A method, equipment, and medium for predicting the impact range of power outages and making emergency decisions.
By constructing a knowledge graph of power outage events and using vectorized embedding technology, the scope of power outage impact and emergency decisions are dynamically output, solving the problem of ineffective utilization of multi-source data in existing technologies. This enables rapid and accurate emergency response and decision-making, adapting to complex disturbance scenarios.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- STATE GRID SHANGHAI MUNICIPAL ELECTRIC POWER CO
- Filing Date
- 2026-01-14
- Publication Date
- 2026-06-02
AI Technical Summary
Existing technologies struggle to effectively integrate multi-source heterogeneous data, making it impossible to adaptively adjust the prediction of the impact range of power outage events and emergency decision-making, resulting in delayed assessments and inaccurate emergency responses.
A knowledge graph of power outage events is constructed with disturbance sources, influencing relationships, and affected objects as its core. Through vectorized embedding and similarity retrieval, the predicted range of power outage impact is dynamically output, and structured emergency decision-making instructions are generated by assessing confidence levels.
It enables rapid and accurate prediction of the impact range of power outages and emergency decision-making, improves the timeliness and reliability of emergency response, has lifelong learning capabilities, and can adapt to complex disturbance scenarios.
Smart Images
Figure CN122134146A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power system emergency command and intelligent decision-making technology, and in particular to a method, equipment and medium for predicting the impact range of power outages and making emergency decisions. Background Technology
[0002] With the continuous expansion of urban power grids and the frequent occurrence of extreme weather events, distribution networks face challenges such as the suddenness of power outages, the diversity of influencing factors, and the high requirements for timely emergency response. Traditional power outage impact analysis mainly relies on passive extrapolation based on power grid topology models and power outage signals, making it difficult to effectively integrate heterogeneous external information such as meteorological disturbances, population movement, user repair reports, and social media sentiment. This results in delayed assessment of the impact range, inaccurate identification of key users, and reliance on human experience for emergency decision-making, failing to meet the core requirements of modern emergency command for speed, accuracy, and intelligence.
[0003] Current mainstream emergency response systems mostly employ rule engines or statistical models for fault location and impact assessment. However, these methods have significant limitations: on the one hand, the rule base is static and cannot continuously evolve with new disturbance scenarios; on the other hand, multi-source data is not utilized in a structured manner, wasting a large amount of implicit causal relationships and scenario knowledge, leading to a significant decrease in the accuracy of judgment under complex coupled disturbances. Some methods have introduced deep learning models. For example, Chinese patent CN118504991A discloses a method, device, equipment, and medium for handling power outages in a power protection area. This method uses a high-precision time synchronization method for time stamping, collects historical operating data from each node of the power system, extracts features using a deep learning algorithm based on convolutional neural networks, establishes a power outage risk assessment model, and optimizes it using Bayesian networks or Markov decision processes to achieve precise monitoring and control of power outage risks. However, this method is a black-box process with poor interpretability.
[0004] In recent years, knowledge graph and artificial intelligence technologies have been initially applied in the power sector, with some studies attempting to build fault knowledge bases for auxiliary diagnosis. However, existing work mostly focuses on equipment-level fault reasoning, lacking systematic knowledge modeling for regional power outage events, and failing to establish a dynamic linkage mechanism between the knowledge base and real-time data. Especially in the impact range prediction stage, it still mainly relies on fixed radii or topological connectivity, unable to adaptively adjust the judgment logic according to the type of disturbance, making it difficult to support differentiated emergency responses.
[0005] Therefore, there is an urgent need for an intelligent judgment and decision-making method that can integrate multi-source heterogeneous data, automatically build and continuously evolve a knowledge system of power outage events, and achieve collaborative driving of data perception and knowledge reasoning, so as to improve the accuracy and timeliness of the power grid's emergency response in complex disturbance scenarios and provide intelligent support for typical businesses such as distribution network emergency repair command and power supply guarantee for important users. Summary of the Invention
[0006] The purpose of this invention is to overcome the defects of the prior art by providing a method, device and medium for predicting the impact range of power outages and making emergency decisions.
[0007] The objective of this invention can be achieved through the following technical solutions: According to a first aspect of the present invention, a method for predicting the impact range of power grid outages and making emergency decisions is provided, comprising the following steps: S1. Collect multi-source heterogeneous data related to power outage events, perform entity recognition and relationship extraction, and construct a power outage event knowledge graph with triplets consisting of disturbance sources, influencing relationships, and affected objects as the core. S2, the knowledge graph of the power outage event is vectorized and embedded to construct a semantic vector space of historical power outage events; when a new power outage event occurs, its structured features are extracted and a real-time feature vector is generated, similar historical events are retrieved in the semantic vector space, and the prediction results of the power outage impact range and the assessment confidence level are dynamically output. S3, when the assessment confidence level is greater than or equal to a preset threshold, the historical best matching scheme is used as a structured emergency decision instruction; when the assessment confidence level is less than the preset threshold, the preliminary response scheme and the historical best matching scheme are merged to generate a structured emergency decision instruction.
[0008] As a preferred technical solution, the disturbance sources include typhoons, lightning strikes, equipment aging, or external damage; the affected objects include important power users and densely populated residential areas; and the impact relationships include line tripping, equipment failure, voltage fluctuations, power outages of important loads, user-side repair requests, and social media public opinion feedback.
[0009] As a preferred technical solution, step S2 includes the following steps: The triples of each historical power outage event are aggregated into an event embedding vector to construct the semantic vector space of historical power outage events; Encode the features of the current power outage event into a real-time feature vector; The similarity between the current event and each historical event is calculated based on the real-time feature vector and the event embedding vector. Select the first preset number of historical events with the highest similarity; The predicted impact radius of the current power outage event is calculated based on the similarity of the selected historical events and their corresponding actual impact radii. The confidence level is calculated based on the similarity of the selected historical events.
[0010] As a preferred technical solution, the method for calculating the predicted impact radius of the current power outage event is as follows: , in, R pred To predict the radius of influence, R i For the first i The actual radius of influence of a historical event K To determine the number of historical events selected, sim i For the current power outage event and the first i The similarity between historical events.
[0011] As a preferred technical solution, the method for calculating the assessment confidence level is as follows: , Where C represents the confidence level, C∈[0,1] K To determine the number of historical events selected, sim i For the current power outage event and the first i The similarity between historical events.
[0012] As a preferred technical solution, the method for determining the structured emergency decision-making instructions in step S3 is as follows: , in, D For structured emergency decision-making instructions; D know The historical best matching scheme invoked by the knowledge-driven layer includes the historical best emergency repair sequence and power protection measures for matching disturbance types. D data The initial response plan output by the data-driven layer is dynamically generated based on the real-time distribution of power-outage users and the location of emergency repair resources. C To assess the confidence level, , where is the confidence threshold; , which is the fusion weighting coefficient.
[0013] As a preferred technical solution, the relation extraction adopts a joint model of named entity recognition and dependency parsing to automatically extract triples from unstructured text.
[0014] As a preferred technical solution, the method further includes: establishing a version management mechanism for the power outage event knowledge graph, supporting automatic updates after the injection of new events and retrospective verification of historical versions. The version management mechanism involves writing the manually reviewed new event triplet into the new version of the knowledge base after each emergency response loop is closed, and retaining the version number, timestamp, and change log.
[0015] According to a second aspect of the present invention, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the program to implement the method described thereon.
[0016] According to a third aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method described thereon.
[0017] Compared with the prior art, the present invention has the following beneficial effects: (1) Existing research is mostly limited to single equipment failures, while the power outage event knowledge graph constructed in this invention describes the regional event chain of <disturbance source-impact relationship-affected object>. This modeling method enables the system to understand and predict the scope and path of the fault. Through vectorized embedding and similarity retrieval, this invention can quickly match the most similar overall event scenario in history, rather than isolated equipment status, thereby outputting a more realistic prediction of the power outage impact range down to specific lines and users, providing key decision-making basis for emergency repair resource allocation and power supply protection for important users.
[0018] (2) This invention uses confidence level assessment to generate decisions. When the confidence level is high, it automatically recommends the best historical matching solution to achieve a fast and automated response and improve efficiency. When the confidence level is low (such as when encountering rare or novel faults), it integrates the preliminary response solution with historical experience to generate a more robust decision. This confidence-based hierarchical decision-making mechanism not only ensures the efficiency of handling routine events, but also provides a risk buffer for extremely complex scenarios, enhancing the practicality and reliability of the system.
[0019] (3) This invention continuously collects multi-source heterogeneous data and automatically constructs a knowledge graph. Each new power outage event and its handling experience are stored in the graph as new triplet knowledge, so that the system can automatically iterate and enrich its knowledge system as the power grid structure changes and new disturbance scenarios emerge. It has lifelong learning capabilities and completely breaks the bottleneck of knowledge updating in traditional systems. Attached Figure Description
[0020] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation
[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0022] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms “a,” “an,” “an,” “the,” and similar words used in this application do not indicate quantity limitation and may indicate singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units not listed, or may include other steps or units inherent to these processes, methods, products, or devices. The terms “connected,” “linked,” “coupled,” and similar words used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following objects are in an "or" relationship. The terms "first," "second," and "third" used in this application are merely to distinguish similar objects and do not represent a specific ordering of the objects.
[0023] Example 1 This embodiment provides a method for predicting the impact range of power outages and making emergency decisions, such as... Figure 1 As shown, it includes the following steps: S1 collects multi-source heterogeneous data related to power outage events from the power grid side, user side, government side, and social side, performs entity identification and relationship extraction, and constructs a power outage event knowledge graph with a triple consisting of disturbance source, influence relationship, and affected object as the core.
[0024] S11 collects heterogeneous data from multiple sources, including power grid operation alarms, user power outage signals, 95598 repair work orders, social media sentiment, weather warnings, and important user ledgers.
[0025] S12 employs a joint model of named entity recognition and dependency parsing to automatically extract triples from unstructured text and construct a knowledge graph of power outage events.
[0026] In this embodiment, the disturbance sources include typhoons, lightning strikes, equipment aging, or external damage; the affected objects include important power users such as hospitals, government agencies, communication base stations, and transportation hubs, as well as densely populated residential areas; the impact relationships include line tripping, equipment failure, power outage, voltage fluctuation, power transfer failure, protection action, grounding short circuit, communication interruption, and user power outage feedback, etc.
[0027] S13 establishes a version management mechanism for the power outage event knowledge graph, supporting automatic updates after new events are injected and historical version retrospective verification.
[0028] In this embodiment, the version management mechanism specifically involves writing the new event triplet, which has been manually reviewed, into the new version of the knowledge base after each emergency response loop is closed, and retaining the version number, timestamp, and change log.
[0029] S2, the knowledge graph of the power outage event is vectorized and embedded to construct a semantic vector space of historical power outage events; when a new power outage event occurs, its structured features are extracted and a real-time feature vector is generated, similar historical events are retrieved in the semantic vector space, and the prediction results of the power outage impact range and the assessment confidence level are dynamically output.
[0030] To support semantic computation, the knowledge graph is vectorized and embedded. Let the embedding dimension be . d The source entity of the disturbance e s Affected entities e o and the relationship of influence r Mapped to vectors respectively e s ∈ R d , e o ∈ R d and r ∈ R d By optimizing the triple reconstruction error, the embedding space achieves semantic consistency, and its scoring function is: , The function The smaller the value, the more reasonable the triple is in the semantic space, which is used to train the knowledge graph embedding model.
[0031] S21, aggregate the triples of each historical power outage event into an event embedding vector. Construct a semantic vector space for historical power outage events, where, i For indexing historical events, d For the embedded dimension.
[0032] S22, Encode the features of the current power outage event into a real-time feature vector. .
[0033] The mapping process is represented as: , in, x raw The original feature vector, For a learnable encoding function (such as a linear transform or a shallow neural network), the output dimension is... d Ensure embedding with historical events v i They exist in the same vector space.
[0034] S23, Calculate the cosine similarity between the current event and each historical event based on the real-time feature vector and event embedding vector: , in, For the first i Cosine similarity between a historical event and the current power outage.
[0035] S24, Select the top two with the highest similarity. K ( K We have (a positive integer) historical events used for inference of the scope of influence. To ensure the reasonableness of the weighted average, the similarity is normalized, and normalization weights are defined. : , Obviously there is ,and .
[0036] S25, Calculate the predicted impact radius of the current power outage event based on the similarity of the selected historical events and their corresponding actual impact radii: , in, R pred To predict the radius of influence, R i For the first i The actual radius of influence of a historical event K To determine the number of historical events selected, sim i For the current power outage event and the first i Cosine similarity of historical events.
[0037] S26, Calculate and determine the confidence level based on the similarity of the selected historical events: , Where C represents the confidence level, C∈[0,1] K To determine the number of historical events selected, sim i For the current power outage event and the first i Cosine similarity of historical events.
[0038] S3, when the assessment confidence level is greater than or equal to a preset threshold, the historical best matching scheme is used as a structured emergency decision instruction; when the assessment confidence level is less than the preset threshold, the preliminary response scheme and the historical best matching scheme are merged to generate a structured emergency decision instruction.
[0039] As a preferred technical solution, the method for determining the structured emergency decision-making instructions in step S3 is as follows: , in, D For structured emergency decision-making instructions; D know The historical best matching scheme invoked by the knowledge-driven layer includes the historical best emergency repair sequence and power protection measures for matching disturbance types. D data The initial response plan output by the data-driven layer is dynamically generated based on the real-time distribution of power-outage users and the location of emergency repair resources. C To assess the confidence level, , where is the confidence threshold; , which is the fusion weighting coefficient.
[0040] In this embodiment, the generated structured emergency decision-making instructions include repair priorities, emergency resource scheduling suggestions, and power supply protection measures for important users.
[0041] Example 2 The difference between this embodiment and Embodiment 1 is that the method for calculating the similarity between the current event and each historical event in step S23 is improved. The specific calculation method is as follows: First, calculate the cosine similarity between the current event and each historical event: , in, For the first i Cosine similarity between a historical event and the current power outage.
[0042] Subsequently, based on cosine similarity, considering spatiotemporal correlation, equipment matching, weather impact, load characteristics, power grid operating status, and event severity, an improved similarity is obtained: , in, For the first i Improvement similarity between historical events and current power outage events, For the corresponding weights, satisfying , For the first i The temporal correlation between a historical event and the current event. For the first iThe spatial correlation between historical events and current events For the first i Device type matching degree between historical events and current events For the first i The weather similarity between a historical event and the current event. For the first i The degree of matching between the load characteristics of historical events and current events. For the first i The degree of matching between the power grid operation status of historical events and current events. For the first i The similarity in severity between a historical event and a current event.
[0043] The calculation method for each parameter is explained in detail below: 1) Time correlation Time correlation To measure the temporal correlation between historical and current events, it is possible to integrate three dimensions: season, time period, and cycle type. 1. Seasonal Matching: Divide the year into four seasons: spring, summer, autumn, and winter. (This is related to historical events.) i 1 point if the season is the same as the current event season; otherwise, 0 points.
[0044] 2. Time Period Matching: Divide the day into peak, average, and off-peak electricity consumption periods, and match historical events... i 1 point is awarded if the event occurs during the same time period as the current event; otherwise, 0 points are awarded.
[0045] 3. Period type matching: Distinguish between weekdays, weekends, and holidays, and include historical events. i 1 point is awarded if the event type is the same as the current event type; otherwise, 0 points are awarded.
[0046] Weights are determined using expert experience or the Analytic Hierarchy Process (AHP), and the scores of the three dimensions are weighted and fused to obtain... , .
[0047] 2) Spatial correlation The power grid coverage area is divided into several power supply zones, if historical events i If the event belongs to the same partition as the current event, it receives 1 point; otherwise, based on geographical distance normalization, the score is 1 minus the ratio of the geographical distance between the historical event and the current event to the maximum geographical distance within the power grid coverage area. This score is used as the spatial correlation score. , .
[0048] 3) Equipment type matching degree If historical eventsi If the main equipment type (e.g., transformer, transmission line, circuit breaker) is exactly the same as the main equipment type affected by the current event, 1 point is awarded; if it belongs to the same major category (e.g., power equipment includes transformers and circuit breakers), 0.7 points are awarded; otherwise, 0 points are awarded. The score obtained is used as the equipment type matching degree. , .
[0049] 4) Weather similarity Collect historical events i The weather characteristics of the current event, including weather type and meteorological parameters, are used to form a feature vector. The cosine similarity of the vectors is then calculated to obtain the weather similarity. , .
[0050] 5) Load characteristic matching degree First, perform type matching to distinguish between industrial, residential, and commercial loads, including historical events. i 1 point is awarded if the event type is the same as the current event type; otherwise, 0 points are awarded.
[0051] Then, the Dynamic Time Warping (DTW) algorithm is used to calculate historical events. i The distance from the load curve of the current event is normalized to obtain the curve similarity.
[0052] The load characteristic matching degree is obtained by weighted summation of type matching score and curve similarity. , .
[0053] 6) Power grid operation status matching degree Extracting historical events i The current event's operational state characteristics are normalized to form a vector. The cosine similarity of the vectors is then calculated to obtain the power grid operational state matching degree. , .
[0054] 7) Severity similarity Get historical events i The severity indicators of the current event, including the number of households experiencing power outages, load loss, and outage duration, are normalized to the range [0,1], and their respective severity vectors are constructed. The cosine similarity of these vectors is then calculated to obtain the severity similarity. , .
[0055] Example 3 The difference between this embodiment and Embodiment 1 or Embodiment 2 is that the confidence level of the assessment is improved. The specific improvement method is as follows: First, the initial assessment confidence level is calculated based on the method in Example 1: or , Where C represents the confidence level, C∈[0,1] K To determine the number of historical events selected, sim i For the current power outage event and the first i Cosine similarity of historical events For the first i Improvements in the similarity between historical events and current power outage events.
[0056] Subsequently, the initial judgment confidence level was corrected based on similarity dispersion and event feature entropy to obtain an improved judgment confidence level. : , in, for K The standard deviation of the similarity of historical events measures the degree of dispersion of similarity; The maximum standard deviation of similarity is preset; The characteristic entropy of the current power outage event measures the uncertainty of the event and is obtained by normalizing the information entropy of multi-source data. The maximum feature entropy is preset.
[0057] Example 4 The electronic device of this invention includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in read-only memory (ROM) or loaded from a storage unit into random access memory (RAM). The RAM may also store various programs and data required for device operation. The CPU, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.
[0058] Multiple components in the device are connected to the I / O interface, including: input units such as keyboards and mice; output units such as various types of displays and speakers; storage units such as disks and optical discs; and communication units such as network interface cards (NICs), modems, and wireless transceivers. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0059] The processing unit executes the various methods and processes described above, such as methods S1 to S3. For example, in some embodiments, methods S1 to S3 may be implemented as computer software programs tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via ROM and / or a communication unit. When the computer program is loaded into RAM and executed by the CPU, one or more steps of methods S1 to S3 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute methods S1 to S3 by any other suitable means (e.g., by means of firmware).
[0060] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.
[0061] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.
[0062] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0063] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for predicting the impact range of power grid outages and making emergency decisions, characterized in that, Includes the following steps: S1. Collect multi-source heterogeneous data related to power outage events, perform entity recognition and relationship extraction, and construct a power outage event knowledge graph with triplets consisting of disturbance sources, influencing relationships, and affected objects as the core. S2, the knowledge graph of the power outage event is vectorized and embedded to construct a semantic vector space of historical power outage events; when a new power outage event occurs, its structured features are extracted and a real-time feature vector is generated, similar historical events are retrieved in the semantic vector space, and the prediction results of the power outage impact range and the assessment confidence level are dynamically output. S3, when the assessment confidence level is greater than or equal to a preset threshold, the historical best matching scheme is used as a structured emergency decision instruction; when the assessment confidence level is less than the preset threshold, the preliminary response scheme and the historical best matching scheme are merged to generate a structured emergency decision instruction.
2. The method for predicting the impact range of power grid outages and making emergency decisions according to claim 1, characterized in that, The sources of disturbance include typhoons, lightning strikes, equipment aging, or external damage; the affected objects include important power users and densely populated residential areas; the impact relationships include line tripping, equipment failure, voltage fluctuations, power transfer failures, power outages, and communication interruptions.
3. The method for predicting the impact range of power grid outages and making emergency decisions according to claim 1, characterized in that, S2 includes the following steps: The triples of each historical power outage event are aggregated into an event embedding vector to construct the semantic vector space of historical power outage events; Encode the features of the current power outage event into a real-time feature vector; The similarity between the current event and each historical event is calculated based on the real-time feature vector and the event embedding vector. Select the first preset number of historical events with the highest similarity; The predicted impact radius of the current power outage event is calculated based on the similarity of the selected historical events and their corresponding actual impact radii. The confidence level is calculated based on the similarity of the selected historical events.
4. The method for predicting the impact range of power grid outages and making emergency decisions according to claim 3, characterized in that, The method for calculating the predicted impact radius of the current power outage event is as follows: ， in, R pred To predict the radius of influence, R i For the first i The actual radius of influence of a historical event K To determine the number of historical events selected, sim i For the current power outage event and the first i The similarity between historical events.
5. The method for predicting the impact range of power outages and making emergency decisions according to claim 3, characterized in that, The method for calculating the confidence level is as follows: ， Where C represents the confidence level, C∈[0,1] K To determine the number of historical events selected, sim i For the current power outage event and the first i The similarity between historical events.
6. The method for predicting the impact range of power grid outages and making emergency decisions according to claim 1, characterized in that, In step S3, the method for determining the structured emergency decision-making instructions is as follows: ， in, D For structured emergency decision-making instructions; D know The historical best matching scheme invoked by the knowledge-driven layer includes the historical best emergency repair sequence and power protection measures for matching disturbance types. D data The initial response plan output by the data-driven layer is dynamically generated based on the real-time distribution of power-outage users and the location of emergency repair resources. C To assess the confidence level, , where is the confidence threshold; , which is the fusion weighting coefficient.
7. The method for predicting the impact range of power grid outages and making emergency decisions according to claim 1, characterized in that, The relation extraction employs a joint model of named entity recognition and dependency parsing to automatically extract triples from unstructured text.
8. The method for predicting the impact range of power outages and making emergency decisions according to claim 1, characterized in that, The method further includes: establishing a version management mechanism for the power outage event knowledge graph, supporting automatic updates after the injection of new events and retrospective verification of historical versions. The version management mechanism involves writing the manually reviewed new event triplet into the new version of the knowledge base after each emergency response loop is closed, and retaining the version number, timestamp, and change log.
9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 8.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 8.