A power failure event intelligent diagnosis method and system based on big data analysis

Abstract

A method and system for intelligent diagnosis of power outage events based on big data analysis is proposed. This method monitors switch tripping signals, user-side power loss data, and customer repair order data in real time to determine if a power outage event is identified. If a power outage event is identified, the distribution network topology is traced based on the user-side power loss data or customer repair order data to determine the initial fault area. If a power outage event is identified and a switch tripping signal exists, candidate fault areas are determined based on the switch tripping signal and a preset distribution network topology. The initial and candidate fault areas are verified to determine the core fault area. Within the core fault area, spatiotemporal correlation diagnosis is performed based on multi-source time-series data to output the fault type. Historical fault mode matching is performed based on the core fault area and fault type to obtain handling strategy data and generate a structured diagnostic report. Simultaneously, proactive risk warnings are provided, achieving intelligent and accurate diagnosis of power outage events.

Description

Technical Field

[0001] This invention belongs to the field of intelligent operation and maintenance technology of power systems, and particularly relates to an intelligent diagnosis method and system for power outage events based on big data analysis. Background Technology

[0002] Rapid and accurate power outage diagnosis (including outage determination, fault location, fault type identification, and response strategy development) is a key requirement for power operation and maintenance management. However, existing power outage diagnosis technologies have several limitations: First, outage determination relies on a single data dimension. Some solutions depend solely on customer repair work orders (which are prone to misjudgment due to interference from non-outage repair requests) or solely on switch trip signals (which are prone to missed judgments due to signal delays and equipment failures), lacking a quantitative judgment mechanism that integrates multi-source data. Second, the accuracy of fault area location is insufficient. Traditional methods often rely on manual investigation or simple "user-line" correspondence for rough location, without systematic tracing based on the distribution network topology, which can easily lead to the expansion or omission of fault areas. Third, the data dimension for fault type diagnosis is limited, often relying solely on electrical data without integrating key information such as signal action sequences and environmental data, resulting in low accuracy in fault type identification. Fourth, there is a lack of proactive risk warning mechanisms. Existing technologies are mostly "passive diagnosis after power outages," failing to assess grid operation risks in advance based on data such as equipment load, aging, and historical faults, thus failing to achieve "prevention before the event." Fifth, historical fault data utilization is low, and handling strategies rely heavily on the experience of maintenance personnel, resulting in high subjectivity and low efficiency. The most significant technical problem to be solved is that existing solutions cannot achieve accurate identification, efficient location, accurate diagnosis, and proactive early warning of power outage events through multi-source data fusion, topology correlation analysis, and reuse of historical patterns.

[0003] To address the shortcomings of existing technologies, there is an urgent need for an intelligent power outage event diagnosis solution based on big data analytics. This solution could improve the accuracy, efficiency, and proactivity of power outage diagnosis by integrating multi-source data, performing topological correlation analysis, matching spatiotemporal features, and reusing historical patterns, thereby reducing losses caused by power outages. Summary of the Invention

[0004] The purpose of this application is to provide an intelligent diagnosis method and system for power outage events based on big data analysis, which aims to solve the problems of inaccurate power outage judgment, coarse fault area location, poor fault type identification, lack of risk warning and low utilization of historical data in the existing technology. It achieves intelligent diagnosis through a five-step closed loop of "power outage judgment - fault area location - core area verification - fault type diagnosis - strategy generation", and can also realize proactive early warning of risks. This invention achieves intelligent and accurate diagnosis of power outage events by real-time monitoring of switch tripping signals, user-side power loss data, and customer repair work order data. If a power outage event is determined, the distribution network topology is traced based on the user-side power loss data or customer repair work order data to determine the initial fault area. If a power outage event is determined and a switch tripping signal exists, candidate fault areas are determined based on the switch tripping signal and a preset distribution network topology. The initial fault area and candidate fault areas are verified to determine the core fault area. Within the core fault area, spatiotemporal correlation diagnosis is performed based on multi-source time-series data to output the fault type. Historical fault mode matching is performed based on the core fault area and fault type to obtain handling strategy data and generate a structured diagnostic report. At the same time, proactive risk warning is provided.

[0005] The present invention adopts the following technical solution.

[0006] This invention proposes an intelligent diagnostic method for power outage events based on big data analysis, comprising: Real-time monitoring of switch tripping signals, user-side power loss data, and customer repair work order data to determine power outage events; If a power outage event is determined, the initial fault area is determined based on the user-side power loss data and the customer repair work order data, and based on the preset power grid topology. If the switch tripping signal exists when a power outage occurs, then candidate fault areas are determined based on the switch tripping signal and the distribution network topology. Based on the initial fault region and the candidate fault regions, the core fault region is determined; Within the core fault area, the fault type is determined based on multi-source timing data; wherein, the multi-source timing data includes electrical timing data, signal action sequence data, and environmental timing data; Historical fault patterns are matched based on the core fault area and the fault type to obtain handling strategy data, and a structured diagnostic report is generated in combination with the fault type.

[0007] More preferably, the real-time monitoring of switch tripping signals, user-side power loss data, and customer repair work order data is used to determine power outage events, and the determination criteria include: Based on a preset text classification algorithm, the customer repair order data within a preset time window in the same power supply area is processed to obtain the number of repair orders for power outage faults, and to determine whether the number of repair orders for power outage faults exceeds a preset first threshold. Based on the user-side power outage data, extract the number of users with power outages in the same power supply area, and determine whether the number of users with power outages exceeds the preset second threshold and whether the proportion of users with power outages exceeds the preset third threshold. Determine if the power distribution switch has tripped; If at least one of the above three conditions is true, a power outage event is generated.

[0008] More preferably, if a power outage event is determined, the specific method for determining the initial fault area based on the user-side power loss data and the customer repair work order data, and based on a preset power grid topology, includes: Based on user-side power outage data and customer repair order data within a preset time window, a set of users experiencing power outages is obtained. Based on the power distribution network topology, each user in the set of power-out users is traced upstream to the power supply node directly or indirectly connected to it. Users who share a common upstream power supply node are grouped into the same user group; The downstream power supply area of ​​the common upstream power supply node corresponding to each user group is identified as a pending fault area; Calculate the ratio of the number of power-out users in each pending fault area to the total number of users in that pending fault area, and use this ratio as the power-out user proportion. The undetermined fault area is determined as the initial fault area if the proportion of power outage users is greater than a preset proportion threshold and the number of power outage users is greater than a preset number threshold.

[0009] More preferably, if the switch tripping signal exists when a power outage event occurs, then candidate fault areas are determined based on the switch tripping signal and the distribution network topology. The specific method includes: If a switch trip signal exists, starting from the switch that tripped, all downstream nodes electrically connected to the starting point are traversed based on the distribution network topology. During the traversal, when a node corresponding to a sectionalizing switch that is in the open state is encountered, the traversal operation of the current path is terminated. The actual power grid range corresponding to the connected subgraph formed by the topological nodes and connecting edges visited during the traversal process is determined as the candidate fault region.

[0010] More preferably, the method for determining the core fault region based on the initial fault region and the candidate fault regions includes: If the switch trip signal is not present, the initial fault area will be identified as the core fault area. If the circuit breaker trip signal exists, calculate the ratio of the intersection area of ​​the initial fault area and the candidate fault area to the union area, and determine whether it is greater than a preset first ratio threshold; calculate the ratio of the number of power-outage users jointly covered by the initial fault area and the candidate fault area to the total number of power-outage users, and determine whether it is greater than a preset second ratio threshold; if both of the above determinations are yes, then select the candidate fault area as the core fault area; otherwise, select the area with a high ratio of power-outage users to the total number of power-outage users as the core fault area.

[0011] More preferably, the method for determining the fault type based on multi-source time-series data within the core fault area includes: Feature extraction is performed on the multi-source time-series data to obtain fault feature pattern data, which includes electrical feature data, action logic feature data and environmental time-series feature data. The fault feature pattern data is input into a preset fault feature pattern library for matching to obtain fault type data.

[0012] More preferably, the specific method for inputting the fault feature pattern data into a preset fault feature pattern library for matching to obtain fault type data includes: Extracting topology data of core fault areas from the distribution network topology; The core fault area topology data, the total number of power outage users, fault type data, and fault characteristic pattern data are input into a preset historical fault database for matching to obtain handling strategy data. Generate a structured diagnostic report that includes the scope of the confirmed fault, the fault type, and the handling strategy.

[0013] More preferably, the specific method for inputting the core fault area topology data, the total number of power outage users, fault type data, and fault characteristic pattern data into a preset historical fault database for matching to obtain handling strategy data includes: The core fault area topology data and the fault type data are used as key core dimension data, and weights are assigned to them; the number of power-out users and the fault characteristic pattern data are used as auxiliary adaptation dimension data, and weights are assigned to them; wherein the weight of the key core dimension data is greater than the weight of the auxiliary adaptation dimension data. Calculate the matching degree between the core fault area topology data, the total number of power outage users, the fault type data, and the fault characteristic pattern data and the corresponding data in the preset historical fault database; wherein, the matching degree of the key core dimension data must meet the set threshold condition. If this condition is met, proceed to the following steps. If the matching degree of the key core dimension data meets the conditions, and the overall comprehensive matching degree calculated based on the set weight exceeds the preset total threshold, the handling strategy of the corresponding historical case will be used as the final handling strategy data. If the matching degree of the key core dimension data meets the condition, but the overall comprehensive matching degree does not meet the condition: The comprehensive similarity of the auxiliary adaptation dimension data is calculated based on the weights, and the handling strategies of the top N historical cases with the highest comprehensive similarity are used as the benchmark strategies. The baseline strategy is optimized to obtain a fine-tuned strategy as the disposal strategy data.

[0014] This invention also proposes an intelligent diagnostic system for power outage events based on big data analysis, including a data acquisition module, an initial fault area determination module, a candidate fault area determination module, a core fault area determination module, a fault type determination module, and a handling strategy generation module. The data acquisition module monitors switch tripping signals, user-side power loss data, and customer repair work order data in real time to determine power outage events. The initial fault area determination module determines the initial fault area based on the user-side power loss data and the customer repair work order data, and based on the preset power grid topology, if the event is determined to be a power outage. The candidate fault area determination module determines the candidate fault area based on the switch trip signal and the distribution network topology if the switch trip signal exists when a power outage event occurs. The core fault region determination module determines the core fault region based on the initial fault region and the candidate fault regions; The fault type determination module determines the fault type within the core fault area based on multi-source timing data; wherein, the multi-source timing data includes electrical timing data, signal action sequence data, and environmental timing data; The handling strategy generation module performs historical fault pattern matching based on the core fault area and the fault type to obtain handling strategy data, and generates a structured diagnostic report in combination with the fault type.

[0015] The present invention also proposes a terminal, including a processor and a storage medium: The storage medium is used to store instructions; The processor is used to perform the steps of the above method according to the instructions.

[0016] The present invention also proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention solves the problem of inaccurate power outage judgment: multi-source data fusion + multi-dimensional condition judgment avoids misjudgment and missed judgment caused by single data, and improves the accuracy of power outage event identification; 2. This invention solves the problem of coarse fault area location: by combining topology tracing and dual-area verification, the core fault area is accurately located, the investigation scope is narrowed, and the operation and maintenance costs are reduced; 3. This invention solves the problem of one-sided fault type diagnosis: multi-source time series data spatiotemporal correlation diagnosis covers multiple influencing factors of equipment, signals, and environment, improving the accuracy of fault type identification; 4. This invention enables proactive risk warning: it assesses the operational risks of the power supply area in advance, transforming "passive diagnosis" into "proactive prevention" and reducing losses from large-scale power outages; 5. This invention can improve operation and maintenance efficiency: historical fault modes are reused to generate handling strategies, reducing reliance on manual experience, shortening the fault handling cycle, and structured reports facilitate rapid implementation. Attached Figure Description

[0018] Figure 1 This is a flowchart of the intelligent diagnosis method for power outage events based on big data analysis according to the present invention. Figure 2 This is a flowchart of the intelligent diagnosis method for power outage events based on big data analysis according to Embodiment 1 of the present invention; Figure 3 This is a flowchart illustrating the power outage event determination process of the intelligent power outage event diagnosis method based on big data analysis provided in this embodiment of the invention. Figure 4 This is a flowchart illustrating the determination of the initial fault area in the intelligent diagnosis method for power outage events based on big data analysis provided in this embodiment of the invention. Figure 5 This is a flowchart illustrating the process of determining candidate fault regions in the intelligent diagnosis method for power outage events based on big data analysis provided in this embodiment of the invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0020] The present invention proposes the following technical solution: like Figure 1As shown, this invention proposes an intelligent diagnostic method for power outage events based on big data analysis, comprising: Real-time monitoring of switch tripping signals, user-side power loss data, and customer repair work order data to determine power outage events; The real-time monitoring of switch tripping signals, user-side power loss data, and customer repair work order data is used to determine power outage events. The determination criteria include: Based on a preset text classification algorithm, the customer repair order data within a preset time window in the same power supply area is processed to obtain the number of repair orders for power outage faults, and to determine whether the number of repair orders for power outage faults exceeds a preset first threshold. Based on the user-side power outage data, extract the number of users with power outages in the same power supply area, and determine whether the number of users with power outages exceeds the preset second threshold and whether the proportion of users with power outages exceeds the preset third threshold. Determine if the power distribution switch has tripped; If at least one of the above three conditions is true, a power outage event is generated.

[0021] If a power outage event is determined, the initial fault area is determined based on the user-side power loss data and the customer repair work order data, and based on the preset power grid topology. If a power outage event is determined, the specific method for determining the initial fault area based on the user-side power loss data and the customer repair work order data, and based on a preset power grid topology, includes: Based on user-side power outage data and customer repair order data within a preset time window, a set of users experiencing power outages is obtained. Based on the power distribution network topology, each user in the set of power-out users is traced upstream to the power supply node directly or indirectly connected to it. Users who share a common upstream power supply node are grouped into the same user group; The downstream power supply area of ​​the common upstream power supply node corresponding to each user group is identified as a pending fault area; Calculate the ratio of the number of power-out users in each pending fault area to the total number of users in that pending fault area, and use this ratio as the power-out user proportion. The undetermined fault area is determined as the initial fault area if the proportion of power outage users is greater than a preset proportion threshold and the number of power outage users is greater than a preset number threshold.

[0022] If the switch tripping signal exists when a power outage occurs, then candidate fault areas are determined based on the switch tripping signal and the distribution network topology. If a power outage event occurs and the switch tripping signal is present, then candidate fault areas are determined based on the switch tripping signal and the distribution network topology. Specific methods include: If a switch trip signal exists, starting from the switch that tripped, all downstream nodes electrically connected to the starting point are traversed based on the distribution network topology. During the traversal, when a node corresponding to a sectionalizing switch that is in the open state is encountered, the traversal operation of the current path is terminated. The actual power grid range corresponding to the connected subgraph formed by the topological nodes and connecting edges visited during the traversal process is determined as the candidate fault region.

[0023] Based on the initial fault region and the candidate fault regions, the core fault region is determined; The method for determining the core fault region based on the initial fault region and the candidate fault regions includes: If the switch trip signal is not present, the initial fault area will be identified as the core fault area. If the circuit breaker trip signal exists, calculate the ratio of the intersection area of ​​the initial fault area and the candidate fault area to the union area, and determine whether it is greater than a preset first ratio threshold; calculate the ratio of the number of power-outage users jointly covered by the initial fault area and the candidate fault area to the total number of power-outage users, and determine whether it is greater than a preset second ratio threshold; if both of the above determinations are yes, then select the candidate fault area as the core fault area; otherwise, select the area with a high ratio of power-outage users to the total number of power-outage users as the core fault area.

[0024] Within the core fault area, the fault type is determined based on multi-source timing data; wherein, the multi-source timing data includes electrical timing data, signal action sequence data, and environmental timing data; Within the core fault region, the fault type is determined based on multi-source time-series data. Specific methods include: Feature extraction is performed on the multi-source time-series data to obtain fault feature pattern data, which includes electrical feature data, action logic feature data and environmental time-series feature data. The fault feature pattern data is input into a preset fault feature pattern library for matching to obtain fault type data.

[0025] The specific method for inputting the fault feature pattern data into a preset fault feature pattern library for matching to obtain fault type data includes: Extracting topology data of core fault areas from the distribution network topology; The core fault area topology data, the total number of power outage users, fault type data, and fault characteristic pattern data are input into a preset historical fault database for matching to obtain handling strategy data. Generate a structured diagnostic report that includes the scope of the confirmed fault, the fault type, and the handling strategy.

[0026] The specific method for inputting core fault area topology data, total number of power outage users, fault type data, and fault characteristic pattern data into a preset historical fault database for matching to obtain handling strategy data includes: The core fault area topology data and the fault type data are used as key core dimension data, and weights are assigned to them; the number of power-out users and the fault characteristic pattern data are used as auxiliary adaptation dimension data, and weights are assigned to them; wherein the weight of the key core dimension data is greater than the weight of the auxiliary adaptation dimension data. Calculate the matching degree between the core fault area topology data, the total number of power outage users, the fault type data, and the fault characteristic pattern data and the corresponding data in the preset historical fault database; wherein, the matching degree of the key core dimension data must meet the set threshold condition. If this condition is met, proceed to the following steps. If the matching degree of the key core dimension data meets the conditions, and the overall comprehensive matching degree calculated based on the set weight exceeds the preset total threshold, the handling strategy of the corresponding historical case will be used as the final handling strategy data. If the matching degree of the key core dimension data meets the condition, but the overall comprehensive matching degree does not meet the condition: The comprehensive similarity of the auxiliary adaptation dimension data is calculated based on the weights, and the handling strategies of the top N historical cases with the highest comprehensive similarity are used as the benchmark strategies. The baseline strategy is optimized to obtain a fine-tuned strategy as the disposal strategy data.

[0027] Historical fault patterns are matched based on the core fault area and the fault type to obtain handling strategy data, and a structured diagnostic report is generated in combination with the fault type.

[0028] This invention also proposes an intelligent diagnostic system for power outage events based on big data analysis, including a data acquisition module, an initial fault area determination module, a candidate fault area determination module, a core fault area determination module, a fault type determination module, and a handling strategy generation module. The data acquisition module monitors switch tripping signals, user-side power loss data, and customer repair work order data in real time to determine power outage events. The initial fault area determination module determines the initial fault area based on the user-side power loss data and the customer repair work order data, and based on the preset power grid topology, if the event is determined to be a power outage. The candidate fault area determination module determines the candidate fault area based on the switch trip signal and the distribution network topology if the switch trip signal exists when a power outage event occurs. The core fault region determination module determines the core fault region based on the initial fault region and the candidate fault regions; The fault type determination module determines the fault type within the core fault area based on multi-source timing data; wherein, the multi-source timing data includes electrical timing data, signal action sequence data, and environmental timing data; The handling strategy generation module performs historical fault pattern matching based on the core fault area and the fault type to obtain handling strategy data, and generates a structured diagnostic report in combination with the fault type.

[0029] The present invention also proposes a terminal, including a processor and a storage medium: The storage medium is used to store instructions; The processor is used to perform the steps of the above method according to the instructions.

[0030] The present invention also proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0031] Example 1 Please refer to Figure 2 , Figure 2 This is a flowchart of a power outage event intelligent diagnosis method based on big data analysis according to some embodiments of this application. This power outage event intelligent diagnosis method based on big data analysis is used in terminal devices, such as computers and mobile terminals. The power outage event intelligent diagnosis method based on big data analysis includes the following steps: S11. Real-time monitoring of switch tripping signals, user-side power loss data, and customer repair work order data, and determination of power outage events; S12. If it is determined to be a power outage event, based on the power loss data on the user side and the customer's repair work order data, trace upstream to the preset distribution network topology to determine the initial fault area. S13. If it is determined to be a power outage event and there is a switch trip signal, then the candidate fault area is determined based on the switch trip signal and the preset distribution network topology. S14. Verify the initial fault area and candidate fault areas to determine the core fault area; S15. Within the core fault area, determine the fault type based on multi-source time-series data; S16. Based on the core fault area and fault type, perform historical fault mode matching to obtain handling strategy data, and generate a structured diagnostic report in combination with the fault type.

[0032] It should be noted that this application achieves intelligent diagnosis through a five-step closed loop of "power outage determination - fault area location - core area verification - fault type diagnosis - strategy generation". It can also achieve proactive early warning of risks, aiming to solve the problems of inaccurate power outage determination, coarse fault area location, poor fault type identification, lack of risk warning and low utilization of historical data in the existing technology.

[0033] Please refer to Figure 3 , Figure 3 This is a flowchart illustrating the power outage event determination process of a power outage event intelligent diagnosis method based on big data analysis in some embodiments of this application. According to embodiments of the present invention, the real-time monitoring of switch tripping signals, user-side power loss data, and customer repair work order data, and the determination of power outage events, includes: S21. Based on a preset text classification algorithm, process the customer repair work order data within a preset time window in the same power supply area to obtain the number of repair work orders for power outage faults, and determine whether the number of repair work orders for power outage faults exceeds a preset first threshold. S22. Extract the number of users with power outages in the same power supply area based on the user-side power outage data, and determine whether the number of users with power outages exceeds the preset second threshold and whether the proportion of users with power outages exceeds the preset third threshold. S23. Determine if the power distribution switch has tripped; S24. If at least one of the above judgment conditions is true, then a power outage event is generated.

[0034] It should be noted that the multi-dimensional judgment logic, which involves "processing repair work orders using text classification algorithms (quantifying work order quantity thresholds), extracting the number and proportion of power outage users (dual threshold judgment), and detecting switch tripping status," overcomes the limitations of existing technologies that rely on single data. This avoids misjudgments based solely on repair work orders (susceptible to interference from non-power outage repair requests) and also avoids the shortcomings of relying solely on switch signals (prone to missing outages without tripping), thus improving the accuracy and timeliness of power outage event judgment. Specifically, natural language processing methods, such as a BERT-based text classification algorithm, are used to extract keywords from repair work orders. Work orders containing preset keywords are identified as power outage-related work orders. Spatial clustering is then performed on these power outage-related work orders, grouping work orders within the same power supply area and with similar timing into one category to obtain the number of power outage fault repair work orders.

[0035] As a preferred embodiment of the present invention, work orders within the same power supply area and within 15-60 minutes are grouped into one category (this can be flexibly set according to actual scenarios such as user density and operation and maintenance response standards in the power supply area). The division standard for the same power supply area is clearly defined as the power distribution network power supply zone (rather than an administrative region). The division is based on the hierarchical relationship of "power supply node-line-user" in the power distribution network topology. That is, a power supply area is constituted by a cluster of users covered by the same upstream power supply node and branch lines. This can be directly determined by those skilled in the art based on existing power distribution network planning and design documents, which is a conventional division method in the industry.

[0036] It should be noted that all preset values ​​in this invention are determined through practical applications and can be configured specifically according to the power supply area characteristics, equipment parameters, operation and maintenance standards of the actual project.

[0037] Please refer to Figure 4 , Figure 4 This is a flowchart illustrating the determination of the initial fault region in a power outage event intelligent diagnosis method based on big data analysis, as described in some embodiments of this application. According to embodiments of the present invention, if a power outage event is determined, the initial fault region is determined by tracing the distribution network topology upstream based on user-side power loss data and customer repair work order data, including: S31. Based on the user-side power outage data and customer repair work order data within a preset time window, obtain the set of power outage users; S32. According to the preset power distribution network topology, trace each user in the power outage user set upstream to the power supply node directly or indirectly connected to it. S33. Users with a common upstream power supply node are grouped into the same user group; S34. Determine the downstream power supply area of ​​the common upstream power supply node corresponding to each user group as a pending fault area; S35. Calculate the ratio of the number of power-out users in each pending fault area to the total number of users in that pending fault area, and use this as the power-out user ratio. S36. Determine the pending fault area where the proportion of power-out users is greater than a preset proportion threshold and the number of power-out users is greater than a preset number threshold as the initial fault area.

[0038] It should be noted that this application, based on the correlation of the distribution network topology, transforms scattered user-side power outage data and repair reports into a systematic basis for regional location. Through the steps of "tracing upstream power supply nodes - dividing into the same user group - filtering potential areas based on dual thresholds of power outage proportion and quantity," it ensures a high degree of consistency between the initial fault area and the actual power outage range. Compared to traditional manual, rough location methods, this not only narrows the fault investigation scope but also avoids regional location errors caused by confusion in the "user-line" correspondence.

[0039] Please refer to Figure 5 , Figure 5 This is a flowchart illustrating the process of determining candidate fault regions in a power outage event intelligent diagnosis method based on big data analysis, as described in some embodiments of this application. According to an embodiment of the present invention, if a power outage event is determined and a switch tripping signal exists, determining the candidate fault region based on the switch tripping signal and a preset distribution network topology includes: S41. If a switch trip signal exists, starting from the switch that tripped, based on the preset distribution network topology, a preset breadth-first search algorithm is used to traverse all downstream nodes electrically connected to the starting point. S42. During the traversal, when a node corresponding to a sectionalizing switch that is in the open state is encountered, the traversal operation of the current path is terminated. S43. The actual power grid range corresponding to the connected subgraph formed by the topological nodes and connecting edges visited during the traversal process is determined as the candidate fault area.

[0040] It should be noted that this application, targeting scenarios where "a circuit breaker tripping signal exists," employs a topology analysis logic of "breadth-first search + traversal terminating at the tripping switch." Starting from the tripping switch, it can quickly traverse all electrically connected downstream nodes. Simultaneously, by using the rule of "terminating at the tripping switch," it accurately defines the scope of the tripping impact (avoiding the inclusion of non-faulty areas that cross the tripping switch). Compared to the traditional coarse location method of "the tripping switch directly corresponding to the line," this application can automatically generate candidate fault areas consistent with the actual electrical connections based on the topology structure, improving the efficiency and accuracy of area location, and eliminating the need for manual intervention, thus reducing operation and maintenance costs.

[0041] According to an embodiment of the present invention, the step of verifying the initial fault region and candidate fault regions to determine the core fault region includes: If there is no circuit breaker trip signal, the initial fault area is determined as the core fault area; If a switch trip signal exists, calculate the ratio of the intersection area of ​​the initial fault region and the candidate fault region to the union area, and determine whether it is greater than the preset first ratio threshold. Calculate the proportion of the number of power-out users jointly covered by the initial fault area and the candidate fault area to the total number of power-out users, and determine whether it is greater than the preset second proportion threshold. If all of the above judgments are true, then the candidate fault region is selected as the core fault region. Otherwise, select the area with a high proportion of power outage users to the total number of power outage users as the core fault area, and conduct manual investigation and prompts.

[0042] It should be noted that this application solves the core problem of "inconsistency between the initial area and the candidate area" through the logic of "scenario-based verification + dual-ratio indicators + manual prompts". When there is no tripping signal, the initial area is directly used (adapting to scenarios without hardware signals). When there is a tripping signal, the core area is verified by dual indicators: "intersection area ratio (regional spatial consistency) and the proportion of power-loss users under common coverage (user coverage consistency)". This ensures that the core area conforms to both the topological logic (candidate area) and the actual power loss situation (initial area). If the dual indicators are not met, the area with a high proportion of power-loss users is selected and manual investigation is prompted. This avoids the error of "only trusting topological signals and ignoring actual user data" and provides a manual assistance entry point for complex scenarios, further ensuring the reliability of the core area positioning.

[0043] The core purpose of manual investigation is not only to verify the correctness of the selection of "selecting areas with a high proportion of power outage users as core fault areas", but also to compensate for the deviation between topology signals and actual power outage data when the dual proportion thresholds (intersection area ratio and the proportion of power outage users jointly covered) of the initial fault area and candidate fault areas are not simultaneously met, so as to accurately locate the real core fault area (avoiding misjudgment of areas due to data anomalies or complex power grid topology); and the results of manual investigation will be fed back to the system in real time to correct the final determination of the core fault area, ensuring the accuracy of subsequent fault type diagnosis and handling strategy generation.

[0044] According to an embodiment of the present invention, determining the fault type based on multi-source time-series data within the core fault region includes: Acquire multi-source timing data within the fault time window, including electrical timing data, signal action sequence data, and environmental timing data; Feature extraction is performed on the multi-source time-series data to obtain fault feature pattern data, which includes electrical feature data, action logic feature data and environmental time-series feature data. The fault feature pattern data is input into a preset fault feature pattern library for matching to obtain fault type data.

[0045] It should be noted that this application integrates multi-source information including "electrical timing data, signal action sequence data, and environmental timing data" to extract multi-dimensional fault feature patterns (electrical features reflect equipment status, action logic features reflect switch linkage patterns, and environmental features reflect external influencing factors). These multi-dimensional features are then matched with a pre-defined fault feature pattern library, covering various scenarios such as "equipment failure (e.g., line short circuit), signal failure (e.g., switch malfunction), and environmentally caused failure (e.g., line grounding caused by heavy rain)," significantly improving the accuracy of fault type identification and avoiding misjudgments due to missing data dimensions (e.g., electrical data alone cannot distinguish between "line aging short circuit" and "heavy rain short circuit"). The fault feature pattern library described in this application is constructed by extracting multi-source timing fault feature data (including electrical features, signal action features, and environmental features) from a large number of historical power outage events, and then associating and matching these feature data with the actual fault types of each historical event, ultimately forming a standardized "fault type-feature pattern" correspondence database.

[0046] The electrical characteristic data includes abnormal voltage values ​​and durations, abnormal current values ​​and durations, active / reactive power abrupt changes, and insulation resistance values. The action logic characteristic data includes distribution switch trip / closing timestamps, distribution switch action trigger type data, and protection device action sequence data. The environmental time-series characteristic data includes meteorological data within the fault time window and geographical environmental data of the fault area. Methods for extracting electrical characteristic data include, but are not limited to, time-domain analysis, frequency-domain analysis, and wavelet transform. Methods for extracting action logic characteristic data include, but are not limited to, association rule analysis (Apriori algorithm) and timestamp interval analysis. Methods for extracting environmental time-series characteristic data include, but are not limited to, sliding window statistics and principal component analysis (PCA).

[0047] According to an embodiment of the present invention, the step of performing historical fault pattern matching based on core fault areas and fault types to obtain handling strategy data, and generating a structured diagnostic report in combination with fault types, includes: Extracting topology data of core fault areas from the distribution network topology; The core fault area topology data, the total number of power outage users, fault type data, and fault characteristic pattern data are input into a preset historical fault database for matching to obtain handling strategy data. Generate a structured diagnostic report that includes the scope of the confirmed fault, the fault type, and the handling strategy.

[0048] Specifically, the matching method is implemented based on a weight allocation logic of "prioritizing key core dimensions + supplementing with auxiliary adaptation dimensions". Core fault area topology data and fault type data are key core dimension data (with the highest weight), while the number of power-out users and fault characteristic pattern data are auxiliary adaptation dimension data (with decreasing weights). The importance of each dimension varies, and the weights are set according to the actual situation. The matching rule is a preset similarity threshold (determined based on statistical optimization of historical fault data). When the matching degree of the key core dimensions based on the set weights meets the standard and the overall comprehensive matching degree exceeds the preset threshold, the corresponding historical handling strategy is directly retrieved. If not all matching is possible or the comprehensive matching degree does not meet the standard, the handling strategy corresponding to the historical case with consistent core dimension matching and the highest similarity in auxiliary adaptation dimensions is selected first. Adaptive fine-tuning is then performed based on the specific auxiliary adaptation dimension data of the current fault (such as the scale of power-out users and differences in fault characteristics) to ensure the targeting and feasibility of the handling strategy.

[0049] As a preferred embodiment of the present invention [wty1], the weight of the key core dimension is set to the highest (for example, accounting for more than 60% in the comprehensive matching degree calculation); the number of power outage users and fault characteristic pattern data are defined as auxiliary adaptation dimensions, and their weights decrease sequentially according to the actual diagnostic needs.

[0050] The matching process includes the following steps: Similarity quantification calculation: For each dimension, a conventional similarity algorithm is used for quantitative comparison.

[0051] Topological data: Graph theory-based subgraph isomorphism / approximate matching algorithms (such as VF2 algorithm, graph edit distance calculation) or topological descriptor comparison (such as adjacency matrix, node degree distribution) are used to calculate topological similarity.

[0052] Fault type: Precise matching based on preset fault type codes or semantic similarity calculation based on fault tree / knowledge graph.

[0053] Auxiliary adaptation dimensions: The number of power outage users is compared using relative error or proportion; for fault characteristic time series data, dynamic time warping (DTW) or Pearson correlation coefficient can be used to calculate waveform similarity.

[0054] The quantitative standard for "matching consistency": The "matching consistency of key core dimensions" in this invention refers to the fact that the similarity of a single item reaches a preset strict threshold. For example, the fault type must match precisely (similarity = 1); the similarity of the topology of the core fault area must not be less than 0.9 (this threshold can be determined by statistical optimization based on historical data). Only when this condition is met can the subsequent strategy selection process proceed.

[0055] Strategy matching and selection rules: Ideally, when the core dimensions match and the overall comprehensive matching degree calculated by weighting exceeds the preset total threshold (e.g., 0.85), the system directly retrieves the handling strategy of the corresponding historical case.

[0056] Suboptimal matching and "adaptive fine-tuning": If the core dimensions match, but the overall comprehensive matching degree does not meet the standard, the system enters the strategy optimization process: From the historical database, prioritize the top N (e.g., Top-3) historical cases with consistent core dimension matching and the highest overall similarity in auxiliary adaptation dimensions, and use their handling strategies as the benchmark strategies.

[0057] Fine-tuning the baseline strategy involves rule-based adjustments or parameter optimization. Specific methods include, but are not limited to: Rule-based adjustments: Predefined adjustment rules are applied based on the specific differences between current and historical cases in the auxiliary adaptation dimension. For example, "If the current number of users experiencing power outages is 1.5 times that of historical cases, the 'number of customer service seats notified' parameter will be increased by 50% year-on-year"; "If the current fault characteristic waveform amplitude is 1.2 times that of historical cases, the 'estimated repair time' will be increased by a fixed margin."

[0058] Model-based parameter optimization: Key parameters in the handling strategy (such as power outage isolation range and protection setting adjustment amount) are used as optimization variables. The auxiliary adaptation dimension data of the current fault is used as input. A lightweight regression model or case-based reasoning (CBR) local adjustment mechanism is used to solve for the optimal parameters for the current situation.

[0059] It should be noted that this application establishes a multi-dimensional historical matching mechanism based on "core area + fault type + number of power-outage users + fault characteristics," fully utilizing the value of historical fault data. Compared to the traditional approach of "relying on the experience of maintenance personnel to formulate strategies," it can quickly match the most similar cases to the current scenario from the historical fault database, reducing the subjectivity and time cost of strategy formulation. Simultaneously, it generates structured diagnostic reports to improve overall handling efficiency. The core fault area topology data refers to a set of data extracted from the distribution network topology, containing the hierarchical relationships of "user-distribution transformer-branch line-main line" within the core fault area, unique identifiers of each line (e.g., 10kV main line - branch line #3), and connection logic of each node (e.g., distribution transformer, line switch). The historical fault database is a pre-built standardized reference database storing "core fault area topology data, fault type, fault characteristic pattern data, power-outage user-related data, and corresponding handling strategies" from historical power outage events. The confirmed fault range refers to the set of fault spatial range data that includes specific line identifiers (main / branch lines), corresponding distribution transformer nodes, distribution of power-outage user clusters, and the number of power-outage users in each cluster within the core fault area.

[0060] According to an embodiment of the present invention, it further includes: Obtain real-time current load rate, equipment aging index, and historical fault frequency of each power supply area in the power distribution network topology; The operational risk index of each power supply area is obtained by weighting calculation based on real-time current load rate, equipment aging index and failure history frequency, and the operational risk level of each power supply area is determined. Monitor the number of users experiencing power outages and the number of customer repair requests within each power supply area; If the number of power outage users in a high-risk power supply area exceeds a preset threshold for the number of power outage users and the number of customer repair work orders exceeds a preset threshold for the number of work orders, a risk warning signal will be generated.

[0061] It should be noted that this application transforms power outage fault diagnosis from "passive diagnosis" to "active early warning." By calculating a weighted operational risk index, high-risk areas in the power grid (such as areas with excessive load, aging equipment, and numerous historical faults) can be identified in advance. Furthermore, by monitoring the number of power outage users and repair work orders in high-risk areas, an early warning signal is generated when a threshold is exceeded. This allows for timely intervention before large-scale power outages occur (such as early maintenance of high-risk equipment), preventing the fault from escalating. The equipment aging index is quantified based on the ratio of the equipment's actual operating years to its rated operating years.

[0062] According to an embodiment of the present invention, it further includes: Obtain the historical fault repair time corresponding to the handling strategy from the historical fault database; Real-time monitoring of the number of users whose power is restored per unit time in the core fault area, and calculation of the estimated restoration time based on the total number of users without power; If the deviation between the expected recovery time and the historical fault repair time is greater than a preset deviation threshold, a resource reinforcement request will be generated.

[0063] It should be noted that the current number of users with power outages is calculated based on the total number of users with power outages and the number of users whose power has been restored. The ratio of the current number of users with power outages to the number of users whose power has been restored in the most recent unit of time is used as the estimated remaining restoration time. This ratio is then added to the restoration time already consumed to obtain the estimated restoration time. The estimated restoration time is updated in real time.

[0064] Example 2 This invention also discloses an intelligent diagnostic system for power outage events based on big data analysis, including a data acquisition module, an initial fault area determination module, a candidate fault area determination module, a core fault area determination module, a fault type determination module, and a handling strategy generation module. When executed, it performs the following steps: Real-time monitoring of switch tripping signals, user-side power loss data, and customer repair work order data; and determination of power outage events. If it is determined to be a power outage event, the distribution network topology is traced upstream based on the user-side power loss data and customer repair work order data to determine the initial fault area. If a power outage event is identified and a switch tripping signal is present, then candidate fault areas are determined based on the switch tripping signal and the preset distribution network topology. Verify the initial fault region and candidate fault regions to determine the core fault region; Within the core fault area, spatiotemporal correlation diagnosis is performed based on multi-source time-series data, and the fault type is output. Historical fault patterns are matched based on the core fault area and fault type to obtain handling strategy data, and a structured diagnostic report is generated by combining the fault type.

[0065] It should be noted that this application achieves intelligent diagnosis through a five-step closed loop of "power outage determination - fault area location - core area verification - fault type diagnosis - strategy generation". It can also achieve proactive early warning of risks, aiming to solve the problems of inaccurate power outage determination, coarse fault area location, poor fault type identification, lack of risk warning and low utilization of historical data in the existing technology.

[0066] According to an embodiment of the present invention, the real-time monitoring of switch tripping signals, user-side power loss data, and customer repair work order data, and the determination of power outage events, includes: Based on a preset text classification algorithm, the customer repair work order data within a preset time window in the same power supply area is processed to obtain the number of repair work orders for power outage faults and to determine whether it exceeds a preset first threshold. Based on the user-side power outage data, extract the number of users with power outages in the same power supply area, and determine whether the number of users with power outages exceeds the preset second threshold and whether the proportion of users with power outages exceeds the preset third threshold. Determine if the power distribution switch has tripped; If at least one of the above judgment conditions is true, a power outage event is generated.

[0067] It should be noted that the multi-dimensional judgment logic, which involves "processing repair work orders using text classification algorithms (quantifying work order quantity thresholds), extracting the number and proportion of power outage users (dual threshold judgment), and detecting switch tripping status," overcomes the limitations of existing technologies that rely on single data sources. This avoids misjudgments based solely on repair work orders (susceptible to interference from non-power outage repair requests) and also avoids the shortcomings of relying solely on switch signals (prone to missing outages without tripping), thus improving the accuracy and timeliness of power outage event judgment. Specifically, by performing natural language processing on repair work orders to extract keywords, work orders containing preset keywords are identified as power outage-related work orders. Spatial clustering of power outage-related work orders groups work orders within the same power supply area and with similar timing into one category, thus obtaining the number of power outage fault repair work orders.

[0068] According to an embodiment of the present invention, if a power outage event is determined, the process of tracing the distribution network topology upstream based on user-side power loss data and customer repair work order data to determine the initial fault area includes: Based on user-side power outage data and customer repair order data within a preset time window, a set of users experiencing power outages is obtained. According to the preset power distribution network topology, each user in the power outage user set is traced upstream to the power supply node directly or indirectly connected to it. Users who share a common upstream power supply node are grouped into the same user group; The downstream power supply area of ​​the common upstream power supply node corresponding to each user group is identified as a pending fault area; Calculate the ratio of the number of power-out users in each pending fault area to the total number of users in that pending fault area, and use this ratio as the power-out user proportion. The undetermined fault area is determined as the initial fault area if the proportion of power outage users is greater than a preset proportion threshold and the number of power outage users is greater than a preset number threshold.

[0069] It should be noted that this application, based on the correlation of the distribution network topology, transforms scattered user-side power outage data and repair reports into a systematic basis for regional location. Through the steps of "tracing upstream power supply nodes - dividing into the same user group - filtering potential areas based on dual thresholds of power outage proportion and quantity," it ensures a high degree of consistency between the initial fault area and the actual power outage range. Compared to traditional manual, rough location methods, this not only narrows the fault investigation scope but also avoids regional location errors caused by confusion in the "user-line" correspondence.

[0070] According to an embodiment of the present invention, if a power outage event is determined and a switch tripping signal exists, determining the candidate fault region based on the switch tripping signal and a preset distribution network topology includes: If a switch trip signal exists, starting from the switch that tripped, based on the preset distribution network topology, a preset breadth-first search algorithm is used to traverse all downstream nodes electrically connected to the starting point. During the traversal, when a node corresponding to a sectionalizing switch that is in the open state is encountered, the traversal operation of the current path is terminated. The actual power grid range corresponding to the connected subgraph formed by the topological nodes and connecting edges visited during the traversal process is determined as the candidate fault region.

[0071] It should be noted that this application, targeting scenarios where "a circuit breaker tripping signal exists," employs a topology analysis logic of "breadth-first search + traversal terminating at the tripping switch." Starting from the tripping switch, it can quickly traverse all electrically connected downstream nodes. Simultaneously, by using the rule of "terminating at the tripping switch," it accurately defines the scope of the tripping impact (avoiding the inclusion of non-faulty areas that cross the tripping switch). Compared to the traditional coarse location method of "the tripping switch directly corresponding to the line," this application can automatically generate candidate fault areas consistent with the actual electrical connections based on the topology structure, improving the efficiency and accuracy of area location, and eliminating the need for manual intervention, thus reducing operation and maintenance costs.

[0072] According to an embodiment of the present invention, the step of verifying the initial fault region and candidate fault regions to determine the core fault region includes: If there is no circuit breaker trip signal, the initial fault area is determined as the core fault area; If a switch trip signal exists, calculate the ratio of the intersection area of ​​the initial fault region and the candidate fault region to the union area, and determine whether it is greater than the preset first ratio threshold. Calculate the proportion of the number of power-out users jointly covered by the initial fault area and the candidate fault area to the total number of power-out users, and determine whether it is greater than the preset second proportion threshold. If all of the above judgments are true, then the candidate fault region is selected as the core fault region. Otherwise, select the area with a high proportion of power outage users to the total number of power outage users as the core fault area, and conduct manual investigation and prompts.

[0073] It should be noted that this application solves the core problem of "inconsistency between the initial area and the candidate area" through the logic of "scenario-based verification + dual-ratio indicators + manual prompts". When there is no tripping signal, the initial area is directly used (adapting to scenarios without hardware signals). When there is a tripping signal, the core area is verified by dual indicators: "intersection area ratio (regional spatial consistency) and the proportion of power-loss users under common coverage (user coverage consistency)". This ensures that the core area conforms to both the topological logic (candidate area) and the actual power loss situation (initial area). If the dual indicators are not met, the area with a high proportion of power-loss users is selected and manual investigation is prompted. This avoids the error of "only trusting topological signals and ignoring actual user data" and provides a manual assistance entry point for complex scenarios, further ensuring the reliability of the core area positioning.

[0074] According to an embodiment of the present invention, the step of performing spatiotemporal correlation diagnosis based on multi-source time-series data within the core fault region and outputting the fault type includes: Acquire multi-source timing data within the fault time window, including electrical timing data, signal action sequence data, and environmental timing data; Feature extraction is performed on the multi-source time-series data to obtain fault feature pattern data, including electrical feature data, action logic feature data and environmental time-series feature data; The fault feature pattern data is input into a preset fault feature pattern library for matching to obtain fault type data.

[0075] It should be noted that this application integrates multi-source information including "electrical timing data, signal action sequence data, and environmental timing data" to extract multi-dimensional fault feature patterns (electrical features reflect equipment status, action logic features reflect switch linkage patterns, and environmental features reflect external influencing factors). These multi-dimensional features are then matched with a pre-defined fault feature pattern library, covering various scenarios such as "equipment failure (e.g., line short circuit), signal failure (e.g., switch malfunction), and environmentally caused failure (e.g., line grounding caused by heavy rain)," significantly improving the accuracy of fault type identification and avoiding misjudgments due to missing data dimensions (e.g., electrical data alone cannot distinguish between "line aging short circuit" and "heavy rain short circuit"). The fault feature pattern library described in this application is constructed by extracting multi-source timing fault feature data (including electrical features, signal action features, and environmental features) from a large number of historical power outage events, and then associating and matching these feature data with the actual fault types of each historical event, ultimately forming a standardized "fault type-feature pattern" correspondence database.

[0076] According to an embodiment of the present invention, the step of performing historical fault pattern matching based on core fault areas and fault types to obtain handling strategy data, and generating a structured diagnostic report in combination with fault types, includes: Extracting topology data of core fault areas from the distribution network topology; The core fault area topology data, the total number of power outage users, fault type data, and fault characteristic pattern data are input into a preset historical fault database for matching to obtain handling strategy data. Generate a structured diagnostic report that includes the scope of the confirmed fault, the fault type, and the handling strategy.

[0077] It should be noted that this application establishes a multi-dimensional historical matching mechanism based on "core area + fault type + fault time + number of power-outage users + fault characteristics," fully utilizing the value of historical fault data. Compared to the traditional approach of "relying on the experience of maintenance personnel to formulate strategies," it can quickly match the most similar cases to the current scenario from the historical fault database, reducing the subjectivity and time cost of strategy formulation. Simultaneously, it generates structured diagnostic reports to improve overall handling efficiency. The core fault area topology data refers to a set of data extracted from the distribution network topology, containing the hierarchical relationships within the core fault area ("user-distribution transformer-branch line-main line"), unique identifiers for each line (e.g., 10kV main line - branch line #3), and connection logic of each node (e.g., distribution transformer, line switch). The historical fault database is a pre-built standardized reference database storing "core fault area topology data, fault type, fault characteristic pattern data, power-outage user-related data, and corresponding handling strategies" from historical power outage events. The confirmed fault range refers to the set of fault spatial range data that includes specific line identifiers (main / branch lines), corresponding distribution transformer nodes, distribution of power-outage user clusters, and the number of power-outage users in each cluster within the core fault area.

[0078] According to an embodiment of the present invention, it further includes: Obtain real-time current load rate, equipment aging index, and historical fault frequency of each power supply area in the power distribution network topology; The operational risk index of each power supply area is obtained by weighting calculation based on real-time current load rate, equipment aging index and failure history frequency, and the operational risk level of each power supply area is determined. Monitor the number of users experiencing power outages and the number of customer repair requests within each power supply area; If the number of power outage users in a high-risk power supply area exceeds a preset threshold for the number of power outage users and the number of customer repair work orders exceeds a preset threshold for the number of work orders, a risk warning signal will be generated.

[0079] It should be noted that this application transforms power outage fault diagnosis from "passive diagnosis" to "active early warning." By calculating a weighted operational risk index, high-risk areas in the power grid (such as areas with excessive load, aging equipment, and numerous historical faults) can be identified in advance. Furthermore, by monitoring the number of power outage users and repair work orders in high-risk areas, an early warning signal is generated when a threshold is exceeded. This allows for timely intervention before large-scale power outages occur (such as early maintenance of high-risk equipment), preventing the fault from escalating. The equipment aging index is quantified based on the ratio of the equipment's actual operating years to its rated operating years.

[0080] According to an embodiment of the present invention, it further includes: Obtain the historical fault repair time corresponding to the handling strategy from the historical fault database; Real-time monitoring of the number of users whose power is restored per unit time in the core fault area, and calculation of the estimated restoration time based on the total number of users without power; If the deviation between the expected recovery time and the historical fault repair time is greater than a preset deviation threshold, a resource reinforcement request will be generated.

[0081] It should be noted that the current number of users with power outages is calculated based on the total number of users with power outages and the number of users whose power has been restored. The ratio of the current number of users with power outages to the number of users whose power has been restored in the most recent unit of time is used as the estimated remaining restoration time. This ratio is then added to the restoration time already consumed to obtain the estimated restoration time. The estimated restoration time is updated in real time.

[0082] This invention discloses an intelligent diagnosis method and system for power outage events based on big data analysis, which aims to solve the problems of inaccurate power outage judgment, coarse fault area location, poor fault type identification, lack of risk warning and low utilization of historical data in the existing technology. It achieves intelligent diagnosis through a five-step closed loop of "power outage judgment - fault area location - core area verification - fault type diagnosis - strategy generation", and can also realize proactive early warning of risks.

[0083] Example 3 The present invention also proposes a terminal, including a processor and a storage medium: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to Embodiment 1.

[0084] Example 4 The present invention also proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in Embodiment 1.

[0085] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0086] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0087] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for intelligent diagnosis of power outage events based on big data analysis, characterized in that, include: Real-time monitoring of switch tripping signals, user-side power loss data, and customer repair work order data to determine power outage events; If a power outage event is determined, the initial fault area is determined based on the user-side power loss data and the customer repair work order data, and based on the preset power grid topology. If the switch tripping signal exists when a power outage occurs, then candidate fault areas are determined based on the switch tripping signal and the distribution network topology. Based on the initial fault region and the candidate fault regions, the core fault region is determined; Within the core fault area, the fault type is determined based on multi-source timing data; wherein, the multi-source timing data includes electrical timing data, signal action sequence data, and environmental timing data; Historical fault patterns are matched based on the core fault area and the fault type to obtain handling strategy data, and a structured diagnostic report is generated in combination with the fault type.

2. The intelligent diagnosis method for power outage events based on big data analysis according to claim 1, characterized in that: The real-time monitoring of switch tripping signals, user-side power loss data, and customer repair work order data is used to determine power outage events. The determination criteria include: Based on a preset text classification algorithm, the customer repair order data within a preset time window in the same power supply area is processed to obtain the number of repair orders for power outage faults, and to determine whether the number of repair orders for power outage faults exceeds a preset first threshold. Based on the user-side power outage data, extract the number of users with power outages in the same power supply area, and determine whether the number of users with power outages exceeds the preset second threshold and whether the proportion of users with power outages exceeds the preset third threshold. Determine if the power distribution switch has tripped; If at least one of the above three conditions is true, a power outage event is generated.

3. The intelligent diagnosis method for power outage events based on big data analysis according to claim 1, characterized in that: If a power outage event is determined, the specific method for determining the initial fault area based on the user-side power loss data and the customer repair work order data, and based on a preset power grid topology, includes: Based on user-side power outage data and customer repair order data within a preset time window, a set of users experiencing power outages is obtained. Based on the power distribution network topology, each user in the set of power-out users is traced upstream to the power supply node directly or indirectly connected to it. Users who share a common upstream power supply node are grouped into the same user group; The downstream power supply area of ​​the common upstream power supply node corresponding to each user group is identified as a pending fault area; Calculate the ratio of the number of power-out users in each pending fault area to the total number of users in that pending fault area, and use this ratio as the power-out user proportion. The undetermined fault area is determined as the initial fault area if the proportion of power outage users is greater than a preset proportion threshold and the number of power outage users is greater than a preset number threshold.

4. The intelligent diagnosis method for power outage events based on big data analysis according to claim 1, characterized in that: If a power outage event occurs and the switch tripping signal is present, then candidate fault areas are determined based on the switch tripping signal and the distribution network topology. Specific methods include: If a switch trip signal exists, starting from the switch that tripped, all downstream nodes electrically connected to the starting point are traversed based on the distribution network topology. During the traversal, when a node corresponding to a sectionalizing switch that is in the open state is encountered, the traversal operation of the current path is terminated. The actual power grid range corresponding to the connected subgraph formed by the topological nodes and connecting edges visited during the traversal process is determined as the candidate fault region.

5. The intelligent diagnosis method for power outage events based on big data analysis according to claim 1, characterized in that: The method for determining the core fault region based on the initial fault region and the candidate fault regions includes: If the switch trip signal is not present, the initial fault area will be identified as the core fault area. If the circuit breaker trip signal exists, calculate the ratio of the intersection area of ​​the initial fault area and the candidate fault area to the union area, and determine whether it is greater than a preset first ratio threshold; calculate the ratio of the number of power-outage users jointly covered by the initial fault area and the candidate fault area to the total number of power-outage users, and determine whether it is greater than a preset second ratio threshold; if both of the above determinations are yes, then select the candidate fault area as the core fault area; otherwise, select the area with a high ratio of power-outage users to the total number of power-outage users as the core fault area.

6. The intelligent diagnosis method for power outage events based on big data analysis according to claim 1, characterized in that: Within the core fault region, the fault type is determined based on multi-source time-series data. Specific methods include: Feature extraction is performed on the multi-source time-series data to obtain fault feature pattern data, which includes electrical feature data, action logic feature data and environmental time-series feature data. The fault feature pattern data is input into a preset fault feature pattern library for matching to obtain fault type data.

7. The intelligent diagnosis method for power outage events based on big data analysis according to claim 6, characterized in that: The specific method for inputting the fault feature pattern data into a preset fault feature pattern library for matching to obtain fault type data includes: Extracting topology data of core fault areas from the distribution network topology; The core fault area topology data, the total number of power outage users, fault type data, and fault characteristic pattern data are input into a preset historical fault database for matching to obtain handling strategy data. Generate a structured diagnostic report that includes the scope of the confirmed fault, the fault type, and the handling strategy.

8. The intelligent diagnosis method for power outage events based on big data analysis according to claim 7, characterized in that: The specific method for inputting core fault area topology data, total number of power outage users, fault type data, and fault characteristic pattern data into a preset historical fault database for matching to obtain handling strategy data includes: The core fault area topology data and the fault type data are used as key core dimension data, and weights are assigned to them; the number of power-out users and the fault characteristic pattern data are used as auxiliary adaptation dimension data, and weights are assigned to them; wherein the weight of the key core dimension data is greater than the weight of the auxiliary adaptation dimension data. Calculate the matching degree between the core fault area topology data, the total number of power outage users, the fault type data, and the fault characteristic pattern data and the corresponding data in the preset historical fault database; wherein, the matching degree of the key core dimension data must meet the set threshold condition. If this condition is met, proceed to the following steps. If the matching degree of the key core dimension data meets the conditions, and the overall comprehensive matching degree calculated based on the set weight exceeds the preset total threshold, the handling strategy of the corresponding historical case will be used as the final handling strategy data. If the matching degree of the key core dimension data meets the condition, but the overall comprehensive matching degree does not meet the condition: The comprehensive similarity of the auxiliary adaptation dimension data is calculated based on the weights, and the handling strategies of the top N historical cases with the highest comprehensive similarity are used as the benchmark strategies. The baseline strategy is optimized to obtain a fine-tuned strategy as the disposal strategy data.

9. A power outage event intelligent diagnosis system based on big data analysis using the method of any one of claims 1-8, comprising a data acquisition module, an initial fault area determination module, a candidate fault area determination module, a core fault area determination module, a fault type determination module, and a handling strategy generation module, characterized in that: The data acquisition module monitors switch tripping signals, user-side power loss data, and customer repair work order data in real time to determine power outage events. The initial fault area determination module determines the initial fault area based on the user-side power loss data and the customer repair work order data, and based on the preset power grid topology, if the event is determined to be a power outage. The candidate fault area determination module determines the candidate fault area based on the switch trip signal and the distribution network topology if the switch trip signal exists when a power outage event occurs. The core fault region determination module determines the core fault region based on the initial fault region and the candidate fault regions; The fault type determination module determines the fault type within the core fault area based on multi-source timing data; wherein, the multi-source timing data includes electrical timing data, signal action sequence data, and environmental timing data; The handling strategy generation module performs historical fault pattern matching based on the core fault area and the fault type to obtain handling strategy data, and generates a structured diagnostic report in combination with the fault type.

10. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-8.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-8.

Images