A Method and System for Automatic Generation of Power Grid Dispatch Operation Tickets Based on Multi-Agent Collaboration

By analyzing the multi-agent collaborative system and extrapolating real-time power grid data, combined with a dual verification mechanism, the problems of low efficiency and insufficient security in the preparation of distribution network dispatch operation tickets have been solved, and efficient and safe operation ticket generation and execution have been achieved.

CN123117580APending Publication Date: 2026-11-17STATE GRID SHANDONG ELECTRIC POWER CO LIAOCHENG POWER SUPPLY CO
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611013835.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-08
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In existing technologies, the compilation efficiency of distribution network dispatch operation tickets is low, and the generated tickets need to be manually reviewed item by item, which makes it difficult to adapt to the dynamic changes in power grid operation mode and lacks effective automated verification methods, affecting dispatch response speed and security.

Method used

A multi-agent collaborative approach is adopted, which analyzes maintenance application tickets through a multi-agent collaborative system, performs operational logic deduction by combining real-time power grid data, and implements a dual verification mechanism to generate and synchronize the data to the distribution automation system, thereby realizing a closed loop of the entire process from application ticket to execution.

Benefits of technology

It improves the efficiency of operation ticket preparation, reduces the error rate, ensures the safety and accuracy of operation tickets, reduces manual intervention, and adapts to the dynamic changes in power grid operation modes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117580A_ABST
    Figure CN123117580A_ABST
Patent Text Reader

Abstract

This invention discloses an automatic generation method and system for power grid dispatch operation tickets based on multi-agent collaboration, belonging to the field of power dispatch automation. The method includes: S1, acquiring distribution network maintenance application ticket data, parsing it using a multi-agent collaborative system including scenario judgment, ticket element extraction, and equipment naming, and outputting structured operation information; S2, acquiring real-time power grid topology and equipment operating status data through the MCP protocol, combining the operation information with logical deduction to generate an initial dispatch operation ticket; S3, performing step integrity verification at the whole ticket level and real-time equipment status matching verification at the step level on the initial dispatch operation ticket to generate a target dispatch operation ticket; S4, visually displaying the target dispatch operation ticket and synchronizing it to the distribution automation system for execution. This achieves intelligent generation and multiple security checks of dispatch operation tickets, significantly improving ticket writing efficiency and operation ticket accuracy.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of power grid dispatch automation technology, specifically to a method and system for automatically generating power grid dispatch operation tickets based on multi-agent collaboration. Background Technology

[0002] Dispatch operation tickets are key technical documents used to guide switching operations in power grid dispatching and operation. The accuracy and efficiency of their preparation are directly related to the safety and reliability of power grid operation.

[0003] Currently, various automated methods have emerged for generating operation tickets for distribution network dispatching, such as instantiation and filling based on typical ticket templates and automatic ticket generation based on rule matching. However, in actual dispatching operations, a large number of operation tickets still require manual generation by dispatchers or item-by-item review and correction of automatically generated tickets. Dispatchers need to manually draft or verify the operation steps based on information such as the outage scope, method of operation, and safety measures in the maintenance request ticket, combined with the power grid wiring diagram and equipment status. However, the generation or verification of operation tickets for complex maintenance tasks often takes a long time. During peak maintenance periods, the workload of operation tickets surges, affecting the dispatching response speed.

[0004] Existing technologies include auxiliary ticket generation methods based on typical ticket templates or rule matching. For example, these methods instantiate and populate typical operation ticket templates, or extract information from maintenance requests based on word segmentation and semantic analysis and then match it with a rule base to generate operation tickets. However, these methods mainly rely on static rules and preset templates, making it difficult to adapt to dynamic changes in power grid operation modes. Furthermore, existing solutions primarily focus on the operation ticket generation stage, lacking effective automated verification methods for the generated operation tickets. The generated tickets still require manual review item by item, and the safety control still depends on human intervention.

[0005] Therefore, how to improve the efficiency of scheduling operation ticket preparation while ensuring the safety and accuracy of the operation tickets is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] To address the aforementioned issues, this invention provides a method and system for automatically generating power grid dispatch operation tickets based on multi-agent collaboration. The method utilizes a multi-agent collaborative system to perform structured parsing of maintenance application tickets; acquires topology and status data in real time through a power grid data service interface; performs operational logic deduction based on real-time data; employs a dual verification mechanism combining whole-ticket and step-by-step verification; and generates executable operation tickets and synchronizes them to the distribution automation system. This shortens the operation ticket preparation time, reduces the error rate through dual verification, and achieves a closed-loop process from application ticket to operation ticket to execution.

[0007] In a first aspect, the present invention provides a method for automatically generating power grid dispatch operation tickets based on multi-agent cooperation, comprising the following steps: S1. Obtain distribution network maintenance application ticket data, and use a pre-trained multi-agent collaborative system to perform parallel parsing of the maintenance application ticket data to output structured operation information; wherein, the multi-agent collaborative system includes a sub-agent for determining the scenario, a sub-agent for extracting ticket elements, and a sub-agent for splitting equipment names; S2. Call the power grid data service interface to obtain real-time power grid topology data and equipment operation status data; combine the structured operation information to perform operation logic deduction and generate an initial scheduling operation ticket containing operation tasks, operation steps, safety measures and equipment numbers; wherein, the operation information includes at least the operation scenario type, power outage and restoration range, safety measures and standardized equipment names; S3. Perform dual compliance checks on the initial scheduling operation ticket to generate a target scheduling operation ticket that passes the checks; the dual compliance checks include the operation logic and specification compliance check at the whole ticket level, and the matching check between the operation instructions and the real-time status of the equipment at the step level. S4. Visualize the target dispatch operation ticket and, in response to the dispatcher's confirmation instruction or automatic execution instruction, synchronize it to the power distribution automation system through the interface to execute the operation.

[0008] As a preferred embodiment of the technical solution of the present invention, in step S1: The sub-agent used to determine the scenario is specifically used to: parse the work content and power outage range in the maintenance application ticket data, automatically determine the maintenance scenario type and the attributes of the power outage ticket or power restoration ticket; the maintenance scenario type includes mode adjustment, in-station equipment maintenance, new equipment commissioning, live-line work or opening / closing a single switch; The sub-agent used to extract ticket elements is specifically used to: automatically split and extract the applicant unit, power outage and restoration range, mode arrangement, protection arrangement and on-site safety measures from the maintenance application ticket data, and complete the structured sorting of information; The sub-agent used for splitting equipment naming is specifically used to: standardize and identify the equipment description in the maintenance application data according to the naming and numbering specifications of distribution network equipment, and accurately identify and split the voltage level, equipment type and full name of the equipment; the equipment type includes switches, disconnectors, lines or ring network boxes.

[0009] The sub-intelligent agent for extracting ticket elements can automatically extract core information such as the applicant unit, power outage and restoration range, mode arrangement, protection arrangement, and on-site safety measures, realizing the structured sorting of maintenance application information and avoiding omissions and errors during manual extraction; the sub-intelligent agent for splitting equipment naming can automatically identify voltage level, equipment type, and full name of equipment according to the naming standard of distribution network equipment, avoiding equipment name identification deviation and writing errors from the source.

[0010] As a preferred embodiment of the technical solution of the present invention, the power grid data service interface is invoked to obtain real-time power grid topology data and equipment operating status data, including: The topology and equipment operating status data of the distribution network are obtained through a power grid data call interface based on a preset communication protocol; the preset communication protocol includes Model Context Protocol (MCP), IEC 104 protocol, or IEC 61968 standard interface. The real-time power grid topology data includes the equipment attributes of circuit breakers, load switches, disconnect switches, fuses, and ring main units, as well as the electrical connection relationships between equipment nodes; the equipment operation status data includes at least one of the following: the opening and closing positions of switchgear, the status of grounding switches, telemetry values ​​of line voltage and current, and protection action signals.

[0011] As a preferred embodiment of the technical solution of the present invention, operational logic deduction is performed in conjunction with the structured operation information, including: Based on the real-time power grid topology data, the upstream and downstream relationships and power flow paths of the equipment to be operated are determined; Based on the equipment operating status data, a pre-verification of the current power grid operating environment is performed to determine whether the preconditions for operation are met. Based on the distribution network dispatching operation procedures and the five-prevention rules, the execution sequence of operation steps is dynamically planned to generate an initial dispatching operation ticket containing the operation object, operation action and operation objective.

[0012] Based on real-time power grid topology data, the upstream and downstream relationships and power flow paths of the equipment to be operated are determined to ensure that the operation steps conform to the actual electrical connection relationship of the power grid and avoid operation sequence errors caused by inaccurate topology information. Pre-verification of operation preconditions is performed in conjunction with equipment operating status data to determine whether the current power grid environment meets the operation conditions before execution, and to detect and prevent unsafe operations in advance. The execution sequence of operation steps is dynamically planned according to the distribution network dispatching operation procedures and the five-prevention rules, so that the generated operation tickets not only comply with industry standards but also adapt to specific scenarios, thereby improving the compliance and executability of the operation tickets.

[0013] As a preferred embodiment of the technical solution of this invention, the compliance verification of the operational logic and specifications at the whole ticket level includes: The pre-set scheduling operation procedure library and typical ticket logic library are invoked to perform a full-page, item-by-item scan of the initial scheduling operation ticket; Identify and verify at least one of the following non-compliant situations: missing key operation steps, reversed logical order of operation steps, non-standard equipment naming, or omission of safety measures. If the verification is successful, the initial scheduling operation ticket is confirmed to enter the step-by-step verification stage; if there are non-compliant situations, a full ticket verification report containing the location of missing points and correction suggestions is generated, and the initial scheduling operation ticket is automatically corrected or a manual intervention instruction is generated based on the correction suggestions.

[0014] By calling the pre-set scheduling operation procedure library and typical ticket logic library to perform a full-ticket item-by-item scan, it can systematically discover non-compliant situations such as missing key operation steps, reversed operation order, non-standard equipment naming, or omission of safety measures. After the verification is passed, it automatically enters the step-by-step verification, forming an organic connection between the full ticket verification and the step-by-step verification, and building a complete double verification defense line. For operation tickets with non-compliant situations, it can generate a verification report that includes the location of missing points and correction suggestions, and supports automatic correction or manual intervention, which improves the efficiency of error correction.

[0015] As a preferred embodiment of the technical solution of the present invention, the matching verification between the step-by-step operation instructions and the real-time status of the device includes: For each operation step in the initial dispatch operation ticket, the remote signaling status and telemetry values ​​of the corresponding equipment in the power distribution automation system are retrieved in real time. Based on the five-prevention rules and power grid topology constraints, it is verified whether the initial state of the equipment before the operation meets the preconditions for operation, and whether the target state of the equipment after the operation will cause the power grid to exceed the limit or the topology to be abnormal. If the verification passes, the step is marked as executable; if the verification fails, a state mismatch alarm is generated, the operation step is marked as prohibited, and a verification failure message containing device state correction instructions is output.

[0016] For each operation step, the system retrieves the remote signaling status and telemetry values ​​of the corresponding equipment in real time, achieving precise step-by-step matching between operation commands and actual equipment status, thus avoiding operational risks caused by delayed status information. Based on the five-prevention rules and power grid topology constraints, the system performs dual verification on the initial status of the equipment before operation and the target status of the equipment after operation, covering the two key nodes before and after operation execution. If the verification passes, the step is marked as executable; if it fails, a status mismatch alarm is generated, the step is marked as prohibited from execution, and correction guidance is output, providing dispatchers with clear operation guidance and safety assurance.

[0017] As a preferred embodiment of the technical solution of the present invention, after synchronizing to the power distribution automation system via the interface in step S4, a closed-loop execution monitoring step is further included: In response to the received operation execution instruction, the operation instructions of the target scheduling operation ticket are sent to the distribution automation system in a step-by-step sequence; The system monitors the equipment status signals fed back by the power distribution automation system in real time and compares the equipment status signals with the target status of the current operation step. If the comparison matches, the step is automatically marked as completed and the next operation instruction is triggered. If the comparison is inconsistent, an execution exception alarm will be generated and the execution of subsequent operation instructions will be suspended. After all operation steps are completed, an operation ticket execution report is generated, which includes execution records, execution time, and exception information.

[0018] The operation instructions of the target scheduling operation ticket are automatically sent to the power distribution automation system in a step-by-step sequence, realizing an automated flow from ticket generation to on-site execution and reducing manual forwarding links; real-time monitoring of equipment status feedback signals and comparison with the target status of the operation steps can promptly detect execution deviations; when the comparison is consistent, it is automatically marked as completed and triggers the issuance of the next instruction; when the comparison is inconsistent, an abnormal alarm is generated and subsequent operations are suspended, effectively preventing the risk of cascading accidents caused by single-step operation failures.

[0019] As a preferred embodiment of the technical solution of the present invention, a feedback optimization step is also included: Collect historical modification records of the generated initial or target scheduling operation tickets by the dispatcher. The historical modification records include the addition or deletion of steps, correction of equipment names, adjustment of operation order, or supplementation of safety measures. Semantic difference analysis was performed on historical modification records to extract correction feature vectors and construct an optimized sample library of operation tickets. When the amount of data in the operation ticket optimization sample library reaches a preset threshold, the multi-agent collaborative system is iteratively updated using the corrected feature vector; the iterative update includes: incrementally training the model parameters of the sub-agents based on the corrected feature vector, and updating the prompt word templates or parsing strategy weights of the sub-agents.

[0020] By collecting historical modification records of dispatchers' generated operation tickets, the dispatchers' professional experience is transformed into training samples that can be learned, realizing the explicit utilization of tacit knowledge. Semantic difference analysis of the modification records and extraction of correction feature vectors can accurately identify the difference patterns between the original generated results and the corrected results, providing targeted guidance for model optimization. When the sample library reaches a preset threshold, incremental training of the model and updating of prompt word templates are triggered, enabling the multi-agent system to continuously adapt to changes in dispatchers' operating habits and standards, gradually reducing the rate of manual correction and realizing the system's self-evolution capability.

[0021] As a preferred embodiment of the technical solution of the present invention, before calling the power grid data service interface, a graph data preprocessing step is further included: Acquire source-side diagram data of the power distribution automation system, wherein the source-side diagram data includes SVG graphic files and XML configuration files; Perform topology parsing on the source-end graph data to extract the electrical connection relationships, switch attributes, and interconnection relationships of the device nodes; The parsed graph data is mapped and aligned with the public information model of the control cloud platform to generate a standardized power grid topology data model, which is then stored in the cache area of ​​the power grid data service interface.

[0022] By parsing the SVG graphic files and XML configuration files of the power distribution automation system before data retrieval, the electrical connection relationships, switch attributes, and communication relationships of the device nodes can be extracted in advance, avoiding parsing delays during real-time retrieval. The parsed graphic model data is then mapped and aligned with the public information model of the control cloud platform, resolving the inconsistencies in device naming, switch attributes, and communication relationships between the power distribution automation system and the control cloud platform.

[0023] Secondly, the technical solution of the present invention also provides an automatic generation system for power grid dispatching operation tickets based on multi-agent cooperation, comprising: The data acquisition module is used to acquire distribution network maintenance application ticket data; The multi-agent collaborative module includes a pre-trained sub-agent for judging the scenario, a sub-agent for extracting ticket elements, and a sub-agent for splitting equipment names. It is used to perform parallel parsing of the maintenance application ticket data and output structured operation information. The data service interface module is used to call the power grid data service interface to obtain real-time power grid topology data and equipment operating status data; The operation logic deduction module is used to perform operation logic deduction based on the structured operation information and generate an initial scheduling operation ticket containing operation tasks, operation steps, safety measures and equipment numbers. The dual compliance verification module is used to perform dual compliance verification on the initial scheduling operation ticket and generate a target scheduling operation ticket that passes the verification. The dual compliance verification includes the operation logic and specification compliance verification at the whole ticket level, and the matching verification of operation instructions and real-time equipment status at the step level. The output and execution module is used to visualize the target scheduling operation ticket and, in response to the dispatcher's confirmation instruction or automatic execution instruction, synchronize it to the power distribution automation system through an interface to execute the operation.

[0024] As can be seen from the above technical solutions, this application has the following advantages: Multi-agent collaborative parsing can process the tasks of scene judgment, element extraction and equipment splitting in the maintenance application ticket in parallel, which greatly shortens the information parsing time and improves the efficiency of writing the ticket; Operation logic deduction based on real-time power grid data can adapt to the dynamic changes in the power grid operation mode and avoid the adaptation deviation caused by static templates; Dual compliance verification is performed from the two levels of the whole ticket logic and step-by-step status, which effectively reduces the error rate of the operation ticket and improves the standardization and executability of the operation ticket; Through visualization and synchronous execution with the distribution automation system, a closed loop of the whole process from application ticket to operation ticket to execution is realized, reducing the manual intervention links. Attached Figure Description

[0025] To more clearly illustrate the technical solution of this application, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart illustrating the method provided in an embodiment of the present invention.

[0027] Figure 2 A block diagram of a system provided in an embodiment of the present invention. Detailed Implementation

[0028] To make the purpose, features, and advantages of this application more apparent and understandable, specific embodiments and accompanying drawings will be used to clearly and completely describe the technical solution protected by this application. Obviously, the embodiments described below are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0029] Unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this application and in the specification of this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0030] like Figure 1 As shown, this embodiment of the invention provides a method for automatically generating power grid dispatch operation tickets based on multi-agent cooperation, including the following steps: S1. Obtain distribution network maintenance application ticket data, and use a pre-trained multi-agent collaborative system to perform parallel parsing of the maintenance application ticket data to output structured operation information; wherein, the multi-agent collaborative system includes a sub-agent for determining the scenario, a sub-agent for extracting ticket elements, and a sub-agent for splitting equipment names; Specifically, the distribution network maintenance application ticket data originates from the OMS (Operation Management System) and includes at least information such as maintenance task content, power outage scope, applicant unit, mode arrangement, protection arrangement, and on-site safety measures. The multi-agent collaborative system comprises three sub-agents: a sub-agent for scenario determination, a sub-agent for extracting ticket elements, and a sub-agent for splitting equipment names. These three sub-agents collaborate based on a pre-built unified ticket drafting rule system and a distribution network scheduling professional knowledge base.

[0031] (1) Determine the specific implementation of the scene sub-agent: This sub-agent employs a text classification model based on the Transformer architecture. The input consists of the work content field and the power outage range field from the maintenance request ticket. The model first performs word segmentation and word embedding on the input text, then extracts semantic features through a multi-head self-attention mechanism, and finally outputs a maintenance scenario type label using a Softmax classifier. Maintenance scenario types include: mode adjustment, in-station equipment maintenance, new equipment commissioning, live-line work, or operating a single switch. Simultaneously, the sub-agent also outputs attribute identifiers for the power outage ticket or power restoration ticket.

[0032] (2) Specific implementation of the sub-agent for extracting ticket elements: This sub-agent employs a BiLSTM-CRF-based named entity recognition model. The model predefines seven entity categories: requesting unit, power outage / restoration range, mode of operation, protection arrangement, on-site safety measures, power outage start time, and power outage end time. The input is the complete text of the maintenance request form; the model labels the entity category of each character, ultimately extracting structured element information.

[0033] (3) Specific implementation of splitting device naming sub-intelligent agents: This sub-agent employs a hybrid approach combining regular expressions and rule matching. A device naming rule library is constructed based on relevant specifications, defining naming patterns for voltage level, device type, and device number. The input device description string is subjected to pattern matching, splitting it into three parts: voltage level, device type, and full device name. For any non-matching abnormal names, a manual review process is triggered.

[0034] The three sub-agents can execute in parallel or sequentially according to their dependencies (e.g., determining the scenario first and then extracting elements). This embodiment does not limit their execution order. After parsing, structured operation information is output, including at least: operation scenario type, power outage / restoration range, safety measures, and standardized equipment names.

[0035] In this embodiment of the invention, the multi-agent collaborative system needs to be pre-trained. The training data comes from historical maintenance application tickets and corresponding operation tickets approved by the dispatcher, collecting no less than 5,000 training samples. Each sample includes: maintenance application ticket text, manually labeled maintenance scenario type tags, manually labeled ticket elements, manually labeled equipment name splitting results, and manually approved operation ticket step sequence.

[0036] The scene-judging sub-agent employs a supervised classification training method, using the cross-entropy loss function to fine-tune the Transformer classification model. The ticket element extraction sub-agent uses sequence labeling training, employing maximum likelihood estimation to learn the parameters of the BiLSTM-CRF model. The equipment naming sub-agent constructs an initial rule base based on the power distribution network equipment naming conventions, and uses manually labeled samples for rule verification and correction.

[0037] After the above training is completed, the model parameters of each sub-agent are fixed and deployed on the control cloud platform for online use. Those skilled in the art should understand that the above training method can be implemented using supervised learning techniques known in the art, and the specific hyperparameter settings can be adjusted according to the actual training effect.

[0038] S2. Call the power grid data service interface to obtain real-time power grid topology data and equipment operation status data; combine the structured operation information to perform operation logic deduction and generate an initial scheduling operation ticket containing operation tasks, operation steps, safety measures and equipment numbers; wherein, the operation information includes at least the operation scenario type, power outage and restoration range, safety measures and standardized equipment names; This embodiment obtains data through the power grid data call interface based on the Model Context Protocol (MCP). The obtained data includes two categories: (1) Real-time power grid topology data: including the equipment attributes of circuit breakers, load switches, disconnect switches, fuses, and ring main units, as well as the electrical connection relationships between equipment nodes. The topology data is stored in a graph structure (node-edge model), where nodes represent electrical equipment and edges represent electrical connections between equipment. (2) Equipment operating status data: including at least one of the following: the opening and closing positions of switchgear, the status of grounding switches, the telemetry values ​​of line voltage and current, and protection action signals. This data is collected in real time through the data acquisition and monitoring control module of the distribution automation system, with a refresh frequency of seconds.

[0039] The operational logic deduction is performed by the parent agent and includes the following steps: Step A: Based on the real-time power grid topology data, determine the upstream and downstream relationships and power flow paths of the equipment to be operated. Specifically, the parent agent invokes a graph search algorithm to traverse its neighboring nodes, starting from the node of the equipment to be operated, and constructs a local topology subgraph centered on the equipment. In the subgraph, upstream power sources and downstream loads are identified to determine the power supply range and power flow direction that may be affected by operating the equipment. For example, for a line switch to be opened, its upstream is the power supply bus, and its downstream is the line and its load.

[0040] Step B: Based on the equipment operating status data, perform a pre-verification of the current power grid operating environment to determine whether the preconditions for operation are met.

[0041] The preconditions for operation refer to the safety conditions that relevant equipment in the power grid must meet before executing a certain operation command. Taking the closing of switch 101 on a 10kV line as an example, the preconditions include: (a) switch 101 is currently in the open state; (b) the grounding switches on both sides of switch 101 are in the open state; (c) there are no maintenance signs on the downstream line of switch 101; (d) during the loop closing operation, the voltage difference and phase difference on both sides of the loop closing point are within the allowable range. The parent agent checks the above conditions one by one by querying the equipment operating status data. If all conditions are met, the operation continues; otherwise, a pre-verification failure report is generated and the simulation is terminated.

[0042] Step C: Based on the distribution network dispatching operation procedures and the five-prevention rules, dynamically plan the execution sequence of the operation steps and generate the initial dispatching operation ticket.

[0043] The five-prevention rules are five functions in the power system to prevent misoperation, including: preventing accidental opening and closing of circuit breakers, preventing opening and closing of disconnecting switches under load, preventing grounding wires from being connected while energized, preventing closing circuit breakers / disconnecting switches with grounding wires connected, and preventing accidental entry into energized compartments. The parent intelligent agent inputs the sequence of operations to be executed into the rule engine. The engine rearranges and verifies the operation sequence according to the five-prevention rules to ensure compliance with the anti-misoperation logic. For example, the circuit breaker must be opened before the disconnecting switch can be opened; the grounding switch must be closed before maintenance work can be performed. The final generated initial dispatch operation ticket contains complete elements such as the operation task, operation steps, safety measures, and equipment number.

[0044] S3. Perform dual compliance checks on the initial scheduling operation ticket to generate a target scheduling operation ticket that passes the checks; the dual compliance checks include the operation logic and specification compliance check at the whole ticket level, and the matching check between the operation instructions and the real-time status of the equipment at the step level. The overall ticket level verification is performed by the overall ticket verification submodule, including: The system invokes a pre-built library of dispatching operation procedures and a library of typical ticket logic to perform a full, item-by-item scan of the initial dispatching operation ticket. The dispatching operation procedure library stores the structured rules of normative documents related to power dispatching; the typical ticket logic library stores the logical patterns of verified historical high-quality operation tickets.

[0045] Identify and verify whether the following non-compliance situations exist: missing key operation steps, reversed logical order of operation steps, non-standard equipment naming, and omission of safety measures.

[0046] If the verification passes, the initial dispatch operation ticket is confirmed to enter the step-by-step verification stage; if non-compliance is found, a full ticket verification report is generated, which includes the specific location of missing points and correction suggestions. The system can automatically correct the operation ticket based on the correction suggestions, or generate manual intervention instructions for the dispatcher to handle.

[0047] Step-by-step verification is performed by the step-by-step verification submodule, including: For each operation step in the initial dispatch operation ticket, the remote signaling status and telemetry values ​​of the corresponding equipment in the distribution automation system are retrieved in real time. Based on the five-prevention rules and power grid topology constraints, two types of verification are performed: Verification A: Check whether the initial state of the equipment meets the preconditions for operation. For example, before closing the circuit breaker, confirm that the switch is currently in the open state and that no grounding switch is closed.

[0048] Verification B: Whether the target state of the equipment after the operation will lead to grid operation exceeding limits or topology anomalies. For example, after a closing operation, the system predicts the current distribution after closing based on topology analysis to determine whether it will lead to line overload or unplanned loops.

[0049] If both checks pass, the operation step is marked as executable; if either check fails, a state mismatch alarm is generated, the step is marked as prohibited, and a check failure message containing equipment state correction instructions is output, such as switch 101 is currently in the closed state and cannot be closed, please perform the opening operation first.

[0050] S4. Visualize the target dispatch operation ticket and, in response to the dispatcher's confirmation instruction or automatic execution instruction, synchronize it to the power distribution automation system through the interface to execute the operation.

[0051] Specifically, the target dispatch operation ticket is rendered in tabular form on the human-machine interface, with each row displaying an operation step and a status indicator. The dispatcher can confirm execution step by step or trigger an automatic execution command. After receiving the automatic execution command, the system sends the operation instructions to the distribution automation system or the distribution network dispatch and command network interaction system through the interface in the order of steps.

[0052] In some embodiments, after the operation command is issued, the system performs closed-loop execution monitoring steps: In response to the received operation execution command, the operation command of the target scheduling operation ticket is sent to the distribution automation system in a step-by-step sequence; the equipment status signal fed back by the distribution automation system is monitored in real time, and the equipment status signal is compared with the target status of the current operation step; if the comparison is consistent, the step is automatically marked as completed and the issuance of the next operation command is triggered; if the comparison is inconsistent, an execution abnormality alarm is generated and the execution of subsequent operation commands is suspended. After all operation steps are completed, an operation ticket execution report is generated, which includes execution records, execution time, and exception information.

[0053] In some embodiments, a feedback optimization step is also included: Collect historical modification records of the generated initial or target scheduling operation tickets by the dispatcher. The historical modification records include the addition or deletion of steps, correction of equipment names, adjustment of operation order, or supplementation of safety measures. Semantic difference analysis is performed on historical modification records to extract corrected feature vectors, and an optimized sample library of operation tickets is constructed. Specifically, a sequence alignment algorithm is used to compare the original generated operation tickets with the corrected operation tickets, extracting the difference patterns between the two to form corrected feature vectors. The corrected feature vectors include: difference location, difference type, and semantic category of the difference content. An optimized sample library of operation tickets is constructed. Each modification record and its corresponding corrected feature vector constitute a training sample, which is stored in the sample library.

[0054] When the amount of data in the operation ticket optimization sample library reaches a preset threshold, the multi-agent collaborative system is iteratively updated using the corrected feature vector. The update method includes: Incremental training of the sub-agent's model parameters is performed based on the corrected feature vectors. For example, the classification model of the sub-agent for judging the scene is incrementally trained using corrected samples, and the model weights are adjusted.

[0055] Update the prompt word template or parsing strategy weights for the sub-agent. For example, write the high-frequency correction pattern into the prompt word template so that this step is automatically added in subsequent generation processes.

[0056] In some embodiments, a graph data preprocessing step is included before invoking the power grid data service interface: Acquire source-side diagram data from the power distribution automation system. This source-side diagram data includes SVG graphic files for storing graphical information about the power grid wiring diagram and XML configuration files for storing structured information such as device attributes and connection relationships.

[0057] The source graph data undergoes topology parsing. The parser reads the device node definitions and connection line definitions from the XML configuration file and constructs an adjacency table of device-connection relationships. Simultaneously, it extracts the attributes of each device, including device ID, device name, device type, affiliated substation, voltage level, switch attributes, and connection relationships.

[0058] The parsed graphical model data is mapped and aligned with the Common Information Model (CIM) of the control cloud platform. Because there are differences in device naming and attribute definitions between the distribution automation system and the control cloud platform (e.g., the same device may have different IDs in the two systems), this step establishes a cross-system device mapping table, converting the parsed graphical model data into a data format conforming to the CIM standard.

[0059] A standardized power grid topology data model is generated and stored in the cache area of ​​the power grid data service interface. This cache area uses a high-speed caching database such as Redis for real-time access in step S2, thereby reducing data parsing latency.

[0060] Corresponding to the above method embodiments, such as Figure 2 As shown, this embodiment also provides an automatic generation system for power grid dispatching operation tickets based on multi-agent cooperation, including the following modules: Data acquisition module: Used to acquire data on distribution network maintenance application tickets.

[0061] The multi-agent collaborative module includes pre-trained sub-agents for judging the scenario, extracting ticket elements, and splitting equipment names. These sub-agents are used to parse the maintenance application ticket data in parallel and output structured work information. The specific implementation of each sub-agent is the same as described in the above method embodiments and will not be repeated here.

[0062] Data Service Interface Module: Used to call the power grid data service interface to obtain real-time power grid topology data and equipment operating status data. This module supports multiple communication protocols such as MCP, IEC 104, and IEC 61968.

[0063] Operation logic deduction module: used to perform operation logic deduction based on the structured operation information, and generate an initial scheduling operation ticket containing operation tasks, operation steps, safety measures and equipment numbers.

[0064] Dual compliance verification module: This module performs dual compliance verification on the initial scheduling operation ticket to generate a target scheduling operation ticket that passes the verification. This module further includes a whole-ticket verification submodule and a step-by-step verification submodule.

[0065] Output and Execution Module: This module visualizes the target scheduling operation ticket and responds to the dispatcher's confirmation or automatic execution instructions by synchronizing the operation to the power distribution automation system via an interface.

[0066] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for automatically generating power grid dispatch operation tickets based on multi-agent cooperation, characterized in that, Includes the following steps: S1. Obtain distribution network maintenance application ticket data, and use a pre-trained multi-agent collaborative system to perform parallel parsing of the maintenance application ticket data to output structured operation information; wherein, the multi-agent collaborative system includes a sub-agent for judging the scenario, a sub-agent for extracting ticket elements, and a sub-agent for splitting equipment naming. S2. Call the power grid data service interface to obtain real-time power grid topology data and equipment operation status data; combine the structured operation information to perform operation logic deduction and generate an initial scheduling operation ticket containing operation tasks, operation steps, safety measures and equipment numbers; wherein, the operation information includes at least the operation scenario type, power outage and restoration range, safety measures and standardized equipment names; S3. Perform dual compliance checks on the initial scheduling operation ticket to generate a target scheduling operation ticket that passes the checks; the dual compliance checks include the operation logic and specification compliance check at the whole ticket level, and the matching check between the operation instructions and the real-time status of the equipment at the step level. S4. Visualize the target dispatch operation ticket and, in response to the dispatcher's confirmation instruction or automatic execution instruction, synchronize it to the power distribution automation system through the interface to execute the operation.

2. The method for automatically generating power grid dispatch operation tickets based on multi-agent collaboration according to claim 1, characterized in that, In S1: The sub-agent used to determine the scenario is specifically used to: parse the work content and power outage range in the maintenance application ticket data, automatically determine the maintenance scenario type and the attributes of the power outage ticket or power restoration ticket; the maintenance scenario type includes mode adjustment, in-station equipment maintenance, new equipment commissioning, live-line work or opening / closing a single switch; The sub-agent used to extract ticket elements is specifically used to: automatically split and extract the applicant unit, power outage and restoration range, mode arrangement, protection arrangement and on-site safety measures from the maintenance application ticket data, and complete the structured sorting of information; The sub-agent used for splitting equipment naming is specifically used to: standardize and identify the equipment description in the maintenance application data according to the naming and numbering specifications of distribution network equipment, and accurately identify and split the voltage level, equipment type and full name of the equipment; the equipment type includes switches, disconnectors, lines or ring network boxes.

3. The method for automatically generating power grid dispatch operation tickets based on multi-agent collaboration according to claim 2, characterized in that, Call the power grid data service interface to obtain real-time power grid topology data and equipment operating status data, including: The topology connection relationship and equipment operation status data of the distribution network are obtained through a power grid data call interface based on a preset communication protocol. The real-time power grid topology data includes the equipment attributes of circuit breakers, load switches, disconnect switches, fuses and ring main units, as well as the electrical connection relationship between equipment nodes. The equipment operation status data includes at least one of the following: the opening and closing position of switchgear, the status of grounding switch, the telemetry values ​​of line voltage and current, and protection action signals.

4. The method for automatically generating power grid dispatch operation tickets based on multi-agent cooperation according to claim 3, characterized in that, The operational logic deduction is performed based on the structured job information, including: Based on the real-time power grid topology data, the upstream and downstream relationships and power flow paths of the equipment to be operated are determined; Based on the equipment operating status data, a pre-verification of the current power grid operating environment is performed to determine whether the preconditions for operation are met. Based on the distribution network dispatching operation procedures and the five-prevention rules, the execution sequence of operation steps is dynamically planned to generate an initial dispatching operation ticket containing the operation object, operation action and operation objective.

5. The method for automatically generating power grid dispatch operation tickets based on multi-agent cooperation according to claim 4, characterized in that, Verification of the operational logic and compliance with regulations at the whole-ticket level, including: The pre-set scheduling operation procedure library and typical ticket logic library are invoked to perform a full-page, item-by-item scan of the initial scheduling operation ticket; Identify and verify at least one of the following non-compliant situations: missing key operation steps, reversed logical order of operation steps, non-standard equipment naming, or omission of safety measures. If the verification is successful, the initial scheduling operation ticket is confirmed to enter the step-by-step verification stage; if there are non-compliant situations, a full ticket verification report containing the location of missing points and correction suggestions is generated, and the initial scheduling operation ticket is automatically corrected or a manual intervention instruction is generated based on the correction suggestions.

6. The method for automatically generating power grid dispatch operation tickets based on multi-agent cooperation according to claim 5, characterized in that, The verification of the match between step-by-step operation instructions and the real-time status of the device includes: For each operation step in the initial dispatch operation ticket, the remote signaling status and telemetry values ​​of the corresponding equipment in the power distribution automation system are retrieved in real time. Based on the five-prevention rules and power grid topology constraints, it is verified whether the initial state of the equipment before the operation meets the preconditions for operation, and whether the target state of the equipment after the operation will cause the power grid to exceed the limit or the topology to be abnormal. If the verification passes, the step is marked as executable; if the verification fails, a state mismatch alarm is generated, the operation step is marked as prohibited, and a verification failure message containing device state correction instructions is output.

7. The method for automatically generating power grid dispatch operation tickets based on multi-agent cooperation according to claim 6, characterized in that, After S4 is synchronized to the distribution automation system via the interface, it also includes a closed-loop execution monitoring step: In response to the received operation execution instruction, the operation instructions of the target scheduling operation ticket are sent to the distribution automation system in a step-by-step sequence; The system monitors the equipment status signals fed back by the power distribution automation system in real time and compares the equipment status signals with the target status of the current operation step. If the comparison matches, the step is automatically marked as completed and the next operation instruction is triggered. If the comparison is inconsistent, an execution exception alarm will be generated and the execution of subsequent operation instructions will be suspended. After all operation steps are completed, an operation ticket execution report is generated, which includes execution records, execution time, and exception information.

8. The method for automatically generating power grid dispatch operation tickets based on multi-agent cooperation according to claim 7, characterized in that, It also includes feedback optimization steps: Collect historical modification records of the generated initial or target scheduling operation tickets by the dispatcher. The historical modification records include the addition or deletion of steps, correction of equipment names, adjustment of operation order, or supplementation of safety measures. Semantic difference analysis was performed on historical modification records to extract correction feature vectors and construct an optimized sample library of operation tickets. When the amount of data in the operation ticket optimization sample library reaches a preset threshold, the multi-agent collaborative system is iteratively updated using the corrected feature vector; The iterative update includes: incrementally training the model parameters of the sub-agent based on the corrected feature vector, and updating the prompt word template or parsing strategy weights of the sub-agent.

9. The method for automatically generating power grid dispatch operation tickets based on multi-agent cooperation according to claim 8, characterized in that, Before calling the power grid data service interface, a preprocessing step for the graphic model data is also included: Acquire source-side diagram data of the power distribution automation system, wherein the source-side diagram data includes SVG graphic files and XML configuration files; Perform topology parsing on the source-end graph data to extract the electrical connection relationships, switch attributes, and interconnection relationships of the device nodes; The parsed graph data is mapped and aligned with the public information model of the control cloud platform to generate a standardized power grid topology data model, which is then stored in the cache area of ​​the power grid data service interface.

10. An automatic generation system for power grid dispatching operation tickets based on multi-agent cooperation, characterized in that, include: The data acquisition module is used to acquire distribution network maintenance application ticket data; The multi-agent collaborative module includes a pre-trained sub-agent for judging the scenario, a sub-agent for extracting ticket elements, and a sub-agent for splitting equipment names. It is used to perform parallel parsing of the maintenance application ticket data and output structured operation information. The data service interface module is used to call the power grid data service interface to obtain real-time power grid topology data and equipment operating status data; The operation logic deduction module is used to perform operation logic deduction based on the structured operation information and generate an initial scheduling operation ticket containing operation tasks, operation steps, safety measures and equipment numbers. The dual compliance verification module is used to perform dual compliance verification on the initial scheduling operation ticket and generate a target scheduling operation ticket that passes the verification. The dual compliance verification includes the operation logic and specification compliance verification at the whole ticket level, and the matching verification of operation instructions and real-time equipment status at the step level. The output and execution module is used to visualize the target scheduling operation ticket and, in response to the dispatcher's confirmation instruction or automatic execution instruction, synchronize it to the power distribution automation system through an interface to execute the operation.