Work ticket intelligent management method and device based on multi-agent cooperation and knowledge distillation self-learning

Abstract

The application belongs to the technical field of electric power operation and maintenance, and relates to a work ticket intelligent management method and device based on multi-agent cooperation and knowledge distillation self-learning; the method realizes task decomposition, feasibility verification and work ticket automatic generation by acquiring work basis data and completing intelligent analysis, and constructing a multi-agent cooperation framework; a work scene is constructed based on power grid topology simulation, high-risk work steps are identified and corrected through ticket matching degree screening, compliance verification and panoramic risk deduction; after the work ticket is issued, on-site execution and data archiving are completed, the multi-agent cooperation framework parameters are optimized by using an adaptive knowledge distillation mechanism, and model autonomous iteration is realized. The application can improve the work ticket preparation efficiency and compliance, reduce human errors and operation risks, form a whole-process closed-loop management and control and self-learning evolution ability, and is suitable for intelligent safety management and control of power operation such as a substation, and has high practicality and high popularization value.

Description

Technical Field

[0001] This invention belongs to the field of power operation and maintenance technology, specifically relating to an intelligent management method and device for work tickets based on multi-agent collaboration and knowledge distillation self-learning. Background Technology

[0002] With the continuous expansion of the power grid and the increasing complexity of equipment, the operational scenarios of 500kV and above substations are becoming increasingly diverse. The traditional management model of manually compiling and reviewing work permits can no longer meet the needs of safe, efficient, and standardized operation control in the new power system. Work permits are a core organizational measure for maintaining order and ensuring safety in power operations. Their compilation quality, review efficiency, and compliance directly affect the safety of power grid operations. Currently, existing work permit management still relies on manual experience, which has the following prominent problems: First, the compilation efficiency is low, with manual verification of procedures taking several hours, making it difficult to adapt to the needs of high-frequency operations; second, there are many omissions in safety checks, risk control is lagging behind, manual checks are prone to omissions, and there is a lack of pre-emptive prevention measures; third, the standardization is poor, there is no unified template, and reliance on personal experience leads to inconsistent permit quality; fourth, there is a serious waste of human resources, with a large number of technical personnel being involved in repetitive permit filling work.

[0003] In view of this, it is very necessary to provide a work ticket intelligent management method and device based on multi-agent collaboration and knowledge distillation self-learning to solve the above-mentioned defects in the prior art. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing work order management technologies, such as low efficiency, numerous security verification omissions and lagging risk control, poor standardization, and large amounts of repetitive work leading to wasted human resources. The invention provides a work order intelligent management method and device based on multi-agent collaboration and knowledge distillation self-learning to solve these technical problems.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A work ticket intelligent management method based on multi-agent collaboration and knowledge distillation self-learning includes the following steps: Step S1: Obtain basic task data and perform task parsing on the basic task data to generate task information; the task information includes: task type, task, equipment, requirements, and personnel. Step S2: Construct a multi-agent collaboration framework, input job information, and generate structured work orders and supporting safety measures; Step S3: Simulate and model the topology of the power grid equipment to generate work scenarios; use the ticket matching algorithm to filter compliant work scenarios; use the compliance verification model to verify the compliance of the generated work tickets; use the work risk level judgment algorithm and compliant work scenarios to conduct a panoramic risk simulation, identify high-risk steps and make corrections. Step S4: Issue a work order and perform the work task according to the work order content; after the work is completed, close the work order and archive the work data. Step S5: After the assignment is completed, an adaptive knowledge distillation mechanism is adopted. The historical work tickets and assignment feedback data approved by experts are used as input to transfer expert experience to the student model, update and optimize the internal parameters of the multi-agent collaborative framework, and achieve autonomous iterative optimization.

[0006] Preferably, step S1 specifically includes: Step S11: Connect the primary wiring diagram and operation and maintenance management system in the power system to obtain basic work data; the basic work data includes: equipment information, time information, personnel information, work location, work content, and work method; Step S12: Based on the Guangming Power big data model, perform task analysis on the basic data of the operation to generate operation information.

[0007] This step can achieve the following technical effects: By integrating the wiring diagram with the operation and maintenance management system, basic work data is automatically collected, avoiding errors, redundancy, and inefficiency caused by manual data entry. Relying on the Guangming Power big data model, the data is intelligently analyzed to quickly identify work types, tasks, equipment, requirements, and personnel information, resulting in standardized work information output. This process achieves automatic acquisition of data sources and intelligent understanding of tasks, significantly reducing manual intervention, improving the accuracy and response speed of task breakdown, and providing reliable data support for subsequent multi-agent collaboration, ticket generation, and risk verification, ensuring the efficient and standardized start of the entire work ticket management process.

[0008] Preferably, step S2 specifically includes: A multi-agent collaborative framework is constructed, which includes three agents: a planning agent, a validating agent, and a generating agent. The planning agent breaks down the task into several sub-tasks; Verify the feasibility of the agent calling knowledge graphs and safety rules to perform sub-tasks. The intelligent agent is built on the Guangming Power big data model and the power industry knowledge graph to construct a ticket generation model, and generates work tickets and supporting safety measures based on the verified sub-tasks. The work order includes the following fields: ticket number, ticket type, work information, work location, work content, safety measures, work time, and end time.

[0009] This step can achieve the following technical effects: Through the collaborative operation of multiple agents, the system achieves automatic decomposition of work tasks, feasibility verification, and structured generation of work permits. The planning agent rationally breaks down complex tasks, improving the granularity of task processing and execution efficiency. The verification agent uses knowledge graphs and safety regulations for pre-verification, reducing the risk of violations. The generation agent automatically outputs standardized work permits and safety measures based on compliant sub-tasks. The division of labor and collaboration among multiple agents enhances adaptability to complex scenarios, reduces the burden of manual compilation and review, increases the speed and standardization of work permit generation, and ensures that work permits are complete, logically rigorous, and have adequate safety measures, meeting the requirements for power field operation management.

[0010] Preferably, step S3 specifically includes: Step S31: Use the power grid-operation digital twin model to simulate and model the topology of the power grid equipment to generate an operation scenario; Step S32: Call the ticket matching algorithm to calculate the matching score of the generated job scenario: , in, Match score; The weights for the operation parameters include: equipment information weight, time information weight, personnel information weight, and the weights for the standard safety measures corresponding to the generated operation scenario. This represents the cosine similarity; n is the total number of core verification parameters; This is the standard value for this work scenario; The actual value for generating the work order; Output the matching score; when the matching score reaches the preset matching threshold, it is determined to be a compliant operation scenario; otherwise, the operation scenario is regenerated. Step S33: Invoke the compliance verification model built based on the safety regulations rule base to perform compliance verification on the generated work order: Calculate the verification confidence level of the work order's compliance with safety regulations and the five general systems for substation operation and maintenance. , in, To verify the confidence level, To verify the number of approved policy clauses, To verify the total number of policy clauses; Output compliance confidence level; when the verification confidence level reaches the preset confidence level threshold, it is judged as compliant and a verification report is generated; when the verification confidence level is lower than the preset confidence level threshold, the violations and violation clauses in the work order are marked and corrected.

[0011] Step S34: Perform a comprehensive risk simulation using the operational risk level determination algorithm and compliant operational scenarios: Calculate the comprehensive risk score for each operational step. , Where R is the total risk score; Weighting of operational risk scores; Weighting of equipment status risk score; As a weight for environmental risk scores; Assign a risk score to the operation; Assign a risk score to the equipment status; Assign environmental risk score; Four risk thresholds are preset. If the total risk score is lower than the first risk threshold, it is judged as Level IV low risk; if the total risk score is not lower than the first risk threshold and is lower than the second risk threshold, it is judged as Level III medium risk; if the total risk score is not lower than the second risk threshold and is lower than the third risk threshold, it is judged as Level II high risk; if the total risk score is not lower than the fourth risk threshold, it is judged as Level I emergency danger. When a work step is identified as having a risk level of III or higher, the work risk level determination algorithm automatically corrects the work step and regenerates the work order.

[0012] This step can achieve the following technical effects: Based on digital twin simulation of power grid topology and operation scenarios, and combined with ticket matching algorithm to screen compliant scenarios, the operation is consistent with the actual site conditions. The compliance verification model automatically verifies safety regulations and institutional clauses, improving the ticket compliance rate. The risk level judgment algorithm realizes panoramic risk simulation and classification, accurately identifies high-risk steps and automatically corrects them. The whole process realizes pre-simulation, automatic verification, risk prediction and dynamic optimization, reducing safety hazards caused by human negligence, improving risk prevention and control capabilities, realizing the transformation from passive rectification to proactive prevention, and ensuring that the entire operation process is safe, controllable and traceable.

[0013] Preferably, step S4 specifically includes: Step S41: Map the content of the work order and the risk level of the operation to the work order template and operation data template built into the intelligent agent to generate unified operation data, and support export in PDF and Excel formats. Step S42: Issue a work order and perform the work task according to the contents of the work order; Step S43: After the work is completed, terminate the work order and archive the work data.

[0014] This step can achieve the following technical effects: The entire process is managed in a closed loop, from work order issuance and on-site execution to completion and archiving. Standardized documents such as "three measures and one plan" and risk control cards are automatically generated, supporting export in multiple formats and improving the standardization of work documents. Strict adherence to the work order content during execution ensures compliance. The archiving process uniformly retains documents, records, risk measures, and other materials, achieving full traceability. This step simplifies on-site operations, reduces the burden of manual document processing, improves work execution efficiency and management standardization, and provides a reliable data foundation for subsequent review, auditing, and model iteration.

[0015] Preferably, step S5 specifically includes: Step S51: Collect on-site work execution feedback and expert review and error correction data, using historical work orders after expert review and approval, on-site execution feedback and review and error correction data as input samples; Step S52: Collect and standardize the input samples to construct an labeled dataset containing work steps, safety regulations, risk points, and expert correction opinions; Step S53, perform knowledge distillation training based on the labeled dataset: using the Guangming Power large model as the teacher model and the ticket generation model as the student model, the following total loss function is used for training: , in, This is the total loss value. These are the weighting coefficients. For student model loss, The teacher model uses a soft-label loss function; this loss function is used to transfer expert experience, safety regulations knowledge, and on-site practical experience to the student model. Step S54: Update the power industry knowledge graph based on the distillation training results, and add entities and relationships such as new operation scenarios, safety regulations, and equipment risks; Step S55: Optimize the task allocation weight parameters of each agent in the multi-agent collaboration framework, as well as the parameters of the ticket generation model, to improve the accuracy of ticket generation, verification, and risk inference. Step S56: The model iteration takes effect, realizing the autonomous iterative optimization of the ticket generation model and the power knowledge graph, providing more accurate support for subsequent tasks.

[0016] This step can achieve the following technical effects: Through an adaptive knowledge distillation mechanism, expert experience and on-site feedback are transformed into model learning samples, enabling autonomous iterative optimization of parameters and decision-making models in a multi-agent collaborative framework. Knowledge distillation transfers the implicit experience of the Guangming Power large model to student models, continuously improving the accuracy of task allocation, ticket generation, compliance verification, and risk identification. Model iteration can automatically adapt to new equipment, new scenarios, and new rules, continuously improving the system's intelligence level and generalization ability, reducing manual maintenance costs, and forming a self-learning closed loop of "execution-feedback-optimization-improvement," enabling the system to maintain high accuracy and high applicability over the long term.

[0017] In addition, the present invention also provides a work ticket intelligent management device based on multi-agent collaboration and knowledge distillation self-learning, including: a data acquisition and task parsing module, a multi-agent collaboration and ticket generation module, a compliance verification and risk simulation module, a work ticket execution and document archiving module, and a model iteration and optimization module; The data acquisition and task parsing module includes: Acquire basic task data and perform task parsing on the basic task data to generate task information; the task information includes: task type, task, equipment, requirements, and personnel. The multi-agent collaboration and ticket generation module includes: Construct a multi-agent collaboration framework, input job information, and generate structured work orders and supporting safety measures; The compliance verification and risk simulation module includes: Simulation modeling of power grid equipment topology is performed to generate work scenarios; ticket matching algorithm is used to screen compliant work scenarios; compliance verification model is used to verify the compliance of the generated work tickets; and a panoramic risk simulation is performed using an operation risk level determination algorithm and compliant work scenarios to identify and correct high-risk steps. The work order execution and data archiving module includes: Issue work orders and carry out the work tasks according to the work order content; after the work is completed, close the work order and archive the work data; The model iterative optimization module includes: After the assignment is completed, an adaptive knowledge distillation mechanism is adopted. Using historical work orders and assignment feedback data approved by experts as input, the expert experience is transferred to the student model to update and optimize the internal parameters of the multi-agent collaborative framework, thereby achieving autonomous iterative optimization.

[0018] The beneficial effects of this invention are as follows: By automatically generating structured work orders through the Guangming Power large-scale model and the ticket generation model, and processing work tasks in conjunction with a multi-agent collaborative framework, the time for conventional ticket generation is greatly shortened, and efficiency is significantly improved; the work orders are double-verified through a compliance verification model and a work risk level determination algorithm, and dynamic risk simulation is performed using a power grid-work digital twin model, reducing the human error rate by 95% and intercepting high-risk situations; the entire process is standardized by using unified work orders and supporting work data templates, eliminating differences in human experience and achieving closed-loop control of the entire process of generation, verification, execution, termination, and evaluation; an adaptive knowledge distillation mechanism is adopted, enabling the model to autonomously learn from expert experience, periodically update the knowledge graph and model parameters, adapt to power grid upgrades in the long term, and reduce the need for frequent manual maintenance, thus reducing the waste of human resources. In addition, the modules of this invention are lightweight, have low deployment costs, and can be widely promoted.

[0019] Therefore, it is evident that the present invention has outstanding substantive features and significant progress compared with the prior art, and the beneficial effects of its implementation are also obvious. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0021] Figure 1 This is a flowchart of an intelligent work ticket management method based on multi-agent collaboration and knowledge distillation self-learning provided by the present invention.

[0022] Figure 2 This is a schematic diagram of the principle of an intelligent work ticket management device based on multi-agent collaboration and knowledge distillation self-learning provided by the present invention.

[0023] Figure 3 This is a flowchart of the intelligent generation and verification process for work tickets provided by the present invention.

[0024] Figure 4 This is a flowchart of the panoramic risk simulation provided by the present invention.

[0025] Figure 5 This is a flowchart of the self-learning iterative process based on the knowledge distillation mechanism provided by the present invention.

[0026] The module consists of five parts: 1-Data Acquisition and Task Parsing; 2-Multi-Agent Collaboration and Ticket Generation; 3-Compliance Verification and Risk Inference; 4-Work Ticket Execution and Document Archiving; and 5-Model Iteration and Optimization. Detailed Implementation

[0027] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. The following embodiments are explanations of the present invention, but the present invention is not limited to the following implementation methods.

[0028] Example 1: like Figure 1 As shown in the figure, this embodiment provides an intelligent management method for work tickets based on multi-agent collaboration and knowledge distillation self-learning, which includes the following steps: Step S1: Obtain basic task data and perform task parsing on the basic task data to generate task information; the task information includes: task type, task, equipment, requirements, and personnel. Step S1 specifically includes: Step S11: Connect the primary wiring diagram and operation and maintenance management system in the power system to obtain basic work data; the basic work data includes: equipment information, time information, personnel information, work location, work content, and work method; Step S12: Based on the Guangming Power big data model, perform task analysis on the basic data of the operation to generate operation information.

[0029] This step can achieve the following technical effects: By integrating the wiring diagram with the operation and maintenance management system, basic work data is automatically collected, avoiding errors, redundancy, and inefficiency caused by manual data entry. Relying on the Guangming Power big data model, the data is intelligently analyzed to quickly identify work types, tasks, equipment, requirements, and personnel information, resulting in standardized work information output. This process achieves automatic acquisition of data sources and intelligent understanding of tasks, significantly reducing manual intervention, improving the accuracy and response speed of task breakdown, and providing reliable data support for subsequent multi-agent collaboration, ticket generation, and risk verification, ensuring the efficient and standardized start of the entire work ticket management process.

[0030] Step S2: Construct a multi-agent collaboration framework, input job information, and generate structured work orders and supporting safety measures; Step S2 specifically includes: A multi-agent collaborative framework is constructed, which includes three agents: a planning agent, a validating agent, and a generating agent. The planning agent breaks down the task into several sub-tasks; Verify the feasibility of the agent calling knowledge graphs and safety rules to perform sub-tasks. The intelligent agent is built on the Guangming Power big data model and the power industry knowledge graph to construct a ticket generation model, and generates work tickets and supporting safety measures based on the verified sub-tasks. The work order includes the following fields: ticket number, ticket type, work information, work location, work content, safety measures, work time, and end time.

[0031] To ensure efficient collaboration among agents, task weights are assigned to each agent: , in, The task weight for the k-th agent; The agent's capability coefficient; For the efficiency of intelligent agent execution; The total number of intelligent agents.

[0032] This step can achieve the following technical effects: Through the collaborative operation of multiple agents, the system achieves automatic decomposition of work tasks, feasibility verification, and structured generation of work permits. The planning agent rationally breaks down complex tasks, improving the granularity of task processing and execution efficiency. The verification agent uses knowledge graphs and safety regulations for pre-verification, reducing the risk of violations. The generation agent automatically outputs standardized work permits and safety measures based on compliant sub-tasks. The division of labor and collaboration among multiple agents enhances adaptability to complex scenarios, reduces the burden of manual compilation and review, increases the speed and standardization of work permit generation, and ensures that work permits are complete, logically rigorous, and have adequate safety measures, meeting the requirements for power field operation management.

[0033] Step S3, as follows Figure 4 As shown, the power grid equipment topology is simulated and modeled to generate work scenarios; a ticket matching algorithm is called to filter compliant work scenarios; a compliance verification model is called to verify the compliance of the generated work tickets; and a work risk level determination algorithm is used to perform a panoramic risk simulation with compliant work scenarios to identify high-risk steps and make corrections. Step S3 specifically includes: Step S31: Use the power grid-operation digital twin model to simulate and model the topology of the power grid equipment to generate an operation scenario; Step S32: Call the ticket matching algorithm to calculate the matching score of the generated job scenario: , in, Match score; The weights for the operation parameters include: equipment information weight (0.3), time information weight (0.2), personnel information weight (0.1), and the weight of the standard safety measures corresponding to the generated operation scenario (0.4). Each weight is determined based on expert experience. This represents the cosine similarity; n is the total number of core verification parameters; This is the standard value for this work scenario; The actual value for generating the work order; Output the matching score; when the matching score reaches 0.85, it is determined to be a compliant operation scenario; otherwise, the operation scenario is regenerated. Step S33: Invoke the compliance verification model built based on the safety regulation rule base to perform compliance verification on the generated work order, such as... Figure 3 As shown: Calculate the confidence level of the ticket's compliance with safety regulations and the five general systems for substation operation and maintenance: , in, To verify the confidence level, To verify the number of approved policy clauses, To verify the total number of policy clauses; Output compliance confidence level; when the verification confidence level reaches the preset confidence level value of 90%, it is judged as compliant and a verification report is generated; when the verification confidence level is lower than the preset confidence level value of 90%, the violations and violation clauses in the work order are marked and corrected.

[0034] Step S34: Perform a comprehensive risk simulation using the operational risk level determination algorithm and compliant operational scenarios: Calculate the comprehensive risk score for each operational step. , Where R is the total risk score; The weight for the operational risk score is set at 0.5. The weight for the equipment status risk score is 0.3. The environmental risk score weight is set to 0.2; each weight is derived based on expert experience. Assign a risk score to the operation; Assign a risk score to the equipment status; Assign environmental risk score; The system has four preset risk thresholds. If the total risk score is below 40, it is classified as Level IV (low risk); if the total risk score is not lower than 40 but lower than 60, it is classified as Level III (medium risk); if the total risk score is not lower than 60 but lower than 80, it is classified as Level II (high risk); and if the total risk score is not lower than 80, it is classified as Level I (emergency danger). When a work step is identified as having a risk level of III or higher, the work risk level determination algorithm automatically corrects the work step and regenerates the work order.

[0035] This step can achieve the following technical effects: Based on digital twin simulation of power grid topology and operation scenarios, and combined with ticket matching algorithm to screen compliant scenarios, the operation is consistent with the actual site conditions. The compliance verification model automatically verifies safety regulations and institutional clauses, improving the ticket compliance rate. The risk level judgment algorithm realizes panoramic risk simulation and classification, accurately identifies high-risk steps and automatically corrects them. The whole process realizes pre-simulation, automatic verification, risk prediction and dynamic optimization, reducing safety hazards caused by human negligence, improving risk prevention and control capabilities, realizing the transformation from passive rectification to proactive prevention, and ensuring that the entire operation process is safe, controllable and traceable.

[0036] Step S4: Issue a work order and perform the work task according to the work order content; after the work is completed, close the work order and archive the work data. Step S4 specifically includes: Step S41: Map the content of the work order and the risk level of the operation to the work order template and operation data template built into the intelligent agent to generate unified operation data, and support export in PDF and Excel formats. Step S42: Issue a work order and perform the work task according to the contents of the work order; Step S43: After the work is completed, terminate the work order and archive the work data.

[0037] This step can achieve the following technical effects: The entire process is managed in a closed loop, from work order issuance and on-site execution to completion and archiving. Standardized documents such as "three measures and one plan" and risk control cards are automatically generated, supporting export in multiple formats and improving the standardization of work documents. Strict adherence to the work order content during execution ensures compliance. The archiving process uniformly retains documents, records, risk measures, and other materials, achieving full traceability. This step simplifies on-site operations, reduces the burden of manual document processing, improves work execution efficiency and management standardization, and provides a reliable data foundation for subsequent review, auditing, and model iteration.

[0038] Step S5: After the assignment is completed, an adaptive knowledge distillation mechanism is adopted. The historical work tickets and assignment feedback data approved by experts are used as input to transfer expert experience to the student model, update and optimize the internal parameters of the multi-agent collaborative framework, and achieve autonomous iterative optimization.

[0039] like Figure 5 As shown, step S5 specifically includes: Step S51: Collect on-site work execution feedback and expert review and error correction data, using historical work orders after expert review and approval, on-site execution feedback and review and error correction data as input samples; Step S52: Collect and standardize the input samples to construct an labeled dataset containing work steps, safety regulations, risk points, and expert correction opinions; Step S53, perform knowledge distillation training based on the labeled dataset: using the Guangming Power large model as the teacher model and the ticket generation model as the student model, the following total loss function is used for training: , in, This is the total loss value. The weight coefficients are set to 0.3. The Adam optimizer is used to optimize the student model and the weight coefficients. Joint iterative optimization, backpropagation to update parameters, and output of optimal weight coefficients when loss converges, which are set to 0.3 after training and optimization; For student model loss, The teacher model uses a soft-label loss function; this loss function is used to transfer expert experience, safety regulations knowledge, and on-site practical experience to the student model. Step S54: Update the power industry knowledge graph based on the distillation training results, and add entities and relationships such as new operation scenarios, safety regulations, and equipment risks; Step S55: Optimize the task allocation weight parameters of each agent in the multi-agent collaboration framework, as well as the parameters of the ticket generation model, to improve the accuracy of ticket generation, verification, and risk inference. Step S56: The model iteration takes effect, realizing the autonomous iterative optimization of the ticket generation model and the power knowledge graph, providing more accurate support for subsequent tasks.

[0040] This step can achieve the following technical effects: Through an adaptive knowledge distillation mechanism, expert experience and on-site feedback are transformed into model learning samples, enabling autonomous iterative optimization of parameters and decision-making models in a multi-agent collaborative framework. Knowledge distillation transfers the implicit experience of the Guangming Power large model to student models, continuously improving the accuracy of task allocation, ticket generation, compliance verification, and risk identification. Model iteration can automatically adapt to new equipment, new scenarios, and new rules, continuously improving the system's intelligence level and generalization ability, reducing manual maintenance costs, and forming a self-learning closed loop of "execution-feedback-optimization-improvement," enabling the system to maintain high accuracy and high applicability over the long term.

[0041] Example 2: like Figure 2 As shown in the figure, this embodiment provides a work ticket intelligent management device based on multi-agent collaboration and knowledge distillation self-learning, including: a data acquisition and task parsing module 1, a multi-agent collaboration and ticket generation module 2, a compliance verification and risk simulation module 3, a work ticket execution and document archiving module 4, and a model iteration and optimization module 5. The data acquisition and task parsing module 1 includes: Acquire basic task data and perform task parsing on the basic task data to generate task information; the task information includes: task type, task, equipment, requirements, and personnel. The multi-agent collaboration and ticket generation module 2 includes: Construct a multi-agent collaboration framework, input job information, and generate structured work orders and supporting safety measures; The compliance verification and risk simulation module 3 includes: Simulation modeling of power grid equipment topology is performed to generate work scenarios; ticket matching algorithm is used to screen compliant work scenarios; compliance verification model is used to verify the compliance of the generated work tickets; and a panoramic risk simulation is performed using an operation risk level determination algorithm and compliant work scenarios to identify and correct high-risk steps. The work order execution and data archiving module 4 includes: Issue work orders and carry out the work tasks according to the work order content; after the work is completed, close the work order and archive the work data; The model iterative optimization module 5 includes: After the assignment is completed, an adaptive knowledge distillation mechanism is adopted. Using historical work orders and assignment feedback data approved by experts as input, the expert experience is transferred to the student model to update and optimize the internal parameters of the multi-agent collaborative framework, thereby achieving autonomous iterative optimization.

[0042] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. The methods disclosed in the embodiments are described simply because they correspond to the systems disclosed in the embodiments; relevant details can be found in the method section.

[0043] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction 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 and steps 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 this invention.

[0044] In the embodiments provided by this invention, it should be understood that the disclosed systems, methods, and approaches can be implemented in other ways. For example, the system 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 coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.

[0045] 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.

[0046] In addition, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each module can exist physically separately, or two or more modules can be integrated into one unit.

[0047] Similarly, in the various embodiments of the present invention, each processing unit can be integrated into a functional module, or each processing unit can exist physically, or two or more processing units can be integrated into a functional module.

[0048] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0049] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0050] The above-disclosed embodiments are merely preferred embodiments of the present invention, but the present invention is not limited thereto. Any non-creative variations that can be conceived by those skilled in the art, as well as any improvements and modifications made without departing from the principles of the present invention, should fall within the protection scope of the present invention.

Claims

1. A work ticket intelligent management method based on multi-agent collaboration and knowledge distillation self-learning, characterized in that, Includes the following steps: Step S1: Obtain basic task data and perform task parsing on the basic task data to generate task information; the task information includes at least the task. Step S2: Construct a multi-agent collaboration framework, input job information, and generate structured work orders and supporting safety measures; Step S3: Simulate and model the topology of the power grid equipment to generate the operation scenario; Use a ticket matching algorithm to filter compliant operation scenarios; The compliance verification model is invoked to perform compliance verification on the generated work order; A comprehensive risk simulation was conducted using an operational risk level assessment algorithm and compliant operational scenarios to identify and correct high-risk steps. Step S4: Issue a work order and perform the work task according to the contents of the work order; After the work is completed, the work order is closed and the work data is archived; Step S5: After the assignment is completed, an adaptive knowledge distillation mechanism is used. The historical work orders and assignment feedback data approved by experts are used as input to transfer expert experience to the student model and update and optimize the internal parameters of the multi-agent collaborative framework.

2. The intelligent management method for work tickets based on multi-agent collaboration and knowledge distillation self-learning as described in claim 1, characterized in that, Step S1 specifically includes: Step S11: Connect the primary wiring diagram and operation and maintenance management system in the power system to obtain basic work data; the basic work data includes: equipment information, time information, personnel information, work location, work content, and work method; Step S12: Based on the Guangming Power big data model, perform task analysis on the basic operation data to generate operation information; the operation information includes: operation type, operation task, operation equipment, operation requirements, and operation personnel.

3. The intelligent management method for work tickets based on multi-agent collaboration and knowledge distillation self-learning as described in claim 1, characterized in that, Step S2 specifically includes: A multi-agent collaborative framework is constructed, which includes three agents: a planning agent, a validating agent, and a generating agent. The planning agent breaks down the task into several sub-tasks; Verify the feasibility of the agent calling knowledge graphs and safety rules to perform sub-tasks. The generated intelligent agent constructs a ticket generation model based on the Guangming Power big data model and the power industry knowledge graph, and generates work tickets and supporting safety measures based on the verified sub-tasks.

4. The intelligent management method for work tickets based on multi-agent collaboration and knowledge distillation self-learning as described in claim 3, characterized in that, The generated work order includes the following fields: ticket number, ticket type, job information, work location, job content, safety measures, work time, and end time.

5. The intelligent management method for work tickets based on multi-agent collaboration and knowledge distillation self-learning as described in claim 1, characterized in that, Step S3 specifically includes: Step S31: Use the power grid-operation digital twin model to simulate and model the topology of the power grid equipment to generate an operation scenario; Step S32: Call the ticket matching algorithm to calculate the matching score of the generated job scenario: ， in, Match score; The weights for the operation parameters include: equipment information weight, time information weight, personnel information weight, and the weights for the standard safety measures corresponding to the generated operation scenario. This represents the cosine similarity; n is the total number of core verification parameters; This is the standard value for this work scenario; The actual value for generating the work order; Output the matching score; when the matching score reaches the preset matching threshold, it is determined to be a compliant operation scenario; otherwise, the operation scenario is regenerated. Step S33: Invoke the compliance verification model built based on the safety regulations rule base to perform compliance verification on the generated work order: Calculate the verification confidence level of the work order's compliance with safety regulations and the five general systems for substation operation and maintenance. ， in, To verify the confidence level, To verify the number of approved policy clauses, To verify the total number of policy clauses; Output compliance confidence level; when the verification confidence level reaches the preset confidence level threshold, it is judged as compliant and a verification report is generated; when the verification confidence level is lower than the preset confidence level threshold, the violations and violation clauses in the work order are marked and corrected. Step S34: Use the operation risk level determination algorithm and compliant operation scenario to conduct a panoramic risk simulation, calculate the comprehensive risk score of each operation step, and obtain the total risk score of each operation step. Four risk thresholds are preset. If the total risk score is lower than the first risk threshold, it is judged as Level IV low risk; if the total risk score is not lower than the first risk threshold and is lower than the second risk threshold, it is judged as Level III medium risk; if the total risk score is not lower than the second risk threshold and is lower than the third risk threshold, it is judged as Level II high risk; if the total risk score is not lower than the fourth risk threshold, it is judged as Level I emergency danger. When a work step is identified as having a risk level of III or higher, the work risk level determination algorithm automatically corrects the work step and regenerates the work order.

6. The intelligent management method for work tickets based on multi-agent collaboration and knowledge distillation self-learning as described in claim 5, characterized in that, The method for calculating the total risk score for each work step in step S34 is as follows: ， Where R is the total risk score; Weighting of operational risk scores; Weighting of equipment status risk score; As a weight for environmental risk scores; Assign a risk score to the operation; Assign a risk score to the equipment status; Environmental risk score.

7. The intelligent management method for work tickets based on multi-agent collaboration and knowledge distillation self-learning as described in claim 1, characterized in that, Step S4 specifically includes: Step S41: Map the content of the work order and the risk level of the operation to the work order template and operation data template built into the intelligent agent to generate unified operation data, and support export in PDF and Excel formats. Step S42: Issue a work order and perform the work task according to the contents of the work order; Step S43: After the work is completed, terminate the work order and archive the work data.

8. The intelligent management method for work tickets based on multi-agent collaboration and knowledge distillation self-learning as described in claim 1, characterized in that, Step S5 specifically includes: Step S51: Collect on-site work execution feedback and expert review and error correction data, using historical work orders after expert review and approval, on-site execution feedback and review and error correction data as input samples; Step S52: Collect and standardize the input samples to construct an labeled dataset containing work steps, safety regulations, risk points, and expert correction opinions; Step S53, perform knowledge distillation training based on the labeled dataset: use the Guangming Power large model as the teacher model and the ticket generation model as the student model, and use the knowledge distillation loss function for training; Step S54: Update the power industry knowledge graph based on the distillation training results, and add entities and relationships such as new operation scenarios, safety regulations, and equipment risks; Step S55: Optimize the task allocation weight parameters of each agent in the multi-agent collaboration framework, as well as the parameters of the ticket generation model. Step S56: Complete the autonomous iterative optimization of the ticket generation model and the power knowledge graph.

9. The intelligent management method for work tickets based on multi-agent collaboration and knowledge distillation self-learning as described in claim 8, characterized in that, The knowledge distillation loss function in step S53 is: ， in, This is the total loss value. These are the weighting coefficients. For student model loss, The soft label loss is used for the teacher model.

10. A work ticket intelligent management device based on multi-agent collaboration and knowledge distillation self-learning, characterized in that, include: The module includes: data acquisition and task parsing, multi-agent collaboration and ticket generation, compliance verification and risk simulation, work order execution and document archiving, and model iteration and optimization. The data acquisition and task parsing module includes: Acquire basic task data and perform task parsing on the basic task data to generate task information; the task information includes: task type, task, equipment, requirements, and personnel. The multi-agent collaboration and ticket generation module includes: Construct a multi-agent collaboration framework, input job information, and generate structured work orders and supporting safety measures; The compliance verification and risk simulation module includes: Simulation modeling of power grid equipment topology is performed to generate work scenarios; ticket matching algorithm is used to screen compliant work scenarios; compliance verification model is used to verify the compliance of the generated work tickets; and a panoramic risk simulation is performed using an operation risk level determination algorithm and compliant work scenarios to identify and correct high-risk steps. The work order execution and data archiving module includes: Issue work orders and carry out the work tasks according to the work order content; after the work is completed, close the work order and archive the work data; The model iterative optimization module includes: After the assignment is completed, an adaptive knowledge distillation mechanism is adopted, using historical work orders and assignment feedback data approved by experts as input to transfer expert experience to the student model and update and optimize the internal parameters of the multi-agent collaborative framework.

Images