distribution automation defects, used to execute the dynamic classification method for distribution automation defects as described above, includes a defect acquisition module, an intelligent classification module, a dynamic classification module, a resource matching module, and a closed-loop monitoring module connected by communication, wherein:

[0024] The defect acquisition module is used to collect basic information on distribution automation defects, real-time power grid operation data, and operating parameters of defect-related equipment, and to perform standardized processing.

[0025] The intelligent grading module is used to calculate the initial risk score of defects through a weighted scoring model and dynamically adjust the weights of each dimension based on real-time power grid operation data.

[0026] The dynamic upgrade module is used to normalize the operating parameters of defect-related equipment, calculate the real-time risk index, identify electrical-related defects and determine the coupling influence factor, calculate the coupling correction score, and finally dynamically determine the final defect level based on the real-time risk index and the coupling correction score.

[0027] The resource matching module is used to match the defect elimination time limit according to the final defect level and schedule operation and maintenance resources through a lightweight intelligent scheduling model.

[0028] The closed-loop supervision module is used to supervise the entire defect elimination process, verify its effectiveness, and realize closed-loop feedback of defect elimination data and iterative optimization of model parameters.

[0029] And a computer device including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the method described above.

[0030] A non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described above.

[0031] Compared with the prior art, the present invention and its preferred embodiments have at least the following beneficial effects:

[0032] 1. Achieved precise and dynamically adaptive defect classification: By constructing a multi-dimensional quantitative scoring model that integrates equipment criticality, defect severity, and network impact, and dynamically adjusting weights based on real-time power grid operating conditions, the defect classification results are made more closely aligned with the actual risks of the power grid. Furthermore, the introduction of a real-time risk index coupled with a defect analysis mechanism enables keen perception of changes in operating status and potential cascading risks, allowing for dynamic review and upgrading of defect levels, effectively overcoming the lag and crudeness of traditional static classification methods.

[0033] 2. Improved the intelligence and efficiency of operation and maintenance resource scheduling: By combining dynamically determined final defect levels with differentiated defect elimination time limits and a lightweight intelligent scheduling model, precise assignment of operation and maintenance tasks and efficient optimization of resource allocation are achieved. This ensures rapid response to high-priority defects while taking into account the overall load balancing and utilization efficiency of operation and maintenance resources, thus optimizing operation and maintenance costs at the system level.

[0034] 3. A complete management closed loop and continuous optimization capability have been formed: Through a closed-loop monitoring and feedback mechanism, data from the entire defect elimination process is used to iteratively optimize the hierarchical model, upgrade algorithm, and scheduling strategy, enabling the system to have the ability to learn and continuously improve itself. This not only ensures the effectiveness of individual defect handling but also improves the adaptability and intelligence level of the entire defect management system in the long run, promoting a fundamental shift in the operation and maintenance model from reactive repair to proactive prevention. Attached Figure Description

[0035] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0036] Figure 1 This is a flowchart illustrating the implementation of the dynamic classification scheme for power distribution automation defects in an embodiment of the present invention. Detailed Implementation

[0037] To make the features and advantages of the present invention more apparent and understandable, specific embodiments are described below in detail:

[0038] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0039] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0040] This invention aims to overcome the shortcomings of existing technologies and provide a dynamic classification system and method for defects in power distribution automation. It includes five core modules: defect acquisition, intelligent classification, dynamic upgrading, resource matching, and closed-loop monitoring. Through the collaboration of multiple modules, it realizes full lifecycle management of defects from discovery, classification, handling to verification, so as to achieve refined, dynamic and intelligent defect management.

[0041] Its systematic implementation includes a defect acquisition module, an intelligent classification module, a dynamic escalation module, a resource matching module, and a closed-loop monitoring module. The method includes: collecting distribution network automation defect information; dynamically classifying defects into four levels—critical, severe, Class I general, and Class II general—based on switch location, defect type, and impact on grid operation; dynamically escalating the escalation based on defect coupling relationships and real-time risks; intelligently matching defect elimination resources and issuing differentiated time-limited work orders; and conducting closed-loop monitoring and evaluation of the entire defect elimination process. This invention solves the problems of traditional, extensive, delayed, and resource mismatched defect management, achieving precise, proactive, and efficient defect handling, significantly improving the reliability of distribution network power supply and the level of lean operation and maintenance.

[0042] like Figure 1 As shown, the implementation process of the embodiment of the present invention includes the following steps:

[0043] Step S1: The defect acquisition module collects information on distribution automation defects in real time, including real-time acquisition of the operating status, alarm information and abnormal data of the distribution network and terminals.

[0044] As a preferred implementation, the defect acquisition module also performs standardized preprocessing and feature fusion on the acquired multi-source data. In addition to the operating status of the distribution network and terminals, alarm information, and abnormal data, the acquired data also includes the commissioning years of the defective equipment, the frequency of historical defects, the peak and valley load periods of the power grid in the area, the power supply association level of important users, and real-time environmental meteorological data (such as extreme weather indicators such as rainstorms, high temperatures, and typhoons). During data preprocessing, missing values ​​are filled with the average value of similar equipment in the distribution network during the same period, outliers are removed using the 3σ criterion, and time-series data is normalized at a 1-minute granularity. Finally, a standardized defect dataset containing basic equipment attributes, operating status, environmental associations, and power grid conditions is generated, providing comprehensive and accurate feature data support for subsequent intelligent classification.

[0045] Step S2: The intelligent grading module calls the multi-dimensional quantitative scoring model based on the switch location of the terminal and the defect type to preliminarily determine the defect level;

[0046] Furthermore, the intelligent grading module calculates the comprehensive score of the defect by calling the multi-dimensional quantitative scoring model, and preliminarily determines the defect level as critical, serious, Class I general, or Class II general based on the preset scoring threshold range; the switch positions include tie switches, main switches, and other switches; the defect types include at least protection failure to operate, protection malfunction, remote control failure, and other abnormal defects.

[0047] Furthermore, the multidimensional quantitative scoring model in step S2 operates as follows:

[0048] Equipment criticality dimension score (C1): The defective equipment (switch) is assigned a value based on its position and function in the network topology. Among them, the tie switch is assigned the highest score A1, the main switch is assigned a score A2, and the branch switch is assigned a score A3, and A1>A2>A3. If the equipment supplies power to a specific important user, an additional fixed score ΔC is added.

[0049] Defect Severity Dimension Score (C2): A base score is preset based on the defect type; among them, defects that directly affect the correct operation of protection, such as protection failure or maloperation, are assigned the highest score B1; defects that affect control, such as remote control failure or communication interruption, are assigned a score of B2; defects such as data anomalies or minor alarms are assigned a score of B3, and B1>B2>B3.

[0050] Network Impact Dimension Score (C3): The impact is assessed based on the network topology analysis at the time the defect occurs; high scores are assigned for defects that cause non-N-1 transmission, load loss, or affect important power supply areas; low scores are assigned for defects that only affect a local area and can be automatically isolated.

[0051] The overall score is as follows:

[0052] S = α×C1 + β×C2 + γ×C3,

[0053] Where α, β, and γ are the weight coefficients for each dimension, and α + β + γ = 1.

[0054] As a preferred implementation, the weight coefficients α, β, and γ for each dimension are not fixed values, but are dynamically adjusted based on the real-time operating conditions of the power grid and the operating characteristics of the equipment. Specifically, during peak load periods of the power grid (such as 8-10 am and 6-9 pm), the weight γ for the network impact dimension is increased by 0.1-0.2, while the corresponding value of the weight α for the equipment criticality dimension is decreased to ensure that α+β+γ=1. If the defective equipment has been in operation for more than 15 years or the frequency of similar defects occurring ≥3 times in the past year, the weight β for the defect severity dimension is increased by 0.05-0.1. If the defective equipment supplies power to first-level important users such as hospitals, municipalities, and data centers, the weight α for the equipment criticality dimension is increased by 0.1. In the equipment criticality dimension score C1, the additional score ΔC is set according to the level of important users: ΔC=10 for first-level important users, ΔC=5 for second-level important users, ΔC=2 for third-level important users, and ΔC=0 for no important users, so that the comprehensive score is more in line with the risk impact characteristics of the actual operation of the distribution network.

[0055] As a further optimized solution, to achieve precise dynamic adjustment of the weighting coefficients, the baseline weights α0, β0, and γ0 are first calibrated (recommended values ​​are α0=0.4, β0=0.4, and γ0=0.2, satisfying α0+β0+γ0=1). Then, quantitative correction coefficients are designed for three scenarios: peak-valley load, equipment operating characteristics, and power supply for important users. The priority of scenario superposition is defined as: power supply for important users > equipment operating characteristics > peak-valley load. The weight correction and final calculation for each scenario are performed according to the following rules, and the correction always satisfies α+β+γ=1.

[0056] 1. Peak-valley load correction: Only the weights of the criticality dimension α of intermodulation equipment and the network impact dimension γ are adjusted, while the defect severity dimension β remains unchanged; during peak load periods of the power grid (regional load rate ≥ 80%), Δγ1∈[0.1,0.2], Δα1=-Δγ1, Δβ1=0; during off-peak load periods (regional load rate < 80%), all correction coefficients are 0, and the weights are restored to the baseline values.

[0057] 2. Equipment operation characteristic correction: Only the defect severity dimension β is positively adjusted, while α and γ are reduced proportionally according to the baseline weight ratio; if the equipment has been in operation for more than 15 years or the frequency of the same type of defect in the past year is ≥3 times, Δβ2∈[0.05,0.1], Δα2=-Δβ2×α0 / (α0+γ0), Δγ2=-Δβ2×γ0 / (α0+γ0); if not satisfied, the correction coefficient is 0;

[0058] 3. Power supply correction for important users: Only adjust the critical dimensions α of the device positively, and β and γ are reduced proportionally according to the reference weight ratio; for first-level important users, Δα3 = 0.1, for second-level important users, Δα3 = 0.05, and for third-level and below / no important users, Δα3 = 0; corresponding Δβ3 = -Δα3 × β0 / (β0 + γ0), Δγ3 = -Δα3 × γ0 / (β0 + γ0);

[0059] The general formula for the final calculation of weights is:

[0060] α = α0 + Δα3 + Δα2 + Δα1,

[0061] β = β0 + Δβ3 + Δβ2 + Δβ1,

[0062] γ = γ0 + Δγ3 + Δγ2 + Δγ1;

[0063] If a certain weight ≤ 0 after correction, the correction coefficient of the corresponding scenario will be adjusted down to a value just greater than 0 for the weight.

[0064] Preset scoring thresholds: S ≥ T1 is a critical defect; T2 ≤ S < T1 is a serious defect; T3 ≤ S < T2 is a Class I general defect; S < T3 is a Class II general defect.

[0065] As a preferred implementation, the grading thresholds T1, T2, and T3 will be dynamically calibrated according to the peak-valley load conditions of the distribution network. During the peak load period of the power grid, T1 will be reduced by 5% - 8%, T2 will be reduced by 3% - 5%, and T3 will remain unchanged, making the grading determination of defects more stringent during the peak period and promptly identifying the risk of low-score defects under the high-load state of the power grid; during the flat-valley load period of the power grid, the thresholds will be restored to the reference values, and the calibration amplitude of the thresholds will be adjusted in real time according to the load rate of the power grid in the region. When the overall load rate of the regional power grid ≥ 80%, it will be reduced according to the above amplitude, and when the load rate < 80%, no threshold adjustment will be made, achieving the dynamic matching of defect grading and the overall operation load of the power grid.

[0066] Step S3: The dynamic upgrading module combines the real-time operation risk of the power grid, defect relevance, and historical data to conduct a dynamic upgrading review of the initially determined defect level and generate the final defect level; based on the calculation of the real-time risk index and defect coupling analysis, this dynamic upgrading module conducts a dynamic review of the preliminary level in step S2, and conducts a dynamic upgrading assessment of the levels of defects that have been determined to cause FA (Feeder Automation) failure and are found during the morning automation operation process. If the upgrading conditions are met, confirm the final defect level;

[0067] Furthermore, the dynamic upgrading algorithm in step S3 operates as follows:

[0068] Real-time risk index calculation: For defects that have been preliminarily classified, parameters such as the load rate L(t), ambient temperature H(t), and voltage deviation rate U(t) of their associated equipment are monitored in real time. After dimensionless normalization, the risk growth factor R(t) is calculated, and a constrained real-time risk index RI = S × [1 + η·R(t)] is defined, where η is the risk sensitivity coefficient. To avoid abnormal increases in RI leading to distortion in the upgrade judgment, a dual constraint design is implemented for η and RI, and a lower limit protection for RI is set. The specific rules are as follows:

[0069] Risk sensitivity coefficient η value constraints: η is differentiated according to the defect level and always satisfies η∈(0,0.5]. Recommended values ​​are: critical defect η=0.2, serious defect η=0.25, Class I general defect η=0.3, Class II general defect η=0.35; the higher the defect level, the smaller the value of η, to avoid excessive amplification of RI for high-level defects.

[0070] RI Upper Threshold Constraint: RI must not exceed 1.5 times the highest threshold of the current level, and must not exceed the critical defect baseline threshold T1+10 (e.g., if T1=90, then RI≤100), that is:

[0071] RI=min(S×[1+η·R(t)],max(1.5×S,T1+10));

[0072] RI lower limit protection constraint: If the calculated result RI < S, then RI = S is taken. That is, when the risk decreases, RI is not lower than the original comprehensive score S, so as to avoid the situation where the defect level decreases in the opposite direction, which is in line with the core design logic of dynamic upgrading. After the above constraint, RI always fluctuates slightly around the original comprehensive score S, and the upgrading judgment result is highly consistent with the actual risk level of the power grid.

[0073] Defect Coupling Analysis: Identify other defects that are electrically or logically related to the current defect (such as the same line, the same busbar, or the same protection zone). If multiple defects exist, assess their cumulative effect. Define the coupling influence factor K.

[0074] As a preferred implementation, the dynamic upgrade module also integrates a short-term risk trend prediction submodule for distribution network scenarios. It uses a lightweight time series prediction model (such as ARIMA or LSTM) to predict the load rate and voltage qualification rate of associated equipment for the next 15-30 minutes. The prediction incorporates the operating status of other equipment on the same distribution network line as constraints. When the prediction results show that the load rate of associated equipment will be ≥95% and the voltage qualification rate ≤90% within the next 15 minutes, a high-risk warning signal is generated. Simultaneously, in the calculation of the risk growth factor R(t), a new grid voltage deviation rate U(t) is added as a correction term. First, the load rate L(t), ambient temperature H(t), and voltage deviation rate U(t) are normalized to a dimensionless range of 0-1. Then, the corrected risk growth factor is calculated through a weighted sum.

[0075] R(t)=wL·Lnorm(t)+wH·Hnorm(t)+wU·Unorm(t),

[0076] Where wL=0.5, wH=0.3, and wU=0.2, to satisfy wL+wH+wU=1 and R(t)∈[0,1]; the normalization rules for each parameter are as follows:

[0077] Load rate Lnorm(t): Convert the actual load rate percentage to a dimensionless value of 0 to 1, i.e.: Lnorm(t) = L(t) / 100. If the result is greater than 1, take 1 (exceeding the safety threshold of the distribution network).

[0078] Ambient temperature Hnorm(t): Linearly normalized based on the national standard operating temperature range of power distribution terminal equipment (reference value is -10℃ to 50℃).

[0079] Hnorm(t)=(H(t)-Hmin) / (Hmax-Hmin),

[0080] If H(t)≤Hmin, take the value 0; if H(t)≥Hmax, take the value 1.

[0081] Voltage deviation rate Unorm(t): Converts the actual voltage deviation percentage into a dimensionless value of 0~1, i.e.: Unorm(t)=U(t) / 100. If the result is >1, then take 1 (exceeding the national standard deviation threshold for distribution networks). Voltage deviation rate U(t) is the percentage deviation between the actual operating voltage and the rated voltage of the associated equipment, making the calculation of risk growth factor more comprehensively reflect the operating status of the power grid.

[0082] In addition, for the high-risk warning signals generated by the prediction, a pre-trigger condition for upgrading the judgment is set. If the warning signal is red (load rate ≥ 95% and voltage qualification rate ≤ 90%), the upgrading review is directly triggered without needing to meet the threshold requirements of other upgrading conditions.

[0083] Promotion determination logic:

[0084] Condition 1: If RI≥T_up (the upgrade threshold, which is higher than the original level's corresponding T1 / T2 / T3), then the upgrade to the next level is triggered.

[0085] Condition 2: If defect coupling exists, and the virtual comprehensive score S' (S'=S×K) after coupling is greater than or equal to the threshold of a higher level, then a level up is triggered.

[0086] Condition 3: If the duration of the defect t_d exceeds the preset tolerance time limit T_tol for this level (for example, a Class II general defect that has not been processed for more than 15 days), it will be automatically upgraded to the next level.

[0087] Dynamic upgrades can be triggered if any one of the conditions is met.

[0088] As a preferred implementation, the upgrade judgment logic may also include condition four: if the line where the defective equipment is located is in a state where the N-1 check of the power grid is not met, or the area where it is located is a high-incidence area of ​​distribution network faults (fault frequency ≥ 5 times / 100 kilometers in the past 3 months), then the defect level is automatically upgraded by one level; at the same time, the coupling influence factor K is quantitatively defined, and the K value is set according to the number and level of coupled defects. If there is one serious defect in the same line, K=1.2; if there is one critical defect in the same line, K=1.5; if there are two or more general defects in the same protection domain, K=1.1; if there are no coupled defects, K=1. This makes the calculation of the virtual comprehensive score S' after coupling clear and feasible quantitative standard, avoiding the ambiguity of coupling analysis.

[0089] Step S4: The resource matching module intelligently schedules maintenance resources based on the final level of the defect and its geographical location, and generates work orders that include differentiated defect elimination time limits.

[0090] Specifically, the resource matching module matches the corresponding defect elimination time limit strategy based on the final defect level and the calculation results of the resource optimization scheduling model, and generates and dispatches differentiated defect elimination work orders; among them, the defect elimination time limit for critical defects is 24 hours, for serious defects it is 7 days, for Class I general defects it is 15 days, and for Class II general defects it is 30 days.

[0091] The optimization objective of the model is:

[0092] Minimize(Σw_i×D_i),

[0093] Where w_i is the weight of defect level i (critical is the highest), and D_i is the actual delay in eliminating defect i; at the same time, it is supplemented by minimizing the team load variance and maximizing the spare parts inventory turnover rate.

[0094] Constraints:

[0095] Time constraints: The actual time for defect elimination must be within the maximum time limit specified for the corresponding level (24 hours for critical cases, 7 days for serious cases, etc.).

[0096] Skills constraint: The assigned work team must possess the skills and qualifications required to handle this type of defect.

[0097] Geographical constraints: Consider the distance between the current location of the work team and the defect point, and prioritize dispatching orders to the nearest location.

[0098] Load constraint: The number of uncompleted work orders in a work group must be lower than its maximum load capacity.

[0099] Inventory constraint: The inventory of critical spare parts required for defect elimination in the designated warehouse must exceed the demand.

[0100] As a preferred implementation, the resource optimization scheduling model is solved using a lightweight Deep Q-Network (DQN) reinforcement learning algorithm for distribution network scenarios. The state space of this algorithm is quantized into a 6-dimensional feature vector, specifically: [Defect level quantification value (critical = 4 / severe = 3 / Class I general = 2 / Class II general = 1), distance between defect point and work group (km), number of uncompleted work orders in the work group, work group skill qualification matching degree (complete match = 1 / partial match = 0.5 / mismatch = 0), spare parts inventory sufficiency (sufficient = 1 / insufficient = 0), regional traffic congestion index (0-1)]. The action space is the work order dispatch decision for each maintenance work group, and the reward function is:

[0101] R = ω1R1 + ω2R2 + ω3R3 - ω4R4

[0102] Among them, R1 is the defect elimination timeliness rate (on-time completion = 1 / not on time = 0), R2 is the resource utilization rate (number of defects handled by the team in a single defect elimination / maximum processing capacity of the team), R3 is the power supply restoration rate (proportion of power supply restored in the affected area after defect elimination), and R4 is the operation and maintenance cost coefficient (actual operation and maintenance cost / standard operation and maintenance cost). The weight coefficients ω1=0.4, ω2=0.2, ω3=0.2, ω4=0.2, and ω1+ω2+ω3+ω4=1. The intelligent agent iteratively updates the strategy network through the actual reward value of each dispatch, realizing the precise scheduling of regional operation and maintenance resources of the distribution network. At the same time, the lightweight DQN algorithm retains only 1 hidden layer (number of neurons = 32), which is suitable for the real-time requirements of distribution network operation and maintenance scheduling.

[0103] The model solves the problem using heuristic algorithms or integer programming, and outputs the recommended work group, dispatch time, estimated completion time, and required spare parts requisition list for each defect.

[0104] As a preferred implementation, the resource matching module will also assign differentiated priority scores to the generated defect elimination work orders. The priority scores are as follows:

[0105] P = 0.6P1 + 0.3P2 + 0.1 × P3

[0106] Wherein, P1 is the quantified value of defect level (critical = 4 / serious = 3 / Class I general = 2 / Class II general = 1), P2 is the quantified value of the number of users affected by the defect (affected users ≥ 1000 = 4 / 500-999 = 3 / 100-499 = 2 / < 100 = 1), and P3 is the quantified value of the distance between the defect point and important users (≤ 1km = 4 / 1-3km = 3 / 3-5km = 2 / 5km = 1). Work orders are sorted from high to low priority score, and maintenance teams perform defect elimination operations according to the sorting results. If multiple work orders have the same priority score, they are sorted according to the time of defect discovery. At the same time, the priority score is synchronized to the dispatch screen of the distribution automation master station in real time to realize the visualization and orderliness of work order dispatch.

[0107] Step S5: The closed-loop monitoring module monitors the entire process of the defect elimination work order and performs effect verification and closed-loop archiving after the defect elimination is completed. It is used to track, monitor and evaluate the execution status and quality of the defect elimination work order throughout the entire process.

[0108] As a preferred implementation, the closed-loop monitoring module also constructs a distribution network defect elimination knowledge graph based on the entire defect elimination process data. The knowledge graph nodes include six categories: equipment type, defect type, elimination team, elimination measures, elimination duration, and defect cause. Edges represent the relationships between nodes and are assigned association weights (calculated based on association frequency). Simultaneously, the closed-loop defect elimination data is categorized and fed back according to three modules: intelligent grading, dynamic upgrading, and resource matching. For the intelligent grading module, the matching degree between the actual defect elimination results and the grading results is fed back. If the matching degree is <80%, a secondary calibration of the grading model weights and thresholds is triggered. For the dynamic upgrading module… The system provides feedback on the accuracy of risk escalation judgments and the precision of risk predictions, optimizing the quantitative coefficients of risk growth factors and coupled influencing factors. For the resource matching module, it provides feedback on the timeliness of defect elimination and resource utilization in dispatch decisions, iteratively updating the reward function weights of the reinforcement learning scheduling model. In addition, it sets quantitative evaluation indicators for defect elimination effectiveness, including defect recurrence rate (the proportion of similar defects recurring within 3 months after elimination), average defect elimination time, and unit consumption of operation and maintenance resources. If the defect recurrence rate is ≥10%, a secondary analysis of defect causes is initiated and added to the knowledge graph, achieving a closed-loop iterative optimization of defect management data, model, and strategy.

[0109] Table 1 is a defect level display table of the dynamic defect grading system and method for power distribution automation. It is a specific quantitative correspondence of the grading rules based on the two dimensions of switch position and defect type. It clarifies the initial grading standards and corresponding defect elimination time benchmarks for various types of defects under different switch positions, and provides a direct reference for the grading judgment of the intelligent grading module.

[0110] Table 1 Defect Level Table

[0111]

[0112] Based on the design of this embodiment, in the overall dynamic classification system for power distribution automation defects, the actions performed by each module include:

[0113] The defect acquisition module obtains various alarms, abnormal status variables, communication interruptions, and other information in real time from the power distribution automation master station system and terminal equipment through the data interface, and cleans and formats the information to generate standard defect records.

[0114] The intelligent grading module receives standard defect records. Its built-in multi-dimensional quantitative scoring model initiates calculations. For example, a protection failure to operate (C2=B1=95 points) occurring at a tie switch (C1=A1=90 points) will, according to topology analysis, result in the inability to transfer loads to two downstream lines when the switch is disconnected (C3=85 points). Assuming weights α=0.4, β=0.4, γ=0.2, the comprehensive score is: S=0.4×90+0.4×95+0.2×85=92 points. With preset thresholds T1=90, T2=70, T3=40, this defect is initially classified as "critical."

[0115] The dynamic escalation module continuously monitors this "critical" defect. It calculates the real-time load rate of its associated lines. Assuming the load rate consistently exceeds 90%, the risk growth factor R(t) is high, causing the real-time risk index RI to reach 98 points, exceeding the lower threshold for maintaining the critical level (assumed to be >95 points), but not reaching the escalation threshold T_up (assumed to be 100 points), therefore condition one is not triggered. Upon inspection, no strongly coupled defects are found (condition two is not met). The defect is processed within 24 hours, without timeout (condition three is not met). Therefore, the defect's final level remains "critical".

[0116] The resource matching module receives defect tasks that are ultimately classified as "critical." Its built-in resource optimization scheduling model begins operation. The model retrieves available work teams from the database: Team X possesses protection and debugging skills, is currently located 10 kilometers away, and has one outstanding work order; Team Y also possesses the skills, is located 5 kilometers away, and has no outstanding work orders. Spare parts inventory is sufficient. After calculation, the model selects and dispatches Team Y, generates a defect resolution work order requiring completion within 24 hours, and automatically associates it with the required spare parts list.

[0117] The closed-loop monitoring module tracks the work order status: dispatch, receipt, departure, arrival, processing, and completion. After defect elimination, the system automatically or manually verifies whether the defect has been eliminated, and the relevant data (including actual time taken, resources used, and verification results) is archived and can be used for subsequent model parameter tuning.

[0118] As a preferred embodiment, the specific implementation process of this invention during peak load periods of the power grid is as follows: Taking the remote control failure defect of the main switch as an example, the defective equipment has been in operation for 12 years, supplies power to secondary important users, and the power grid load rate is 85% during peak periods. At this time, the equipment criticality dimension C1=80+ΔC=85, the defect severity dimension C2=B2=70, and the network impact dimension C3=80. Due to the dynamic adjustment weights α=0.35, β=0.45, and γ=0.2 during peak periods, the comprehensive score S=0.35×85+0.45×70+0.2×80=77.25. The peak period thresholds T1 and T2 are lowered to 85 and 66, respectively, so the defect is initially judged as a serious defect. The dynamic upgrade module calculates L(t)=88% and H(t)=35℃. U(t) = 5%, after correction R(t) = 0.5 × 0.88 + 0.3 × 0.35 + 0.2 × 0.05 = 0.555, RI = 77.25 × (1 + 0.02 × 0.555) = 77.99, which does not reach the upgrade threshold. However, the line where the defect is located is in an N-1 check failure state, triggering upgrade condition four. Therefore, it is finally upgraded to a critical defect. The resource matching module assigns it a priority score P = 0.64 + 0.33 + 0.13 = 3.6. The lightweight DQN algorithm is used to schedule the nearest team with remote debugging skills to generate a 24-hour defect elimination work order. After the defect elimination is completed, the closed-loop supervision module reports a 100% matching degree and adds the defect elimination measures (replacing the remote control module) to the knowledge graph to complete the full process closed loop.

[0119] Compared with existing technologies, the advantages of the solution provided by this invention include:

[0120] (1) Scientific and precise classification: An innovative classification rule base based on the dual dimensions of "switch position + defect type" was constructed, and the defects were refined into four levels: "critical, serious, Class I general, and Class II general". In particular, the principle of upgrading the classification of defects of tie switches and main switches was clarified, so that the classification results are highly consistent with the degree of impact on power grid safety.

[0121] (2) Dynamic adaptive control: The introduction of a dynamic upgrade mechanism enables dynamic adjustment of defect levels based on multi-defect coupling and real-time operational risks, effectively warning of potential chain risks and realizing a leap from static classification to dynamic risk assessment.

[0122] (3) Efficient resource allocation: The defect level is bound to a strict defect elimination time limit (24 hours, 7 days, 15 days, 30 days), and the operation and maintenance resources are intelligently matched to achieve refined resource scheduling of "rapid response to important defects and planned handling of general defects", which greatly improves the operation and maintenance efficiency.

[0123] (4) Traceable management loop: Through full-process supervision and effect verification, a complete management loop of "discovery-evaluation-disposal-verification" is formed to ensure that every defect is effectively eliminated and to provide data support for the continuous optimization of operation and maintenance strategies.

[0124] Based on the same inventive concept, this invention also provides a computer device, comprising: one or more processors, and a memory for storing one or more computer programs; the programs include program instructions, and the processor executes the program instructions stored in the memory. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, used to implement one or more instructions, specifically for loading and executing one or more instructions stored in a computer storage medium to implement the above-described method.

[0125] It should be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium storing a computer program, which, when executed by a processor, performs the above-described method. This storage medium can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0126] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0127] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

[0128] This invention is not limited to the preferred embodiment described above. Anyone inspired by this invention can derive other forms of dynamic classification systems and methods for power distribution automation defects. All equivalent changes and modifications made within the scope of the claims of this invention shall fall within the scope of this invention.

Claims

1. A dynamic classification method for defects in power distribution automation, characterized in that, include: Collect basic information on distribution automation defects, real-time power grid operation data, and operating parameters of defect-related equipment. The basic information on defects includes the location information of the defective equipment in the network topology and the defect type information. Based on the basic information of the defect, an initial risk score of the defect is calculated through a weighted scoring model. The weights of each scoring dimension in the weighted scoring model are dynamically adjusted according to real-time power grid operation data. The operating parameters of the defect-associated equipment are normalized, and the risk growth factor composed of the load rate, ambient temperature, and voltage deviation rate of the associated equipment is calculated. The initial risk score is adjusted with preset constraints in combination with the risk sensitivity coefficient associated with the initial level of the defect to obtain the real-time risk index of the defect. Identify other defects that are electrically associated with the current defect, determine the defect coupling impact factor based on the number and level of associated defects, and calculate the coupling correction score, which is the product of the initial risk score and the defect coupling impact factor; The final level of the defect is dynamically determined based on the real-time risk index and the coupled correction score.

2. The method for dynamic classification of defects in power distribution automation according to claim 1, characterized in that: The weighted scoring model includes three scoring dimensions: equipment criticality, defect severity, and network impact. The initial risk score is a weighted sum of the three scoring dimensions, and the sum of the weight coefficients of each dimension is 1. The weights of each dimension are adjusted according to the priority order of the power supply level of important users, the operating characteristics of defective equipment, and the peak and valley load of the power grid. After adjustment, the sum of the weight coefficients of each dimension is still 1.

3. The method for dynamic classification of defects in power distribution automation according to claim 1, characterized in that: The preset constraints include an upper threshold constraint and a lower protection constraint. The upper threshold is the maximum value between the multiple of the initial risk score and the additional fixed value of the critical defect benchmark score threshold. The lower protection is that the real-time risk index is not lower than the initial risk score. The risk sensitivity coefficient is calibrated differently according to the initial defect level. The higher the initial defect level, the smaller the value of the risk sensitivity coefficient.

4. The method for dynamic classification of defects in power distribution automation according to claim 1, characterized in that: Other defects that are electrically related to the current defect are defects within the same line or the same protection domain; The defect coupling influence factor is set in a stepwise manner according to the level and quantity of associated defects. When there are no electrical associated defects, the coupling influence factor is 1.

5. The method for dynamic classification of defects in power distribution automation according to claim 1, characterized in that: The final level of the defect is dynamically determined based on the real-time risk index and the coupling correction score. The defect level is raised by one level if any of the following conditions are met: the real-time risk index is greater than or equal to the upgrade threshold of the corresponding defect level; or there is an electrical associated defect and the coupling correction score is higher than the score threshold of the higher level. The defect lasts longer than the preset tolerance time limit of its initial level; the line where the defective equipment is located is in a state where the N-1 check of the power grid is not met or the area where it is located is a historically high-incidence area of ​​faults.

6. The method for dynamic classification of defects in power distribution automation according to claim 1, characterized in that: The initial risk score is divided into four defect levels, which correspond to preset scoring threshold ranges. The scoring threshold ranges are dynamically calibrated according to the peak and valley load conditions of the power grid. During peak load periods, the scoring thresholds for high-level defects are lowered, while the scoring thresholds for low-level defects remain unchanged.

7. The method for dynamic classification of defects in power distribution automation according to claim 1, characterized in that: Based on the final level of the defect, a differentiated defect elimination time limit strategy is matched. In combination with the geographical location of the defect and the characteristics of operation and maintenance resources, a lightweight intelligent scheduling model adapted to the real-time requirements of power distribution automation operation and maintenance is adopted to schedule operation and maintenance resources.

8. The method for dynamic classification of defects in power distribution automation according to claim 1, characterized in that: It also includes short-term risk trend prediction of the operating status of defect-related equipment, using a lightweight time series prediction model to make short-term predictions of the load rate and voltage qualification rate of related equipment, and generating an early warning signal when the prediction results reach the preset high-risk threshold and directly triggering the upgrade review of the defect level.

9. The method for dynamic classification of defects in power distribution automation according to claim 1, characterized in that: It also includes full-process supervision of defect elimination, verification of defect elimination effect and closed-loop optimization. The defect elimination process data is classified and fed back to the relevant links of weighted scoring model, real-time risk index calculation and operation and maintenance resource scheduling, and the model parameters and quantification coefficients of each link are iteratively optimized.

10. A dynamic classification system for distribution automation defects, used to execute the dynamic classification method for distribution automation defects as described in any one of claims 1 to 9, characterized in that, It includes a defect acquisition module for communication connections, an intelligent grading module, a dynamic grading module, a resource matching module, and a closed-loop monitoring module, among which: The defect acquisition module is used to collect basic information on distribution automation defects, real-time power grid operation data, and operating parameters of defect-related equipment, and to perform standardized processing. The intelligent grading module is used to calculate the initial risk score of defects through a weighted scoring model and dynamically adjust the weights of each dimension based on real-time power grid operation data. The dynamic upgrade module is used to normalize the operating parameters of defect-related equipment, calculate the real-time risk index, identify electrical-related defects and determine the coupling influence factor, calculate the coupling correction score, and finally dynamically determine the final defect level based on the real-time risk index and the coupling correction score. The resource matching module is used to match the defect elimination time limit according to the final defect level and schedule operation and maintenance resources through a lightweight intelligent scheduling model. The closed-loop supervision module is used to supervise the entire defect elimination process, verify its effectiveness, and realize closed-loop feedback of defect elimination data and iterative optimization of model parameters.

Images