Fault handling auxiliary decision-making system and method

Abstract

The invention discloses a fault handling auxiliary decision-making system and method, relates to the technical field of power equipment fault handling, and solves the problem that static procedure knowledge and dynamic power grid data are disjointed in an existing method. According to the embodiment of the invention, regulations, plans and historical cases of the power system are fused through the knowledge graph, so that the system can automatically determine compliant and effective processing steps based on the current fault information; and by combining simulation deduction of the real-time power grid model, the effect and risk of the treatment measures are pre-judged, so that the suggested scheme is changed from theoretical feasibility to safe feasibility in the current power grid state, effective fusion of static knowledge and a dynamic power grid model is realized, and scientificity and high efficiency of power system fault treatment are improved.

Description

Technical Field

[0001] This invention relates to the field of power equipment fault handling technology, and in particular to a fault handling auxiliary decision-making system and method. Background Technology

[0002] With the rapid development of society and the economy, the scale of the power system is constantly expanding, and the power grid structure is becoming increasingly complex, placing increasingly stringent requirements on the security and reliability of power supply. When a power system failure occurs, the ability to quickly and accurately formulate a scientific and reasonable handling plan is directly related to controlling the scope of the failure's impact, shortening the duration of power outages, and ensuring the safe and stable restoration of the power grid.

[0003] Currently, in the process of handling power system faults, operation and maintenance personnel mainly rely on consulting a large number of paper or electronic regulations, emergency plans, and other materials, combined with their own accumulated work experience, to formulate handling solutions. Among them, regulations and emergency plans serve as the core basis for power system fault handling, covering general handling principles, operating procedures, and safety specifications for various faults, providing basic guidance for fault handling. At the same time, the scientific nature and feasibility of fault handling are highly dependent on the current real-time grid topology and electrical operating status, such as dynamic data like grid power flow distribution, reserve capacity in various regions, and equipment operating parameters. Only by formulating handling solutions based on the actual real-time grid status can the effectiveness and operability of the solutions be ensured.

[0004] However, regulations, emergency plans, and other similar documents are static knowledge resources. Their content is universal and cannot be directly adapted to the real-time power grid status under different fault scenarios. Meanwhile, dynamic data such as real-time power grid topology and power flow distribution are scattered across different monitoring systems, failing to be effectively integrated with static regulations. This forces maintenance personnel to spend a significant amount of time switching between reviewing static documents and integrating dynamic data when handling faults. This not only leads to low decision-making efficiency but also a high reliance on personal experience, making it easy for human error to result in unreasonable handling plans, thereby amplifying the impact of the fault.

[0005] Therefore, a fault handling auxiliary decision-making system and method are needed. Summary of the Invention

[0006] To address the disconnect between static procedural knowledge and dynamic power grid data in existing technologies, this invention provides a fault handling auxiliary decision-making system and method. This system effectively integrates static knowledge with dynamic power grid models, improving the scientific rigor and efficiency of power system fault handling. The specific technical solution is as follows: In a first aspect, embodiments of this application provide a fault handling auxiliary decision-making system, which includes a knowledge base construction module, a power grid status perception module, a fault scenario formatting module, a two-layer matching and evaluation engine, and a report generation module; the two-layer matching and evaluation engine is connected to the knowledge base construction module, the power grid status perception module, the fault scenario formatting module, and the report generation module, respectively. This knowledge base construction module is used to acquire power system regulations, fault handling plans, and historical cases, and to construct a structured handling knowledge graph based on these regulations, plans, and cases. The power grid state perception module is used to acquire electrical data of the power grid before and after a fault, and to generate a power grid model based on the electrical data. The fault scenario formatting module is used to receive power grid fault information and generate a standard query vector based on the power grid fault information; wherein, the standard query vector includes the faulty equipment, the fault type, and the sequence of protection actions taken on the faulty equipment after the fault. The two-layer matching and evaluation engine is used to match the standard query vector with the disposal knowledge graph to obtain preliminary disposal steps, and to perform safety simulation and risk assessment on the preliminary disposal steps based on the power grid model to obtain multiple disposal paths; The report generation module is used to perform multi-objective optimization and ranking of the disposal path and generate a comprehensive auxiliary decision-making report containing executable instructions and risk explanations; wherein, the executable instructions correspond to the disposal steps of the disposal path.

[0007] Preferably, the knowledge base construction module is specifically used for: extracting entities and corresponding relationships in the power field from the text data of the power system regulations, the fault handling plan, and the historical cases. These relationships at least include conditions, actions, and consequences. The conditional relationship connects the entity performing the handling action with the entity providing the preconditions; the action relationship connects the entity describing the fault scenario with the entity performing the handling action; and the consequence relationship connects the entity performing the handling action with the entity describing the expected result and expected risk. The module also extracts condition-action pairs from the text data to describe the handling rules; and uses the entity as the first node, and the relationship as the second node. The system is used as the first side to construct the conceptual layer; the condition-action pair is used as the second node, and the logical relationship between the second nodes is used as the second side to construct the rule layer; the historical case is used as an independent case subgraph, and the similarity relationship between the case subgraphs is used as the third side to construct the case layer. The case subgraph includes scenario nodes for describing faults, response nodes for describing protection actions or handling actions, and state nodes for recording snapshots of the power grid state before and after the action; the fourth side is constructed based on the semantic relationship between the first node and the second node, and the fifth side is constructed based on the similarity relationship between the case subgraph and the second node, thus obtaining the handling knowledge graph.

[0008] Preferably, the power grid state perception module is specifically used for: upon receiving a fault event notification, setting a pre-fault time window, a post-fault transient time window, and a post-fault steady-state time window based on the fault time; collecting first multi-source electrical data within the pre-fault time window, second multi-source electrical data within the post-fault transient time window, and third multi-source electrical data within the post-fault steady-state time window from the energy management system, wide-area measurement system, and data acquisition and monitoring system; constructing a basic model based on the third multi-source electrical data; and dynamically correcting the basic model based on the first and second multi-source electrical data to obtain the power grid model.

[0009] Preferably, the fault scenario formatting module is specifically used for: parsing the fault information based on the parser corresponding to the data source type of the fault information; semantically labeling the parsing results based on a preset power fault domain ontology framework to obtain the faulty device, the fault type, the protection action sequence, and the events that occur during the fault handling process; constructing the causal-temporal chain between the events; and constructing the standard query vector based on the faulty device, the fault type, the protection action sequence, the event, and the causal-temporal chain.

[0010] Preferably, the two-layer matching and evaluation engine includes a connected knowledge matching layer and a real-time inference and evaluation layer. The knowledge matching layer is used to calculate the similarity between the standard query vector and the conditions in the second node; based on the similarity, the average matching degree between the second node and associated historical cases, and the confidence level of the second node, the activation intensity of the second node is calculated; based on the activation intensity, a set of preliminary handling steps is generated; the real-time inference and evaluation layer is used to perform sequential simulation of the steps in the set of preliminary handling steps based on the power grid model, and after each simulation step, the power flow calculation is called to verify the equipment overload and voltage over-limit safety constraints, while calculating the comprehensive risk value after the step is executed; based on the comprehensive risk value, multiple handling paths from the post-fault state to the target recovery state of the power grid model are evaluated and determined.

[0011] Preferably, the report generation module is specifically used for: receiving multiple disposal paths and corresponding evaluation indicators output by the two-layer matching and evaluation engine, wherein the evaluation indicators include at least the total path risk, total operation time, and expected power restoration; adjusting the weights of each evaluation indicator based on the urgency of the current power grid fault, and ranking the disposal paths using the approximation ideal solution ranking method to obtain a Pareto optimal solution set; and generating the comprehensive auxiliary decision-making report based on the Pareto optimal solution set, wherein the report includes an execution summary for decision-makers, actionable operation suggestions for operators, and a technical detail including risk attribution analysis and rule tracing.

[0012] Preferably, the formula for calculating the comprehensive risk value includes: ;in, This indicates the degree to which the m-th type of safety constraint is exceeded. This is the penalty coefficient for the corresponding constraint.

[0013] Secondly, embodiments of this application provide a fault handling auxiliary decision-making method, applied to the system described in the first aspect, the method comprising: The knowledge base construction module acquires power system regulations, fault handling plans, and historical cases, and constructs a structured handling knowledge graph based on these resources. The power grid state perception module acquires electrical data of the power grid before and after a fault, and generates a power grid model based on this data. The fault scenario formatting module receives power grid fault information and generates a standard query vector based on this information. This standard query vector includes the faulty equipment, the fault type, and the sequence of protective actions taken against the faulty equipment after the fault. The two-layer matching and evaluation engine matches this standard query vector with the handling knowledge graph to obtain preliminary handling steps. Based on the power grid model, it performs safety simulation and risk assessment on these preliminary handling steps to obtain multiple handling paths. The report generation module performs multi-objective optimization and ranking on these handling paths and generates a comprehensive auxiliary decision-making report containing executable instructions and risk explanations. The executable instructions correspond to the handling steps of the handling path.

[0014] Thirdly, embodiments of this application provide a computing device, including: a memory for storing a program; and a processor for loading the program to execute the method as described in the second aspect.

[0015] Fourthly, embodiments of this application provide a computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to perform the method described in the second aspect.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: by integrating power system regulations, contingency plans, and historical cases through knowledge graphs, the system can automatically determine compliant and effective handling steps based on current fault information; and by combining real-time power grid model simulations, the effectiveness and risks of handling measures can be predicted, making the proposed solutions not only theoretically feasible but also safe and feasible under the current power grid conditions, thus achieving effective integration of static knowledge and dynamic power grid models and improving the scientificity and efficiency of power system fault handling. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0018] Figure 1 A system architecture diagram of a fault handling auxiliary decision-making system provided in this application embodiment; Figure 2 A flowchart illustrating a fault handling auxiliary decision-making method provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a computing device provided in an embodiment of this application. Detailed Implementation

[0019] 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 are within the scope of protection of the present invention.

[0020] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0021] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0022] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0023] To address the disconnect between static procedural knowledge and dynamic power grid data in traditional methods, this invention provides a fault handling auxiliary decision-making system and method that can effectively integrate static knowledge with dynamic power grid models, thereby improving the scientific rigor and efficiency of power system fault handling.

[0024] Please see Figure 1 , Figure 1This is a system architecture diagram of the fault handling auxiliary decision-making system provided in the embodiments of this application. Figure 1 As shown, the system includes a knowledge base construction module 10, a power grid status perception module 20, a fault scenario formatting module 30, a two-layer matching and evaluation engine 40, and a report generation module 50; the two-layer matching and evaluation engine 40 is connected to the knowledge base construction module 10, the power grid status perception module 20, the fault scenario formatting module 30, and the report generation module 50, respectively.

[0025] The aforementioned modules and engines can be deployed on the same computing device or on different computing devices; this application does not specifically limit this.

[0026] Specifically, the computing device can be a terminal or a server. The terminal can be a personal computer, laptop, tablet, etc.; the server can be a blade server, high-density server, rack server, cabinet server, general-purpose server, graphics processing unit (GPU) server, data processing unit (DPU) server, or artificial intelligence (AI) server, etc.

[0027] Specifically, the computing device can communicate with the power system and obtain power grid status information and fault information from the relevant database of the power system.

[0028] Among them, the knowledge base construction module 10 is the brain of the system, responsible for transforming unstructured text knowledge into a structured processing knowledge graph that can be understood and reasoned by machines, providing a static knowledge foundation for the two-layer matching and evaluation engine 40.

[0029] Specifically, the knowledge base construction module 10 is used to acquire power system regulations, fault handling plans and historical cases, and to construct a structured handling knowledge graph based on the power system regulations, fault handling plans and historical cases.

[0030] Among them, the knowledge base construction module 10 can obtain documents such as power system safety regulations, dispatching operation procedures, and substation on-site handling plans from official regulations and contingency plans; it can obtain structured or semi-structured fault analysis reports on historical faults from the production management system; and it can also obtain experiential knowledge from outputs such as relevant meeting minutes, accident event databases, and expert interview recordings transcribed into text.

[0031] Then, the knowledge base construction module 10 can convert all documents into a uniform plain text format, remove irrelevant headers, footers, table lines, and special characters, and then divide large blocks of text into chapters, clauses, and natural sentences to form text units to be processed; then, natural language processing technology is applied for processing.

[0032] Preferably, the knowledge base construction module 10 is specifically used to extract entities and corresponding relationships in the power field from the text data of the power system regulations, the fault handling plan, and the historical cases.

[0033] The knowledge base construction module 10 can first use a pre-trained Named Entity Recognition (NER) model to identify entities in the power sector from text units. For example, the basic entities in the power sector include equipment entities, fault or event entities, action entities, and quantity entities. Quantity entities can include overload a%, voltage below bpu, and delay c seconds.

[0034] Then, based on a predefined dependency system, the grammatical dependencies between words in a sentence are analyzed using rule templates. This dependency system is a fine-grained relation type specifically designed for fault handling, including relationships with conditions (hasCondition), actions (hasAction), leads to results (leadsTo), objects (hasObject), action parameters (hasParameter), and similarities (isSimilarTo). Specifically, the conditional relationship is used to connect the entity of the action and the entity of the precondition, specifically pointing from the entity of the action to the entity of the precondition; the action relationship is used to connect the entity of the fault scenario and the entity of the action, specifically pointing from the entity of the fault scenario to the entity of the action; the result-leading relationship is used to connect the entity of the action and the entities of the expected result and expected risk; the action parameter relationship is used to connect the entity of the action and the quantity entity; and the similarity relationship is used to connect two matching fault scenarios.

[0035] Specifically, a fault scenario is a composite node that represents a specific fault state scenario requiring decision-making and response. It includes a fault fact sub-entity, a system response state sub-entity, and an environmental context sub-entity. For example, the fault fact sub-entity includes the faulty device, the fault type, and the fault location.

[0036] Then, the knowledge base construction module 10 extracts condition-action pairs from the text data to describe the disposition rules. Understandably, when a rule is complex and multi-layered, containing multiple conditions and / or multiple actions, a tuple of condition set-action set can be generated based on the complex rule.

[0037] Then, the knowledge base construction module 10 can construct a concept layer using the entity as the first node and the relationship as the first side; construct a rule layer using the condition-action pair as the second node and the logical relationship between the second nodes as the second side; construct a case layer using the historical case as an independent case subgraph and the similarity relationship between the case subgraphs as the third side, where the case subgraph includes scenario nodes for describing faults, response nodes for describing protection actions or handling actions, and state nodes for recording the state of the power grid before and after the action; construct a fourth side based on the semantic relationship between the first node and the second node, and construct a fifth side based on the similarity relationship between the case subgraph and the second node, to obtain the handling knowledge graph.

[0038] The concept layer defines the relevant entities for fault handling, ensuring the consistency of the knowledge structure; the rule layer stores rule nodes and their links derived from condition-action pairs or condition-action tuples, representing static procedural knowledge; the case subgraph in the case layer not only contains fault scenarios and handling actions, but also connects to the rule layer and other cases through the isSimilarTo relationship, and is associated with the original handling results and evaluation reports.

[0039] In this way, the handling rules and cases can be linked together. When receiving fault information in the future, the system can first match similar handling rules or cases, and then match the other case based on the one with the highest similarity.

[0040] Once the system is running, each successful case is formatted and sent to the knowledge base construction module 10. The knowledge base construction module 10 links the new case subgraph to existing rules and case nodes in the graph through similarity calculation. If the handling mode of the new case conflicts with existing rules, a prompt will be generated and submitted to experts for review. Experts can decide whether to revise the rules, add exceptions, or store the new case as a special case under a specific scenario in the case layer.

[0041] Among them, the power grid state perception module 20 is the system's sensory nerve, responsible for acquiring real-time electrical quantities and topology of the power grid from external systems, providing a dynamic data foundation for the two-layer matching and evaluation engine 40. Specifically, the power grid state perception module 20 is used to acquire electrical data of the power grid before and after a fault, and generate a power grid model based on this electrical data.

[0042] Preferably, the power grid state perception module 20 is specifically used for: upon receiving a fault event notification, setting a pre-fault time window, a post-fault transient time window, and a post-fault steady-state time window based on the fault time; collecting first multi-source electrical data within the pre-fault time window, second multi-source electrical data within the post-fault transient time window, and third multi-source electrical data within the post-fault steady-state time window from the energy management system, the wide-area measurement system, and the data acquisition and monitoring system; constructing a basic model based on the third multi-source electrical data; and dynamically correcting the basic model based on the first and second multi-source electrical data to obtain the power grid model.

[0043] Among them, the power grid status sensing module 20 continuously listens for fault event notifications from the fault diagnosis system. When it receives the fault event notification, the power grid status sensing module 20 will cancel the polling mode and send supplementary data requests with specific time window parameters to each data source.

[0044] Specifically, the power grid state sensing module 20 can obtain steady-state electrical variable data of the power grid from the Energy Management System (EMS), including active power, reactive power and voltage amplitude at all measurement points; it can also obtain synchronous phasor measurement data of key nodes from the Wide Area Measurement System (WAMS), including voltage phase angle and frequency f; and it can also obtain the open / close status signals of all circuit breakers, disconnectors, grounding switches and other switching equipment in the power grid from the Supervisory Control and Data Acquisition (SCADA) system.

[0045] For example, the pre-fault time window is 1 second before the fault time, the post-fault transient time window is the time from the occurrence of the fault to the fault clearing, and the post-fault steady-state time window is the time from the completion of fault clearing to the system oscillation subsiding and reaching a new steady state.

[0046] Then, the power grid state sensing module 20 can construct a basic model based on the third multi-source electrical data under steady-state conditions after the fault. This third multi-source electrical data includes the final positions of all switches, the new power flow distribution, and voltage levels after the system oscillation subsides, used to describe which islands the power grid has been divided into, the voltage of each bus, and how the power flow has been redistributed.

[0047] Then, the power grid status sensing module 20 can establish a baseline of the power grid health status based on the first multi-source electrical data, and adjust the nominal parameters of each component in the basic model in combination with the fault impact status reflected by the second multi-source electrical data.

[0048] Preferably, the power grid status sensing module 20 can determine whether the line is a permanent fault or a transient fault by analyzing the fault current waveform and whether the line still has an insulation fault after the fault is cleared; if it is a permanent fault, the line is disconnected in the basic model; if it is a transient fault, the line is marked as testable for closing in the basic model.

[0049] Preferably, the power grid status sensing module 20 can calculate the magnitude and duration of the short-circuit current flowing through the transformer; if the transformer is subjected to a short-circuit current exceeding its tolerance capacity, a vulnerability marker is added to the transformer in the basic model, and it is recommended to avoid subjecting it to heavy loads or impacts in the short term in subsequent simulations.

[0050] For equipment carrying large short-circuit currents, the winding resistance can be temporarily adjusted based on the thermal effect of the short-circuit current to simulate the heating effect. If a generator continues to be strongly excited during a fault, the upper limit of reactive power output can be temporarily lowered in its model to simulate rotor overheating limitation; if the power angle fluctuates violently, its damping coefficient can be temporarily increased.

[0051] In this way, after parameter adjustment, a high-fidelity digital mirror model of the power grid can be obtained, which deeply integrates real-time data after the fault with the fault analysis conclusions.

[0052] Specifically, the resulting power grid model can include node admittance matrices, node injection vectors, and a set of switching states. The node injection vectors contain the net active and reactive power of each node.

[0053] Among them, the fault scenario formatting module 30 is the system's translator, responsible for receiving the original report from the external fault diagnosis system, extracting its key information and encoding it into a machine-readable standard query vector, which serves as the problem description that triggers the entire decision-making process.

[0054] Specifically, the fault scenario formatting module 30 is used to receive power grid fault information and generate a standard query vector based on the power grid fault information; wherein, the standard query vector includes the faulty equipment, the fault type, and the sequence of protection actions taken on the faulty equipment after the fault.

[0055] Preferably, the fault scenario formatting module 30 is specifically used for: parsing the fault information based on the parser corresponding to the data source type of the fault information; semantically labeling the parsing results based on a preset power fault domain ontology framework to obtain the faulty device, the fault type, the protection action sequence, and the events that occur during the fault handling process; constructing the causal-temporal chain between the events; and constructing the standard query vector based on the faulty device, the fault type, the protection action sequence, the event, and the causal-temporal chain.

[0056] The fault scenario formatting module 30 can continuously monitor fault event notifications from external systems through predefined interfaces, message queues, or service buses.

[0057] The fault event notification consists of an event announcement section and a data payload section. Specifically, the event announcement section is a structured message header containing metadata such as event ID, occurrence time, data source ID, and event type; the data payload section is a detailed description of the fault, which can be in the form of structured data, semi-structured text, or a free text report.

[0058] A fault event refers to a specific key link, state change, or operational behavior that objectively exists, can be recorded and described, during the entire process of the occurrence, development, and handling of a power system fault. It is the basic unit that constitutes the complete fault process in a fault report.

[0059] The fault scenario formatting module 30 can preset a multimodal resolver pool and a resolver routing table. The routing table is used to dynamically select the most suitable resolver instance based on the form of the fault description.

[0060] Among them, the fault scenario formatting module 30 pre-defines an ontology framework for the power fault domain, defining core semantic roles, including triggerer, sufferer, action, fault attributes, and state.

[0061] Specifically, the trigger is the protection device or the fault itself, the receiver is the equipment whose action is performed, the fault attributes include type, phase, ranging value, etc.; the status includes the state before and after the fault, as well as whether the action was successful or failed, etc.

[0062] The fault scenario formatting module 30 can perform fine-grained semantic role labeling on entities based on the ontology framework after performing named entity recognition and syntactic analysis on the text data. For example, for the event described in the sentence "the differential protection action tripped the 2201 switch", the parsing result is: [Trigger: differential protection, action: tripped, bearer: 2201 switch, action result: success].

[0063] The fault report contains multiple events that occur in chronological order. The fault scenario formatting module 30 can reconstruct discrete events into a directed event chain, namely the causal-temporal chain, by analyzing the event timestamps and logical words in the text data.

[0064] Among them, the fault scenario formatting module 30 can define a standardized JSON Schema or similar data structure as a template for standard query vectors.

[0065] Specifically, the template can be a multi-dimensional vector template, which includes a fault fact layer, an inference evidence layer, a spatiotemporal context layer, and an event logic graph. The fault fact layer contains fault fact entities directly extracted from text data, such as the set of faulty equipment, possible fault types, and protection action sequences. The inference evidence layer contains the original statements, confidence levels, and evidence sources used to support the fault facts. The spatiotemporal context layer contains automatically associated fault times, plant names, and voltage levels. The event logic graph is used to embed standard query vectors in a graph structure, with events as nodes and time-series or causal relationships as edges, to fully preserve the complex relationships between events.

[0066] Understandably, the confidence level of the original statement can be due to the inherent uncertainty of the data. Some advanced fault recorders or protection devices will include an accuracy range or confidence level indicator when outputting fault location results. Alternatively, it can be obtained based on the mapping of fuzzy words in the statement and the analysis of sentence strength. For example, words such as "sure" and "certain" can map the confidence level of the corresponding statement to 0.95, while words such as "possibly" and "probably" can map the confidence level to 0.7.

[0067] After filling the aforementioned multi-dimensional information into the standard query vector template, the standard query vector corresponding to the real-time power grid fault information can be obtained. Then, the fault scenario formatting module 30 can link and align the entity names in the standard query vector with the entity names in the handling knowledge graph.

[0068] Among them, the two-layer matching and evaluation engine 40 is the central processing unit of the system. It receives problem descriptions from the fault scenario formatting module 30 and real-time data from the power grid state perception module 20, and calls knowledge in the knowledge graph to execute the key matching-evaluation-optimization loop.

[0069] Specifically, the two-layer matching and evaluation engine 40 is used to match the standard query vector with the disposal knowledge graph to obtain preliminary disposal steps, and to perform safety simulation and risk assessment on the preliminary disposal steps based on the power grid model to obtain multiple disposal paths.

[0070] Preferably, the two-layer matching and evaluation engine 40 includes a connected knowledge matching layer 41 and a real-time inference and evaluation layer 42. The knowledge matching layer 41 is used to calculate the similarity between the standard query vector and the conditions in the second node. Based on the similarity, the average matching degree between the second node and associated historical cases, and the confidence level of the second node, the activation intensity of the second node is calculated. Based on the activation intensity, a set of preliminary handling steps is generated. The real-time inference and evaluation layer 42 is used to perform sequential simulation of the steps in the set of preliminary handling steps based on the power grid model. After each simulation step, power flow calculation is called and the equipment overload and voltage over-limit safety constraints are verified. At the same time, the comprehensive risk value after the step is executed is calculated. Based on the comprehensive risk value, multiple handling paths from the post-fault state to the target recovery state of the power grid model are evaluated and determined.

[0071] Among them, the knowledge matching layer 41 can first use the elements in the standard query vector as query conditions to initiate a subgraph matching query to the knowledge graph to find all related rule nodes in the knowledge graph and obtain the relevant rule set.

[0072] Then, the knowledge matching layer 41 matches each rule in the set. Calculate an activation strength The calculation formula includes: ; in, , The weighting coefficients for each piece of evidence. For standard query vectors and rules Structured similarity of preconditions In accordance with the rules The matching degree and result weight of related historical success cases, For rules The confidence factor itself.

[0073] Specifically, Standard query vectors and rules can be used. The second node's preconditions are a weighted sum of entity matching, relation matching, and attribute matching. Entity matching can be the Jaccard similarity coefficient or semantic distance between the entity set in the standard query vector and the entity set in the rule preconditions. Relation matching can be obtained by detecting and comparing whether the relationships between the two entities are consistent. Attribute matching can be obtained by comparing the values ​​of key attributes. For example, if all key attributes match, it is 1; if partially matched, it is 0.6; and if not matched, it is 0. Key attributes include fault type and faulty equipment.

[0074] Specifically, The calculation formulas include: ;in, This represents the similarity between the current standard query vector and its associated historical case k-scenario vectors. This is the weight value for whether the handling of this historical case was successful or not, with 1 for success and 0 for failure; Let k be the time decay factor for historical case k. It is the time difference between the current time and the time when the case occurred.

[0075] Specifically, during knowledge base construction, the knowledge base construction module 10 can, based on domain experts, refine the rules according to their source and generality. The confidence level is initially assigned; then, the rule is actually used. The actual implementation effect of the subsequent treatment plan is used to dynamically adjust the confidence level, resulting in a real-time confidence factor. .

[0076] Then, all activation strengths Rules that exceed the threshold are activated in parallel, and their corresponding action sets are extracted to form a preliminary candidate pool of action steps.

[0077] Then, for each handling step, the real-time simulation and evaluation layer 42 can instantiate it into a specific operable instruction. The rule conclusions in the knowledge graph are usually abstract, such as load transfer. The real-time simulation and evaluation layer 42 can combine the specific context in the query vector with the power grid model to obtain the specific operation object of the action, and then instantiate it to obtain a specific operable instruction. For example, for the context "the fault occurred at the 220kV I bus", the connection relationship of "220kV I bus" in the power grid model can be traversed to find all its outgoing lines, and then the target lines for load transfer can be selected according to the reserve capacity attribute, thereby instantiating the load transfer step as transferring the load from line L1 to lines L2 and L3.

[0078] Then, the real-time simulation and evaluation layer 42 can sort each disposal step according to the activation intensity, rule source authority and step type, and disposal principle of the corresponding rule in the candidate pool of preliminary disposal steps; or it can normalize these factors and calculate a weighted score to obtain a ranking score, and sort the rules based on the ranking score.

[0079] The handling principles include prioritizing safety before restoring power supply, addressing the main lines before the branches, and handling high voltage before low voltage.

[0080] In this way, we can obtain a preliminary set of processing steps that have been pre-sorted and instantiated, where each step is associated with its source rule and activation strength.

[0081] Then, the real-time simulation evaluation layer 42 can consider the temporal dependencies and electrical state continuity between operations, on a copy of the power grid model, according to the order of each step in the set of preliminary handling steps. Execute the following loop: 1. Execute on a copy of the power grid model .

[0082] 2. Perform power flow calculations on the new power grid state after the operation.

[0083] 3. Check critical equipment overload constraints, voltage over-limit constraints, and N-1 static safety constraints; among these, equipment overload constraints are checked for all lines and transformers. Whether it is valid. (Among them) Apparent power, The rated capacity is used; the voltage over-limit constraint checks whether the voltage exceeds the limit for all nodes; the N-1 static safety constraint simulates the possible failure of any component in the future and checks whether the system can remain stable.

[0084] 4. Based on the nature of the constraint violation, for Calculate a comprehensive risk value Risk value The immediate security pressure faced by the system after performing this step was quantified.

[0085] Preferably, the formula for calculating the comprehensive risk value includes: ;in, This indicates the degree to which the m-th type of safety constraint is exceeded. This is the penalty coefficient for the corresponding constraint.

[0086] 5. If If it is below the threshold, then step If deemed acceptable, the resulting grid state will serve as the new baseline state for evaluating subsequent steps. .if If the risk is too high, it may be marked as infeasible, and alternative steps or adjustments to the operating parameters may be sought.

[0087] Since some steps may have multiple alternatives, the real-time simulation and evaluation layer 42 naturally forms a solution path search tree during the simulation process. Each node represents a power grid state, and each edge represents an operational step and its risk. At this point, the real-time simulation and evaluation layer 42 can use a heuristic search algorithm to explore multiple paths from the post-fault state to the target recovery state. For each complete path, its total path risk and key performance indicators are calculated; then, branches with extremely high risk or extremely poor performance are dynamically pruned to obtain multiple feasible solution paths.

[0088] Understandably, these feasible solutions cannot surpass each other in terms of total risk, estimated recovery time, operational complexity, and estimated power restoration; once the first solution is superior to the second solution in all indicators, the second solution will be eliminated.

[0089] The output of the two-layer matching and evaluation engine 40 can be a decision package with rich context, which includes 1-3 complete disposal paths ordered by priority, which can be determined based on the total risk of the path, the expected recovery time and the operational complexity; a comparison of key electrical quantities before and after simulation, risk value and expected time for each step of each path; equipment that needs to be monitored during execution; and the original procedure clauses and cases associated with each operation step.

[0090] Among them, the report generation module 50 is the system's decision expression and executor. It is responsible for receiving the engine's processing results and transforming them into an optimized handling sequence and comprehensive report with clear risk ranking that can be directly used by the scheduler.

[0091] Specifically, the report generation module 50 is used to perform multi-objective optimization and ranking of the disposal path and generate a comprehensive auxiliary decision-making report containing executable instructions and risk explanations; wherein the executable instructions correspond to the disposal steps of the disposal path.

[0092] Preferably, the report generation module 50 is specifically used for: receiving multiple disposal paths and corresponding evaluation indicators output by the two-layer matching and evaluation engine, wherein the evaluation indicators include at least the total path risk, total operation time, and expected power restoration; adjusting the weights of each evaluation indicator based on the urgency of the current power grid fault, and ranking the disposal paths using the approximation ideal solution ranking method to obtain a Pareto optimal solution set; and generating the comprehensive auxiliary decision-making report based on the Pareto optimal solution set, wherein the report includes an execution summary for decision-makers, actionable operation suggestions for operators, and a technical detail including risk attribution analysis and rule tracing.

[0093] The level of urgency can be determined based on instability indicators such as frequency drop and load loss ratio of the power grid. When these instability indicators exceed the threshold, it can be determined as an emergency mode, and the weight of the total operation time will be increased. Conversely, it can be determined as a safe mode, and the weight of the total path risk will be the highest.

[0094] It should be noted that, in practice, the system architecture of the fault handling auxiliary decision-making system can be any architecture, including... Figure 1 A similar architecture to that in [the text]. The embodiments of this application do not limit the specific composition of this system architecture. Furthermore, Figure 1 The architectural components shown do not constitute a limitation on the system architecture, except... Figure 1In addition to the devices shown, the system architecture may include more or fewer modules than illustrated.

[0095] In this embodiment, by integrating procedures, contingency plans, and historical cases using a knowledge graph, the system can automatically determine compliant and effective handling steps based on current fault information. Furthermore, by combining real-time power grid model simulations, the system can predict the effectiveness and risks of handling measures, transforming theoretically feasible solutions into safe and feasible solutions under the current power grid conditions. This achieves effective integration of static knowledge and dynamic power grid models, enhancing the scientific rigor and efficiency of power system fault handling.

[0096] In a specific embodiment, the operation of the above system can be divided into three stages, as follows: Phase 1: Fault Triggering.

[0097] After a power grid fault occurs, the protection device executes its protective action, and the fault is cleared. Then, an external independent fault diagnosis system performs fault analysis and generates a preliminary report containing information such as the faulty equipment, fault type, and protection action sequence.

[0098] For example, if a permanent ground fault occurs in phase A of the 220kV voltage bus, the protection device fails to perform the reclosing protection action.

[0099] The fault scenario formatting module 30 receives and parses the report. Specifically, after selecting the corresponding parser, the fault scenario formatting module 30 extracts the key fields from the report according to the predefined schema and generates a structured standard query vector. For example: {Faulty equipment: "220kV voltage bus", Fault type: "AG", Protection action: ["Differential protection", "Reclosing"], Action result: "Failure"}.

[0100] At the same time, the power grid status sensing module 20 is triggered synchronously, and it immediately requests the EMS and WAMS systems to obtain real-time snapshot data of the entire network before and after the fault, including the voltage of each node, the power flow of each branch, and the on / off status of all switches, and builds or updates a real real-time power grid calculation model based on this.

[0101] Phase Two: Dual-Engine Collaborative Decision Making.

[0102] After acquiring the outputs from the power grid status perception module 20 and the fault scenario formatting module 30, the two-layer matching and evaluation engine 40 begins operation. First, the graph matching layer 41 starts, performing graph query and semantic matching between the standard query vector and the handling knowledge graph. For example, it matches the handling rule nodes associated with nodes such as "220kV line", "permanent grounding fault", and "reclosing failure" in the knowledge graph, such as "Article XX of the Dispatch Regulations: A forced power restoration should be attempted once", and the historical case node "A similar fault occurred on X month X day, 2023, and forced power restoration was successful after load switching".

[0103] Based on the matching results, the graph matching 41 can form a preliminary set of handling steps. For example: [Step 1: Confirm the isolation status of the fault point, Step 2: Attempt to force power restoration once, Step 3: If the force restoration fails, consider transferring the load to the backup line...].

[0104] Then, the real-time simulation and evaluation layer 42 is activated to obtain the preliminary set of handling steps and the real-time power grid calculation model; and performs the following operations automatically and sequentially: a) Take out a procedure, such as closing the contact switch K123.

[0105] b) Simulate the operation in a real-time power grid model, i.e. change the state of switch K123 in the model to "closed".

[0106] c) Call the built-in power flow calculation engine to perform fast power flow calculation and deduce the system state after the operation.

[0107] d) Assess risks: Calculate and check whether critical electrical quantities exceed limits under the simulated state, such as whether related lines are overloaded, whether the voltage of critical nodes drops to dangerous levels, and whether the system stability margin is sufficient.

[0108] e) Generate a risk assessment label for this step, such as: After execution, the load rate of line L45 will reach 115%, there is an overload risk, and the risk level is high.

[0109] f) Repeat process ae for each step in the set of preliminary treatment steps.

[0110] After completing this operation cycle, the real-time simulation and evaluation layer 42 can obtain a set of disposal steps with detailed risk assessment labels.

[0111] Phase 3: Report generation and output.

[0112] The report generation module 50 receives a set of handling steps with risk assessment, and filters, sorts, and merges all steps according to a predefined strategy to generate a final comprehensive decision support report. Specifically, the predefined strategy can be a priority strategy, such as prioritizing the fastest power restoration option or prioritizing the option with the lowest risk.

[0113] For example, steps with a "high" risk level can be downgraded or marked with a warning; steps that can be executed in parallel can be packaged together; ultimately forming an executable sequence of actionable suggestions ordered by recommended priority.

[0114] The report includes not only the optimized sequence of actions, but also the basis for each step (linked to procedural clauses), risk assessment, required operational targets, and necessary contextual information extracted from the real-time model.

[0115] Then, the report generation module 50 pushes the report to the dispatcher through the human-machine interface. The dispatcher makes decisions and executes them based on the report. If the handling is successful, the fault scenario, the measures taken, and the final result can be used as a new success case. After review, the case is fed back to the knowledge base construction module to update and enrich the knowledge graph, enabling the system to learn continuously.

[0116] By combining graph matching with real-time simulation and evaluation, the decision-making process is upgraded from rule-based retrieval to sandbox simulation and optimization based on the current state. This transforms the process from a simple information query into a dynamic planning process that considers constraints and pursues the optimal action steps.

[0117] The system portion of the embodiments of this application has been described above. The method portion of the embodiments of this application will be described below.

[0118] Please see Figure 2 , Figure 2 This application provides a flowchart illustrating a fault handling auxiliary decision-making method, which is applied to the aforementioned system. Figure 2 As shown, the method includes: Step 201: The knowledge base construction module acquires power system regulations, fault handling plans, and historical cases, and constructs a structured handling knowledge graph based on these regulations, plans, and cases.

[0119] Step 202: The power grid status perception module acquires electrical data of the power grid before and after the fault, and generates a power grid model based on the electrical data.

[0120] Step 203: The fault scenario formatting module receives power grid fault information and generates a standard query vector based on the power grid fault information.

[0121] The standard query vector includes the faulty device, the fault type, and the sequence of protective actions taken against the faulty device after the fault occurred.

[0122] Step 204: The two-layer matching and evaluation engine matches the standard query vector with the disposal knowledge graph to obtain preliminary disposal steps, and performs safety simulation and risk assessment on the preliminary disposal steps based on the power grid model to obtain multiple disposal paths.

[0123] Step 205: The report generation module performs multi-objective optimization and ranking of the disposal path, and generates a comprehensive auxiliary decision-making report containing executable instructions and risk explanations; wherein, the executable instructions correspond to the disposal steps of the disposal path. The fault handling auxiliary decision-making method provided in this application embodiment can be understood by referring to the relevant content in the foregoing system embodiment section, and will not be repeated here.

[0124] like Figure 3 As shown, Figure 3 This is a schematic diagram of a possible logical structure of a computing device provided in an embodiment of this application. The computing device 300 includes a processor 301, a communication interface 302, a memory 303, and a bus 304. The processor 301, the communication interface 302, and the memory 303 are interconnected via the bus 304. In an embodiment of this application, the processor 301 is used to control and manage the operation of the computing device 300. For example, the processor 301 is used to execute... Figure 2 The steps in the embodiments and / or other processes used in the techniques described herein. Communication interface 302 is used to support communication by computing device 300. Memory 303 is used to store program code and data of computing device 300.

[0125] The processor 301 can be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. The processor can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a digital signal processor and a microprocessor, etc. The bus 304 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0126] In another embodiment of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including instructions that, when executed on a computer, cause the computer to perform the above-described... Figure 2 The method described in the embodiments.

[0127] Those skilled in the art will recognize that the units of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of the invention.

[0128] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0129] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0130] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0131] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0132] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0133] 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 them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A fault handling auxiliary decision-making system, characterized in that, The system includes a knowledge base construction module, a power grid status perception module, a fault scenario formatting module, a two-layer matching and evaluation engine, and a report generation module; the two-layer matching and evaluation engine is connected to the knowledge base construction module, the power grid status perception module, the fault scenario formatting module, and the report generation module, respectively. The knowledge base construction module is used to acquire power system regulations, fault handling plans and historical cases, and to construct a structured handling knowledge graph based on the power system regulations, the fault handling plans and the historical cases; The power grid status sensing module is used to acquire electrical data of the power grid before and after a fault, and to generate a power grid model based on the electrical data; The fault scenario formatting module is used to receive power grid fault information and generate a standard query vector based on the power grid fault information; wherein, the standard query vector includes the faulty equipment, the fault type, and the sequence of protection actions taken for the faulty equipment after the fault. The two-layer matching and evaluation engine is used to match the standard query vector with the disposal knowledge graph to obtain preliminary disposal steps, and to perform safety simulation and risk assessment on the preliminary disposal steps based on the power grid model to obtain multiple disposal paths; The report generation module is used to perform multi-objective optimization and sorting of the disposal paths and generate a comprehensive auxiliary decision-making report containing executable instructions and risk explanations; wherein the executable instructions correspond to the disposal steps of the disposal paths.

2. The system according to claim 1, characterized in that, The knowledge base construction module is specifically used for: Entities and corresponding relationships in the power field are extracted from the text data of the power system regulations, the fault handling plan, and the historical cases. The relationships include at least conditions, actions, and results. The relationships with conditions are used to connect entities with handling actions and entities with preconditions. The relationships with actions are used to connect entities with fault scenarios and entities with handling actions. The relationships that lead to results are used to connect entities with handling actions and entities with expected results and expected risks. Extract condition-action pairs from the text data to describe the handling rules; Using the entity as the first node and the relationship as the first side, a concept layer is constructed; using the condition-action pair as the second node and the logical relationship between the second nodes as the second side, a rule layer is constructed. Using the historical cases as independent case subgraphs and the similarity relationships between the case subgraphs as third edges, a case layer is constructed. The case subgraphs include scenario nodes for describing faults, response nodes for describing protection actions or handling actions, and status nodes for recording the state of the power grid before and after the action. A fourth edge is constructed based on the semantic relationship between the first node and the second node, and a fifth edge is constructed based on the similarity relationship between the case subgraph and the second node, thus obtaining the disposal knowledge graph.

3. The system according to claim 1, characterized in that, The power grid status sensing module is specifically used for: Upon receiving a fault event notification, a time window is set based on the fault time, including a pre-fault time window, a post-fault transient time window, and a post-fault steady-state time window. The first multi-source electrical data within the pre-fault time window, the second multi-source electrical data within the post-fault transient time window, and the third multi-source electrical data within the post-fault steady-state time window are collected from the energy management system, the wide-area measurement system, and the data acquisition and monitoring system. A basic model is constructed based on the aforementioned third multi-source electrical data; Based on the first multi-source electrical data and the second multi-source electrical data, the basic model is dynamically corrected to obtain the power grid model.

4. The system according to claim 1, characterized in that, The fault scenario formatting module is specifically used for: The fault information is parsed using a parser corresponding to the data source type of the fault information. Based on a pre-defined ontology framework for power fault domain, semantic role labeling is performed on the parsing results to obtain the faulty equipment, the fault type, the protection action sequence, and the events that occur during the fault handling process. Construct causal-temporal chains among the events; The standard query vector is constructed based on the faulty device, the fault type, the protection action sequence, the event, and the causal-temporal chain.

5. The system according to claim 2, characterized in that, The dual-layer matching and evaluation engine includes a connected knowledge matching layer and a real-time inference and evaluation layer; The knowledge matching layer is used to calculate the similarity between the standard query vector and the conditions in the second node; Based on the similarity, the average matching degree between the second node and the associated historical cases, and the confidence degree of the second node, the activation intensity of the second node is calculated; A set of preliminary treatment steps is generated based on the activation intensity; The real-time simulation and evaluation layer is used to perform sequential simulation of the steps in the set of preliminary handling steps based on the power grid model. After each simulation step, the power flow calculation is called to verify the safety constraints of equipment overload and voltage over-limit, and the comprehensive risk value after the step is executed is calculated. Based on the comprehensive risk value assessment, multiple disposal paths are determined for the power grid model from the post-fault state to the target recovery state.

6. The system according to claim 5, characterized in that, The formula for calculating the comprehensive risk value includes: ； in, This indicates the degree to which the m-th type of safety constraint is exceeded. This is the penalty coefficient for the corresponding constraint.

7. The system according to any one of claims 1-6, characterized in that, The report generation module is specifically used for: The system receives multiple processing paths and corresponding evaluation indicators output by the dual-layer matching and evaluation engine. The evaluation indicators include at least the total path risk, total operation time, and expected power restoration. The weights of each evaluation index are adjusted based on the urgency of the current power grid fault, and the disposal paths are sorted using the approximation of ideal solution sorting method to obtain the Pareto optimal solution set. Based on the Pareto optimal solution set, the comprehensive decision support report is generated. The report includes an executive summary for decision-makers, actionable recommendations for operators, and a technical detail including risk attribution analysis and rule tracing.

8. A fault handling auxiliary decision-making method, characterized in that, Applied to the system according to any one of claims 1-7, the method comprises: The knowledge base construction module acquires power system regulations, fault handling plans, and historical cases, and constructs a structured handling knowledge graph based on the power system regulations, fault handling plans, and historical cases; The power grid status sensing module acquires electrical data of the power grid before and after a fault, and generates a power grid model based on the electrical data; The fault scenario formatting module receives power grid fault information and generates a standard query vector based on the power grid fault information; wherein, the standard query vector includes the faulty equipment, the fault type, and the sequence of protection actions taken for the faulty equipment after the fault. The two-layer matching and evaluation engine matches the standard query vector with the disposal knowledge graph to obtain preliminary disposal steps, and performs safety simulation and risk assessment on the preliminary disposal steps based on the power grid model to obtain multiple disposal paths; The report generation module performs multi-objective optimization and sorting on the disposal path and generates a comprehensive auxiliary decision-making report containing executable instructions and risk explanations; wherein the executable instructions correspond to the disposal steps of the disposal path.

9. A computing device, characterized in that, include: Memory, used to store programs; A processor for loading the program to execute the method as described in claim 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method of claim 8.

Images