Graphical automatic ticket forming method based on NLP

Abstract

The invention provides a graphical automatic ticket forming method based on NLP. In the scheme, an operator sends an operation ticket forming request to a server through a first terminal device, the server analyzes operation content, extracts key information and generates a graphical operation ticket by using an NLP technology, the operation ticket is sent to a second terminal device for auditing, and after the auditing is passed, the operation ticket forming request is sent to the second terminal device. The work ticket is determined as a target work ticket and is returned to the first terminal equipment. Through the above method, the NLP-based graphical automatic ticket forming method effectively reduces human errors and operation risks, guarantees the safety and normalization of the operation process, provides comprehensive, systematic and intelligent support for power grid management, and greatly improves the quality and safety standard of operation management.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a graphical automatic ticketing method based on NLP. Background Art

[0002] With the continuous expansion of the power grid and the development of information technology, the company has put forward new standards and requirements for the two-ticket management. The two-ticket management is the basis for ensuring the safety of power maintenance, testing, infrastructure and other work.

[0003] At present, in the process of power production, the work ticket is filled out by maintenance personnel, which needs to include information such as work tasks, safety measures, work permit holders, work supervisors, etc., and then signed by the responsible person. The operation ticket is filled out by the operator according to the operation task and equipment status, and is reviewed and issued by the responsible person. The operator must strictly follow the instructions on the operation ticket to ensure the safety of the operation process.

[0004] However, the current manual filling of work tickets is cumbersome and complicated, and the process of generating work tickets is inefficient. Summary of the invention

[0005] The present application provides a graphical automatic ticket generation method based on NLP to solve the problems of cumbersome manual filling of work tickets, complicated operations, and low efficiency in the process of generating work tickets, and provides a technical solution for realizing the computerization and automation of the work ticket management system.

[0006] In a first aspect, the present application provides a graphical automatic ticketing method based on NLP, which is applied to a power grid operation ticket management system, wherein the power grid operation ticket management system includes a server, a first terminal device, and a second terminal device, and the method includes: The first terminal device sends a job ticket generation request to the server, wherein the job ticket generation request carries the job content related to the current job input by the operator; The server processes the job content according to the job ticket creation request using a natural language processing model NLP to obtain key information corresponding to the job; The server obtains the job ticket according to the key information and the information processing model corresponding to the job ticket type; The server sends a job ticket review request to the second terminal device, wherein the job ticket review request carries the job ticket; The second terminal device reviews the operation ticket according to the operation ticket review request using a pre-acquired power grid management knowledge graph, wherein the power grid management knowledge graph includes multiple operation modes, multiple safety measures, dependencies between different operation modes, and dependencies between safety measures and operation modes; After the job ticket is reviewed and approved, the second terminal device sends review confirmation information to the server. The server determines the job ticket as a target job ticket based on the review confirmation information and sends the target job ticket to the first terminal device.

[0007] In a specific embodiment, the method further comprises: According to the key information, the type of the job ticket for this job is determined by using a pre-configured job ticket identification rule; Among them, the job ticket identification rule includes: if the key information includes multiple job steps and the time corresponding to each job step, then the job ticket type is determined to be an operation ticket; otherwise, the job ticket type is determined to be a work ticket.

[0008] In a specific implementation, the natural language processing model NLP is used to process the job content to obtain key information corresponding to the job, including: By using NLP, the operation content is parsed to obtain the dependency relationship between the words in the operation content and the multiple entities in the operation ticket involved in the operation content; Information is extracted according to the dependency relationship between the words and the multiple entities to obtain the key information.

[0009] In a specific implementation, if the type of the work ticket is an operation ticket, the key information includes: an operator, a supervisor, an operation device, multiple operation modes and operation sequences, the operation time of each operation mode, and safety measures; If the type of the work ticket is a work ticket, the key information includes: staff, reviewer, work location, work time, work methods involved in the work task, and safety measures for each work method; Among them, the operation mode includes the operation mode or the working mode.

[0010] In a specific implementation, the server obtains the job ticket according to the key information and the information processing model corresponding to the job ticket type, including: The server inputs each operation mode and corresponding safety measures in the key information into the information processing model corresponding to the operation ticket type for processing, and outputs the standardized operation mode and corresponding safety measures corresponding to the operation ticket type; The server writes the standardized operation method and the corresponding safety measures, as well as other information in the key information, into a preset template corresponding to the operation ticket type to obtain the operation ticket.

[0011] In a specific embodiment, the method further comprises: For each job ticket type, an initial model is built based on the deep learning model; Acquire a training sample set corresponding to the work ticket type, the training sample set comprising: a plurality of samples, each sample comprising a description of a first work mode and safety measures input by a worker, and a manually annotated description of a standardized second work mode and safety measures corresponding to the first work mode and safety measures; Based on the training sample set, the initial model is trained to obtain the information processing model.

[0012] In a specific implementation, the training of the initial model based on the training sample set to obtain the information processing model includes: In the process of training the initial model according to the training sample set, the model training parameters are optimized until a preset convergence function converges to obtain the information processing model.

[0013] In a specific implementation, the use of the pre-acquired power grid management knowledge graph to review the operation ticket includes: Obtaining personnel information from the work ticket, and reviewing whether each person in the personnel information has the corresponding work authority according to the work authority of each person stored in the database; Determine whether the target operation mode and target safety measures included in the operation ticket conform to the dependency relationship in the power grid management knowledge graph; If each person in the personnel information in the operation ticket has the corresponding operation authority, and the target operation mode and the target safety measures both conform to the dependency relationship in the power grid management knowledge graph, the operation ticket is approved; Otherwise, the job ticket will not pass the review.

[0014] In a specific embodiment, the method further comprises: If the job ticket fails to pass the review, the second terminal device returns error information and / or modification suggestions to the server; The server sends the error information and / or the modification suggestion to the first terminal device.

[0015] In a second aspect, the present application provides a power grid operation ticket management system, including: a server, a first terminal device and a second terminal device, wherein: The first terminal device is used to send a job ticket generation request to the server, wherein the job ticket generation request carries the job content related to the current job input by the operator; The server is used to: According to the job ticket creation request, the job content is processed using a natural language processing model NLP to obtain key information corresponding to the job; Input the key information into a generation model corresponding to the job ticket type to generate a job ticket; Sending a job ticket review request to the second terminal device, wherein the job ticket review request carries the job ticket; The second terminal device is used for: According to the operation ticket review request, the operation ticket is reviewed using a pre-acquired power grid management knowledge graph, wherein the power grid management knowledge graph includes multiple operation modes, multiple safety measures, dependencies between different operation modes, and dependencies between safety measures and operation modes; After the job ticket is reviewed and approved, a review confirmation message is sent to the server; The server is further configured to determine the job ticket as a target job ticket based on the audit confirmation information, and send the target job ticket to the first terminal device.

[0016] The NLP-based graphical automatic ticketing method provided in this application is applied to the power grid operation ticket management system, aiming to improve the efficiency and automation level of the power grid operation ticket management system. The method sends a job ticket generation request to the server through the first terminal device. The server parses the job content through NLP technology, extracts key information, and generates a job ticket according to the preset job ticket type information processing model. And the second terminal device reviews it based on the power grid management knowledge graph to ensure that the content of the job ticket meets the safety and operating specifications. After the review is passed, the server sends the final target job ticket to the first terminal device. Through the above method, the cumbersome steps of manual filling and review are greatly reduced, the efficiency of job ticket generation and management is improved, and the accuracy and standardization of the content of the job ticket are ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0018] Figure 1 A flowchart of a first embodiment of a graphical automatic ticketing method based on NLP provided in an embodiment of the present application; Figure 2 A flowchart of a second embodiment of a graphical automatic ticketing method based on NLP provided in an embodiment of the present application; Figure 3 A flowchart of a third embodiment of a graphical automatic ticketing method based on NLP provided in an embodiment of the present application; Figure 4 A flowchart of a fourth embodiment of a graphical automatic ticketing method based on NLP provided in an embodiment of the present application; Figure 5 A schematic diagram of a specific implementation of a graphical automatic ticketing method based on NLP provided in an embodiment of the present application; Figure 6 A schematic diagram of an application scenario of a graphical automatic ticketing method based on NLP provided in an embodiment of the present application.

[0019] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0020] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0021] Before introducing the embodiments of the present application, the application background of the embodiments of the present application is first explained: The power industry is a key pillar industry for my country's national economy and people's livelihood. With the continuous development of the national economy, the scale of the power grid continues to expand, and the grid structure becomes increasingly complex. The importance of effectively improving the safe and stable operation level of the power industry has become increasingly prominent. The "Electric Power Industry Safety Work Regulations" is the basic law of the power industry. The "two-ticket" system is an important measure to ensure the safety of personnel and equipment in the power system, and is an important part of the "Electric Power Industry Safety Work Regulations".

[0022] The "two tickets" specifically refer to work tickets and operation tickets. At present, the power system requires "work with tickets and operation with tickets" to reduce the probability of misoperation and ensure the safety of personnel and equipment. Therefore, the "two tickets" are known as the "life tickets of the power system". The work ticket refers to a written document filled out by the person in charge of the work and approved by the relevant departments before the inspection, maintenance, construction, testing and other operations of power facilities, equipment, lines, substations, power plants and other work sites. It stipulates the content of the operation, operators, safety measures, operation time, etc., and is used to ensure the safety, standardization and orderly progress of the work tasks. The work ticket is filled out by the person in charge of the work and can only be used after approval by the relevant departments. The approval department should carefully review the content of the work ticket to ensure the safety of the operation. Before the work ticket is issued, it must go through a strict approval process to ensure that all safety measures are in place and all participants are clear about their respective responsibilities and tasks.

[0023] An operation ticket refers to a written document filled out by operators and approved by relevant departments before operating, debugging, testing and other operations at power facilities, equipment, lines, substations, power plants and other workplaces. It stipulates the operation content, operators, safety measures, operation time, etc. It is used to ensure the safe, standardized and orderly operation of power equipment and systems. The operation ticket is filled out by the operator and can only be used after approval by relevant departments. The approval department should carefully review the content of the operation ticket to ensure safe operation. The issuance and execution of the operation ticket must be carried out in strict accordance with the operating procedures and safety regulations of the power system to ensure the accuracy and safety of the operation.

[0024] With the continuous expansion of the scale of power grids and the development of information technology, the traditional method of manually filling out work tickets can no longer meet the requirements of power safety production. The efficient automation and intelligent realization of the two-ticket system has become the main trend of the power system. The traditional manual filling of work tickets is filled out by staff, which requires staff to be extremely familiar with the on-site equipment conditions and have rich knowledge of power system operation. The manual filling is labor-intensive and has low accuracy. All texts need to be filled out manually. If the filling is wrong, the ticket needs to be re-issued, resulting in reduced work efficiency and paper waste. In addition, after the work ticket and operation ticket are executed, they need to be archived for future inquiries, but the ticket loss phenomenon occurs from time to time, especially the work ticket needs to be divided into two copies and kept by maintenance and operation personnel respectively, which is more likely to be lost, thereby affecting the management efficiency of the decision-making department. In addition, the issuance and circulation of work tickets and operation tickets takes a lot of time. The work ticket needs to go through multiple departments from filling to execution. Due to the long distance between the offices of various departments, a lot of time is wasted in the circulation process.

[0025] In summary, realizing the computerization and automation of the work ticket and operation ticket system and providing an automated ticketing method have become urgent issues to be solved.

[0026] Based on the above technical problems, the inventors found that in the process of studying the automatic generation of work tickets, by using natural language processing (NLP) technology and power grid management knowledge graph, the problems of low efficiency, complex information processing, cumbersome audit process and lack of intelligent support in the traditional work ticket generation and review process can be effectively solved. Based on this, the present application provides a graphical automatic ticket generation method based on NLP, which can be applied to the power grid work ticket management system. The system can convert the unstructured natural language text input by the operator into structured data, extract key information, and automatically generate work tickets. The power grid management knowledge graph provides standard operating procedures and safety measures. By comparing the work content with the standards in the knowledge graph, it ensures that the content of the work ticket is reasonable and meets safety specifications. Through the above method, the automation and intelligence of work ticket generation and review are realized, human errors and operational risks are reduced, and the safety and standardization of the work process are ensured.

[0027] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0028] Figure 1 The present invention provides a flowchart of a first embodiment of a graphical automatic ticketing method based on NLP, which can be applied to a power grid operation ticket management system, such as Figure 1 As shown, the graphical automatic ticketing method based on NLP provided in this embodiment specifically includes: S101: The first terminal device sends a job ticket generation request to the server, and the job ticket generation request carries the job content related to the current job input by the operator.

[0029] In this step, in order to ensure that the server can receive the job ticket request sent by the first terminal device in time, so as to send the job ticket review request to the second terminal device in time, the communication connection between the first terminal device, the server and the second terminal device is first established. At the same time, the operation content related to this operation input by the operator is configured on the first terminal device, and the server extracts the key information corresponding to this operation and generates a job ticket. Among them, the first terminal device, as the initiator of the job ticket request, is the entrance for the operator to interact with the server. The server assumes the role of data processing center, information processing and generation center, communication center and review coordinator in the whole system, and completes the whole process from receiving the job ticket request, processing the operation content, generating the job ticket, sending the review request, receiving the review confirmation information, to finally determining and sending the target job ticket. As the executor of the job ticket review, the second terminal device is responsible for receiving the job ticket review request from the server, and converting it into specific review confirmation information through the review process, so as to realize the review and confirmation of the job ticket.

[0030] Operation tickets and work tickets are the two main types of work tickets, which differ in the content and purpose of the tickets. Operation tickets are mainly used to record and guide specific operation steps to ensure the safety and standardization of the operation process. They usually include information such as operation items, operation steps, operation time, operators, operation locations, equipment status and safety measures, etc., to ensure that operators operate according to the predetermined steps and sequence to avoid misoperation and safety accidents. Work tickets are mainly used to record and manage work tasks to ensure the safety and effective execution of the work process. They usually include information such as work task description, operation time, operators, operation location, work tools, work environment, risk assessment and safety measures, etc., to ensure that operators fully understand the specific requirements and potential risks of the work tasks and take appropriate safety measures.

[0031] Specifically, take the example of an operator sending a job ticket request to the server through a laptop computer as the first terminal device. The request carries the job content related to the job input by the operator's voice, and the job content is: "Appoint Zhang San to perform routine operations on the main circuit breaker of the substation on September 24, 2024, and Li Si to supervise. At 8 o'clock in the morning of that day, turn off the main circuit breaker and put a warning sign to ensure safety. Then at 8:15, check the mechanical condition and electrical connection of the circuit breaker to ensure that there are no abnormalities. Perform an insulation test on the circuit breaker at 8:30 and record the test results. After 15 minutes, lubricate the operating mechanism of the circuit breaker to ensure its flexibility. Finally, restore the power supply of the circuit breaker at 9 o'clock and conduct a trial run. After confirming that it is working properly, record the operation and sign for confirmation. During the whole process, you must wear insulating gloves and goggles, and be sure to use insulating tools when operating electrical equipment to ensure that all safety measures are in place." The terminal device uses a voice plug-in to convert the work content entered by the operator into text. After the server receives a ticket request carrying the work content from a laptop computer that has established a communication connection with it, it parses the work content.

[0032] S102: The server processes the job content according to the job ticket creation request using the natural language processing model NLP to obtain key information corresponding to the job.

[0033] In this step, natural language processing NLP aims to enable computers to understand, interpret and generate human language. In a specific implementation, NLP is used to process the job content input by the operator. Specifically, when the server receives the job-related content input by the operator on the first terminal device, it is parsed by NLP, including but not limited to word segmentation, part-of-speech tagging, named entity recognition, syntactic analysis and semantic analysis. These technologies work together to extract key information from the input natural language text, such as job type, operation steps, required tools and safety precautions. Through the above method, the conversion of unstructured natural language text into structured data is completed, laying the foundation for the subsequent generation of job tickets.

[0034] Specifically, an operator sends a job ticket request to the server through a laptop computer. The request carries the job content related to the job input by the operator's voice. The specific job content is as described above and will not be repeated here. The server processes the job content through NLP technology based on the job ticket request sent by the first terminal device to obtain the key information corresponding to the job.

[0035] In a specific implementation, the server parses the job content through NLP to obtain the dependency relationship between words in the job content and multiple entities in the job ticket involved in the job content.

[0036] In this solution, the input work content is first segmented to separate the continuous text into independent words. Segmentation is a basic step in NLP. Through segmentation, the system can break down the input natural language text into basic vocabulary units, which are the basis for subsequent analysis. Then, the system identifies the part of speech of each word, such as noun, verb, adjective, etc., to help it understand the grammatical role of each word in the sentence.

[0037] Specifically, the server first performs word segmentation on the work content entered by the operator through the laptop through NLP, such as "September 24, 2024", "Zhang San", "8 o'clock in the morning of that day", "turn off", "main circuit breaker", "stick", "warning sign", etc. Then these words are tagged with part of speech, "Zhang San" is a noun, "September 24, 2024", "8 o'clock in the morning of that day" are time adverbials, "turn off" is a verb, "main circuit breaker" is a noun, "warning sign" is a noun phrase, etc. Through word segmentation and part of speech tagging, the system can initially understand the basic structure and components of the work content, which is the basis for ensuring the accuracy of subsequent named body recognition.

[0038] In a specific implementation, information is extracted based on dependency relationships between words and multiple entities to obtain key information.

[0039] In this solution, the system further analyzes the dependency relationship between words to determine which words are subjects, predicates, objects, etc., as well as the modification and subordinate relationships between them. For example, the system recognizes that "turn off" in "turn off the main circuit breaker" is a verb, "main circuit breaker" is a noun, and "main circuit breaker" is the object of "close", and "Zhang San" is the subject of "turn off the main circuit breaker". Through this dependency analysis, the system can understand the operation steps and task descriptions in the job content. In addition, the system will also extract multiple entities involved in the job content, including equipment names, operation actions, time stamps, operators, safety measures, etc. For example, in the sentence "turn off the main circuit breaker and put up a warning sign", the system can identify "main circuit breaker" as the equipment name, "turn off" as the operation action, and "put up a warning sign" as a safety measure. By extracting and analyzing these entities, the system can further understand the specific requirements and steps of the job content, thereby obtaining the key information of the job content corresponding to this job.

[0040] S103: The server obtains the job ticket according to the key information and the information processing model corresponding to the job ticket type.

[0041] In this step, the operation ticket generated by the operator's request focuses on recording and guiding the specific operation steps to ensure the standardization and safety of the operation process. The work ticket focuses on the management and supervision of the operation task to ensure the safety and effective execution of the operation process. Through key information, the operation ticket and the work ticket can be clearly distinguished, thereby achieving effective management and control of different types of operations.

[0042] Therefore, the system uses pre-configured job ticket identification rules based on the key information to determine the job ticket type for this job. The job ticket identification rules include: if the key information includes multiple job steps and the time corresponding to each job step, the job ticket type is determined to be an operation ticket; otherwise, the job ticket type is determined to be a work ticket.

[0043] Specifically, the key information extracted by the system includes multiple specific work steps: "Turn off the main circuit breaker and put up a warning sign", "Check the mechanical status and electrical connection of the circuit breaker", "Perform insulation test on the circuit breaker", "Lubricate the operating mechanism of the circuit breaker", "Restore the power supply of the circuit breaker and conduct a trial run", and the time corresponding to each work step: "8 o'clock in the morning of the same day", "8:15", "8:30", "15 minutes later", "9 o'clock". According to the pre-configured job ticket recognition rules, the system determines that the key information includes multiple work steps and the time corresponding to each work step, so it is determined that the job ticket request sent by the operator is specifically an operation ticket request. By clarifying the type of job ticket, not only the accuracy of job ticket creation is improved, but also the efficiency and accuracy of the operation are further improved, so as to provide detailed records in the event of an accident, which is convenient for traceability and analysis.

[0044] In this solution, after the type of the job ticket is clarified, the server outputs the job ticket in a graphical format based on the key information extracted by the system and the information processing model corresponding to the job ticket type. At this point, the job ticket generated by the operator through the first terminal device request has been formed. The job ticket contains all key information. If the job ticket type is an operation ticket, the key information includes at least: operator, supervisor, operating equipment, multiple operation methods and operation sequence, operation time and safety measures for each operation method. If the job ticket type is a work ticket, the key information includes at least: staff, reviewer, work location, working hours, work methods involved in the work task, and safety measures for each work method. Ensure that all details related to this operation are accurately recorded.

[0045] Specifically, the aforementioned operator sends a request for a work ticket to the server through a laptop computer as the first terminal device, which contains the work content carried in the request for a work ticket. The operator is "Zhang San", the supervisor is "Li Si", the operating device is "main circuit breaker", and multiple operation modes and operation sequences are "turn off the main circuit breaker and put up a warning sign", "check the mechanical state and electrical connection of the circuit breaker", "perform insulation test on the circuit breaker", "lubricate the operating mechanism of the circuit breaker", "restore the power supply of the circuit breaker and conduct a trial run", and the operation times are "8 o'clock in the morning of the same day", "8:15", "8:30", "15 minutes later", "9 o'clock", and the safety measures are "Insulating gloves and goggles must be worn during the whole process, and insulating tools must be used when operating electrical equipment". The server outputs the operation ticket in a graphical format based on the above key information extracted by the system and the information processing model corresponding to the operation ticket.

[0046] S104: The server sends a job ticket review request to the second terminal device, and the job ticket review request carries the job ticket.

[0047] In this step, the reviewer reviews the work ticket through the second terminal device to ensure that all contents comply with operating specifications and safety standards, thereby reducing human errors and operational risks and reducing the possibility of accidents.

[0048] In this solution, based on the job ticket generated by the operator's request, the server packages the generated job ticket and sends a job ticket review request to the second terminal device. The request not only contains the detailed content of the job ticket, but also includes the unique identification of the job ticket, the generation time, the operator information and other necessary metadata for easy review and tracking. After the second terminal device receives the job ticket review request, the relevant reviewer can view the detailed content of the job ticket, check the rationality and completeness of each operation step, ensure that all safety measures are in place, and confirm that the job ticket complies with the company and industry standards and specifications.

[0049] S105: The second terminal device reviews the operation ticket according to the operation ticket review request using the pre-acquired power grid management knowledge graph, which includes multiple operation modes, multiple safety measures, dependencies between different operation modes, and dependencies between safety measures and operation modes.

[0050] In this step, the knowledge graph provides a comprehensive understanding and support for all aspects of power grid management. First, the knowledge graph includes a variety of operating methods, such as routine operations, emergency maintenance, equipment maintenance, system upgrades, etc. There are specific dependencies between different operating methods, that is, each operating method has its specific steps, processes and requirements, which are recorded and associated in detail in the knowledge graph. It should be noted that the operating method includes the operation method or the working method. Through the knowledge graph, the system can quickly find and refer to the best practices and standard operating procedures for different operating methods. Secondly, the knowledge graph also includes a variety of safety measures, such as wearing insulating gloves, using goggles, setting warning signs, and isolating equipment. These safety measures are closely related to specific operating methods to ensure that the relevant safety measures can be correctly applied when performing different types of operations. For example, when overhauling high-voltage equipment, safety measures such as equipment isolation and wearing insulating gloves are required. The knowledge graph provides guidance in job ticket review by recording in detail various operating methods, various safety measures, the dependencies between different operating methods, and the dependencies between safety measures and operating methods, ensuring that all operating contents comply with operating specifications and safety standards, and providing comprehensive, systematic and intelligent support for power grid management.

[0051] Specifically, the work content of arranging Zhang San to perform routine operations on the main circuit breaker of the substation on September 24, 2024, and supervised by Li Si, is reviewed and verified through the relevant content in the knowledge graph. For routine operations, the knowledge graph records standard operating procedures and safety measures, such as shutting down equipment, checking status, testing and lubrication, etc., and "checking the mechanical status and electrical connection of the circuit breaker" should be performed after "turning off the main circuit breaker and posting warning signs", and "restoring the power supply of the circuit breaker and conducting a trial run" should be performed after all inspections and lubrication operations are completed. By comparing the operating steps in the work content with the standard steps in the knowledge graph, the system can verify the rationality and completeness of the work content.

[0052] Secondly, the knowledge graph also includes a variety of safety measures, such as wearing insulating gloves, using goggles, setting up warning signs, isolating equipment, grounding, cutting off power, etc. These safety measures are closely related to specific working methods to ensure that relevant safety measures can be correctly applied when performing different types of operations. For example, in the knowledge graph, the operating steps for "turning off the main circuit breaker and posting warning signs" will be associated with the safety measures of wearing insulating gloves and goggles. By comparing the safety measures in the work content with the standard safety measures in the knowledge graph, ensure that all necessary safety requirements are met.

[0053] S106: After the job ticket is reviewed and approved, the second terminal device sends review confirmation information to the server. The server determines the job ticket as a target job ticket based on the review confirmation information, and sends the target job ticket to the first terminal device.

[0054] If the job ticket is approved, the second terminal device sends an approval confirmation message to the server, and the server determines the job ticket as a target job ticket and sends it to the first terminal device.

[0055] If the job ticket fails the review, the second terminal device returns an error message and / or modification suggestion to the server, and the server forwards it to the first terminal device. The operator receives the error message and / or modification suggestion sent by the server on the first terminal device, and sends a modification request to the server through the first terminal device. The server makes corresponding adjustments based on the feedback and regenerates the job ticket, and sends a job ticket review request to the second terminal device again until the job ticket passes the review. The reviewer sends a review confirmation message to the server through the second terminal device. The server determines the job ticket as the target job ticket based on the review confirmation message and sends it to the first terminal device.

[0056] The graphical automatic ticketing method based on NLP provided in the embodiment of the present application is that the operator sends a job ticketing request to the server through the first terminal device. The server uses NLP technology to parse the job content, extract key information and generate a graphical job ticket, and sends the job ticket to the second terminal device for review. After the review is passed, the job ticket is determined as the target job ticket and returned to the first terminal device. Through the above method, the safety and standardization of the operation process are ensured, providing comprehensive, systematic and intelligent support for power grid management, and significantly improving the quality and safety level of operation management.

[0057] Figure 2 The present invention provides a flowchart of a second embodiment of a graphical automatic ticketing method based on NLP, which can be applied to a power grid operation ticket management system, such as Figure 2 As shown, based on the above embodiment, in step S103, the server obtains the job ticket according to the key information and the information processing model corresponding to the job ticket type, including the following steps: S201: The server inputs each operation mode and corresponding safety measures in the key information into the information processing model corresponding to the operation ticket type for processing, and outputs the standardized operation mode and corresponding safety measures corresponding to the operation ticket type.

[0058] In this step, the operation methods and safety measures in the operation content input by the operator may exist in a colloquial form, such as "turn off the main circuit breaker at 8 o'clock in the morning and put up a warning sign", "check the mechanical state and electrical connection of the circuit breaker", "must wear insulating gloves and goggles", etc. In order to ensure the standardization and standardization of the operation ticket, the server converts these colloquial contents into standardized language expressions based on the principles of clear logic, rigorous structure, formal objectivity, and avoiding brief and casual expressions.

[0059] Specifically, the server uses a pre-configured information processing model to standardize the operation methods and safety measures in the extracted key information. The information processing model contains standard operating procedures and safety measures expression requirements, that is, it must comply with the principles of clear logic, rigorous structure, formal objectivity, and avoidance of brief and casual expression, and can convert unstructured natural language text into structured and standardized work ticket content. For example, "turn off the main circuit breaker and post a warning sign" is mapped to "turn off... and..." in the standard operating procedure, and the conversion result is "turn off the main circuit breaker and post a warning sign around the circuit breaker". At the same time, according to the context or pre-configured information, time and location information is added, such as adding "8 o'clock in the morning of the same day" to the operating steps, that is, "08:00, turn off the main circuit breaker and post a warning sign around the circuit breaker". In addition, the extracted safety measures are matched with the standard safety measures requirements, such as "in the whole process, insulating gloves and goggles must be worn" is mapped to "must... meet the standards..." in the standard safety requirements, and the conversion result is "in the whole process, insulating gloves and goggles that meet the standards must be worn".

[0060] S202: The server writes the standardized operation method and the corresponding safety measures, as well as other information in the key information, into a preset template corresponding to the operation ticket type to obtain an operation ticket.

[0061] In this step, based on the processed key information, the predefined template is used to fill the standardized operation method and safety measures, as well as other information in the operation content into the template, for example, "Zhang San", "main circuit breaker", "close the main circuit breaker at 08:00 and post a warning sign near the circuit breaker", "Li Si" and other information are filled into the corresponding places of the operation ticket template. Then, the filled template is formatted into a structured data format that conforms to the specification, and the result is output in an image to obtain the operation ticket.

[0062] The server outputs standardized operating methods and safety measures corresponding to the job ticket type, and generates job tickets that meet company and industry standards. These tickets not only include detailed job content, but also all necessary safety measures. Standardized job tickets help managers to supervise and review, ensuring that the operating process complies with safety standards and operating specifications, thereby ensuring the safety of operators and the normal operation of equipment.

[0063] The NLP-based graphical automatic ticketing method provided in the embodiment of the present application standardizes the spoken work content input by the operator through the server, converts it into standardized work methods and safety measures that are logically clear, rigorously structured, formal and objective, and fills it into the preset work ticket template to generate a work ticket that meets company and industry standards. Through the above method, the standardization and standardization of the work ticket are ensured, which helps managers to supervise and review, ensure that the work process meets safety standards and operating specifications, thereby ensuring the safety of the operator and the normal operation of the equipment.

[0064] Figure 3 The present invention provides a flowchart of a third embodiment of a graphical automatic ticketing method based on NLP, which can be applied to a power grid operation ticket management system, such as Figure 3 As shown, based on the above embodiment, step 103 specifically includes: S301: For each job ticket type, build an initial model based on the deep learning model.

[0065] In this step, for each job ticket type, an initial model is built based on the Long Short-Term Memory (LSTM) network. LSTM can effectively solve the gradient vanishing and gradient exploding problems that exist when traditional recurrent neural networks process long sequence data.

[0066] In this solution, for job tickets and operation tickets, LSTM is first selected as the basis of the deep learning model. LSTM introduces memory units and gating mechanisms, namely input gates, forget gates, and output gates. The architecture of the initial model includes an input layer, multiple LSTM layers, and an output layer.

[0067] The input layer is used to receive the work content input by the operator and convert it into a word vector representation. Multiple LSTM units extract features and patterns in the text by passing information layer by layer. Each LSTM unit contains a memory unit and a gating mechanism. The input gate determines which part of the input information of the current time step needs to be retained and added to the memory unit; the forget gate determines which part of the memory information of the previous time step needs to be retained; the output gate determines which part of the memory unit information of the current time step needs to be output as the hidden state information of the current time step and passed to the next time step. The output layer converts the features extracted by the LSTM layer into structured work ticket data and generates standardized work steps and safety measures descriptions. When building the initial model, parameter settings are also required, such as learning rate, batch size, number of training rounds, etc.

[0068] Specifically, for the operation ticket, the information of each category is first encoded and converted into a vector representation using word embedding. The input layer receives the encoded sequence data, and the input dimension is expressed as (batch_size, sequence_length, embedding_dim), where batch_size represents the number of samples for each training, which is set to 64; sequence_length represents the length of the input sequence, which depends on the maximum amount of information contained in each operation ticket in the actual data, and is set to 25; embedding_dim represents the dimension of the embedding layer, which is used to convert the trained category information into a continuous vector representation, and is set to 128. The input layer inputs these continuous vector sequences into the LSTM model, where the number of hidden units in the LSTM layer, hidden_units, determines the capacity and expression ability of the model, and is set to 256; the number of layers in the LSTM layer, num_layers, is set to 2, to achieve a balance between capturing the temporal relationship and computational overhead. In each LSTM layer, for the current time step input information "arrange Zhang San to perform routine operations on the main circuit breaker of the substation on September 24, 2024, and Li Si will supervise", the input gate is set to receive input information through the Sigmoid function, and the input gate outputs a value between 0 and 1. Assuming that the input gate outputs 0.8, it means that 80% of the input information will be retained and added to the memory unit. The forget gate receives the hidden state information of the previous time step through the Sigmoid function. Assuming that the forget gate outputs 0.7, it means that 70% of the memory information will be retained. The output gate receives the memory unit state of the current time step through the Sigmoid function. Assuming that the output gate outputs 0.9, it means that 90% of the current memory unit state will be output as the hidden state of the current time step and passed to the next time step. Finally, the fully connected layer maps the output of the LSTM layer to the various fields of the operation ticket, and the output layer outputs the final operation ticket.

[0069] For the work ticket, the sequence_length in the input dimension of the initial model represents the length of the input sequence, which depends on the maximum amount of information contained in each work ticket in the actual data and is set to 20. The remaining parameter settings are consistent with the parameters of the initial model of the operation ticket, and the specific implementation principles and technical effects are similar, so they will not be repeated here.

[0070] The design of the initial model not only takes into account the complexity and diversity of the data, but also finds the optimal balance between computing resources and model performance, thereby achieving efficient and accurate job ticket generation.

[0071] S302: Obtain a training sample set corresponding to the job ticket type, the training sample set including: multiple samples, each sample including a description of a first working method and safety measures input by an operator, and a manually labeled description of a standardized second working method and safety measures corresponding to the first working method and safety measures.

[0072] In this step, in order to generate standardized work ticket data corresponding to the work ticket type, the training sample set corresponding to the work ticket type must first be obtained. For the description of the standardized second work mode and safety measures corresponding to the first work mode and safety measures manually labeled, a large amount of work ticket actual text data needs to be collected. These data are usually entered by operators in daily work and stored in the database, describing the specific work mode and corresponding safety measures, including various work types and work modes, such as power equipment maintenance, mechanical equipment maintenance, chemical equipment operation, etc.

[0073] In this scheme, a two-person annotation method through the LightTag platform is adopted, that is, two annotators independently annotate through the LightTag platform, and then compare and merge the annotation results. Specifically, the annotators need to identify the key content in the text data. For example, for the input "arrange Zhang San to perform routine operation on the main circuit breaker of the substation on September 24, 2024, and Li Si will supervise", the annotators convert it into a structured description based on its key information and industry standards and specifications. For example, "Operator: "Zhang San", "Operation time: September 24, 2024", "Operation equipment: routine operation", "Supervisor: Li Si". During the annotation process, the annotators should also pay attention to the text context. For example, some operating methods may need to be understood and converted in combination with the context before and after to ensure that the generated standardized description is complete and accurate. The annotators need to carefully review and verify each operation step and safety measure to ensure that it complies with industry standards and safety specifications.

[0074] S303: Based on the training sample set, the initial model is trained to obtain an information processing model.

[0075] In this step, the initial model is trained based on the aforementioned training sample set to obtain an information processing model that can effectively process and generate standardized work ticket data. The architecture of the initial model includes an input layer, multiple LSTM layers, and an output layer. During the training process, the LSTM model learns through a large number of training samples. Each training sample contains an input sequence and a corresponding target output. The input sequence is the natural language text entered by the operator, and the target output is a manually annotated standardized description. Through the back-propagation algorithm, the model continuously adjusts its parameters, including the weights of the input gate, forget gate, and output gate, to minimize the prediction error. The model's loss function is used to measure the difference between the model's predicted output and the actual target output. By minimizing the loss function, the model learns which information is important at different time steps and which information can be ignored.

[0076] In a specific implementation, during the process of training the initial model according to the training sample set, the model training parameters are optimized until a preset convergence function converges to obtain an information processing model.

[0077] Specifically, the Adam optimization algorithm is used for the operation ticket and work ticket information processing model to ensure that the model performs well when processing complex time series data and diverse work ticket content. Initialize the parameters of the Adam optimization algorithm, including the learning rate learning_rate, the decay rate beta_1 of the first-order moment estimate, the decay rate beta_2 of the second-order moment estimate, and a small constant epsilon to prevent zero division errors. The default parameter settings are: learning_rate=0.001, =0.9, beta_2=0.999 and epsilon=10 -8 .

[0078] In each training step, the gradient of the loss function with respect to the model parameters is first calculated. For the LSTM model, the gradient calculation involves the backpropagation algorithm to handle the dependencies of the time series data. Specifically, the gradient calculation traverses the entire sequence data, gradually calculates the gradient of each time step, and accumulates it. Then, the first-order moment estimate (momentum) and the second-order moment estimate (exponentially decayed average of the square of the gradient) of the gradient are calculated. The formula for the first-order moment estimate is: t = beta_1×m t-1 +(1- beta_1) ×g t , where m t represents the momentum at the current moment, beta_1 represents the momentum hyperparameter, that is, the decay rate of the first-order moment estimate, g t Represents the gradient at the current moment. The formula for second-order moment estimation is: , where v trepresents the second-order moment estimate at the current moment, beta_2 represents the second-order moment hyperparameter, that is, the decay rate of the second-order moment estimate, g t Represents the gradient at the current moment. These two estimates represent the exponentially decaying averages of the first-order and second-order moments of the gradient, respectively, and can capture the momentum and changing trend of the gradient. Next, in order to solve the problem that the first-order and second-order moment estimates at the initial moment may be zero, the Adam algorithm also introduces a bias correction mechanism. The corrected momentum and second-order estimates are calculated as follows: in, and They represent the corrected first-order moment and second-order moment estimates respectively, and t represents the current number of iterations.

[0079] Finally, the corrected moment estimates are used to update the model parameters. The parameter update formula is: Among them, θ t Represents the updated parameters in step t.

[0080] The Adam algorithm combines the ideas of momentum and RMSprop optimization algorithms to adaptively adjust the model's trainable parameters during training, including the weight matrix and bias vector of the input gate, forget gate, output gate, and candidate memory units, as well as the parameters of the fully connected layer and embedding layer, thereby accelerating convergence and reducing oscillations during training until the preset loss function converges. This ensures that the model performs well when processing complex time series data and diverse job ticket content. The specific parameter settings can be fine-tuned based on the actual data and model performance, but the above default parameter settings have provided good training results in actual experiments.

[0081] Specifically, the model continuously adjusts the weights of the input gate, forget gate, and output gate through the back-propagation algorithm during the training process. For example, at a certain time step, the input gate receives input information through the Sigmoid function, and the input gate outputs a value between 0 and 1. Assuming that the input gate outputs 0.9, this means that 90% of the current input information will be retained. However, the specific value of 0.9 is determined by the optimization algorithm and loss function during the training process. The model will gradually learn the importance of "Zhang San" in the task of generating job tickets, thereby retaining this key information in most cases, ensuring that the model can focus on the most relevant and critical pieces of information.

[0082] The graphical automatic ticketing method based on NLP provided in the embodiment of the present application ensures the accuracy and diversity of the data by obtaining a set of training samples corresponding to the job ticket type and manually annotating them. The memory unit and gating mechanism of the LSTM model are used to effectively process long sequence data and capture long-term dependencies. The model parameters are continuously optimized through the back propagation algorithm so that it can learn key information and generate structured job ticket data. This method improves the generalization ability and performance of the model through the stacking of multiple layers of LSTM and the use of pre-trained word vectors, ensuring that the generated job ticket content is accurate and standardized, and meets industry standards and safety specifications.

[0083] Figure 4 The present invention provides a flowchart of a fourth embodiment of a graphical automatic ticketing method based on NLP, which can be applied to a power grid operation ticket management system, such as Figure 4 As shown, based on the above embodiment, in step S105, the second terminal device uses the pre-acquired power grid management knowledge graph to review the operation ticket according to the operation ticket review request. The power grid management knowledge graph includes multiple operation modes, multiple safety measures, dependencies between different operation modes, and dependencies between safety measures and operation modes. The review process specifically includes: S401: Obtain personnel information from the job ticket, and verify whether each person in the personnel information has the corresponding job authority according to the job authority of each person stored in the database.

[0084] In this step, after obtaining the work ticket, the server sends a work ticket review request to the second terminal device, which contains complete work ticket information. The second terminal device uses the pre-acquired power grid management knowledge graph to conduct a comprehensive review of the work ticket. In the first step of the review, all relevant personnel information is extracted from the work ticket, including operators and supervisors, etc. The extracted personnel information will be compared and reviewed with the work permissions of each person stored in the database. The database records the work permissions of each person in detail, including but not limited to the operation permissions of specific equipment, the execution permissions of specific types of operations, and other relevant safety qualifications and certifications.

[0085] Specifically, the system will check whether each person in the job ticket has the authority to perform the job. For example, for operator Zhang San, the system will check whether he has the authority to operate the main circuit breaker; for supervisor Li Si, the system will check whether he has the qualifications to supervise this type of operation. If a person does not have the corresponding authority, the system will automatically mark and issue a warning, and give suggestions for modifying the job ticket, that is, prompting the need to replace the person with the corresponding authority or upgrade the authority. Effectively ensure that the personnel in each operation link have the corresponding qualifications and authority, thereby improving the safety and reliability of the operation.

[0086] S402: Determine whether the target operation mode and target safety measures included in the operation ticket conform to the dependencies in the power grid management knowledge graph.

[0087] Figure 5 A schematic diagram of a specific implementation of a graphical automatic ticketing method based on NLP provided in an embodiment of the present application. In this step, after reviewing the authority of relevant personnel, according to the following Figure 5 The power grid knowledge graph dependency indicator diagram shown determines whether the target operation mode and target safety measures included in the operation ticket conform to the dependency relationship shown in the figure. It should be noted that the power grid knowledge graph dependency indicator diagram is only exemplary and does not represent the only one. This application does not specifically limit the various operation modes and safety measures involved and the dependency relationship between the two.

[0088] In constructing the knowledge graph of power grid management, data screening is first performed, including structured data and unstructured data. Structured data includes operation data, maintenance records, dispatch plans, etc. of power grid equipment, which are usually stored in databases; unstructured data includes text files such as operation manuals, fault reports, and technical documents. In order to ensure data quality, data cleaning, noise data removal, data veracity, and duplicate data processing are required. Next, the dependencies are determined. Power grid operations involve a variety of entities and relationships, such as entities or equipment such as substations, main circuit breakers, and power plants, routine operations, equipment maintenance, and system updates. Based on a large amount of structured data and a variety of unstructured data in the database, data dependencies are determined by combining manual verification and automatic verification methods. Then, according to the operation mode, safety measures, dependencies between different operation modes, and dependencies between safety measures and operation modes, a knowledge graph is constructed through the graph database Neo4j. The operation mode and safety measures are stored as triples (entity-relationship-entity) in the resource description framework RDF format. Its basic structure consists of nodes and edges, where nodes represent entities and edges represent relationships between entities. In the knowledge graph of power grid operation, each operation mode and safety measure is a node. The constructed knowledge graph is regularly updated through the incremental update method to reflect the latest power grid operation status and maintenance records. Through the above method, a comprehensive and dynamic power grid operation knowledge graph is finally constructed, providing strong support for the efficient operation and intelligent dispatch of the power grid.

[0089] In the knowledge graph of power grid management, various operation modes include but are not limited to routine operation, equipment maintenance, equipment overhaul, etc. Each operation mode has its standard operation steps and processes. For example, routine operation of a circuit breaker may include closing the circuit breaker, checking the mechanical status and electrical connection, performing insulation test, lubricating the operating mechanism, restoring the power supply and conducting a trial run, etc. Before conducting insulation test of the circuit breaker, its mechanical status and electrical connection must be checked first; before restoring the power supply of the circuit breaker, all inspections and lubrication work must be completed first. Various safety measures include wearing personal protective equipment, using insulating tools, posting warning signs, setting up safety isolation areas, conducting safety training, etc. The dependency between safety measures and operation modes stipulates that each operation mode requires specific safety measures to ensure its safe operation. For example, during insulation test, operators must wear insulating gloves and goggles to prevent other accidental injuries such as electric shock and splashing.

[0090] Specifically, the system checks each operation method and safety measure in the work ticket one by one to ensure that they comply with the dependencies and standard requirements in the knowledge graph. If there are any non-compliance, the system will automatically mark and issue a warning, and give suggestions for modifying the work ticket. For example, if the work ticket does not mention checking the mechanical status and electrical connections before the insulation test, the system will prompt that this inspection step needs to be added. Through this detailed comparison and review, the system can ensure that each operation method and safety measure in the work ticket complies with the standards and specifications of power grid management, thereby improving the safety and reliability of the operation.

[0091] S403: If each person in the personnel information in the operation ticket has the corresponding operation authority, and the target operation mode and target safety measures conform to the dependency relationship in the power grid management knowledge graph, the operation ticket is approved; otherwise, the operation ticket is not approved.

[0092] In this step, if each person involved in the job ticket has the corresponding job authority, and the operation method and target safety measures comply with the dependencies in the power grid management knowledge graph, the job ticket is reviewed and approved, and the second terminal device sends a review confirmation message to the server. The server determines the job ticket as a target job ticket based on the review confirmation information, and sends the target job ticket to the first terminal device.

[0093] If any non-compliance is found during the review process, such as a person does not have the corresponding work authority, or the target operation method or safety measures do not conform to the dependencies in the knowledge graph, the job ticket will fail the review. The second terminal device sends an error message and / or modification suggestions to the server, and the server sends the error message and / or modification suggestions to the first terminal device. The operator receives the error message and / or modification suggestions sent by the server at the first terminal device, and sends a modification request to the server through the first terminal device. The server makes corresponding adjustments based on the feedback and regenerates the job ticket, and sends a job ticket review request to the second terminal device again until the job ticket passes the review. The second terminal device sends a review confirmation message to the server. The server determines the job ticket as the target job ticket based on the review confirmation information and sends it to the first terminal device.

[0094] In the NLP-based image automatic ticketing method provided in the embodiment of the present application, the server sends a job ticket review request to the second terminal device, and the second terminal device uses the power grid management knowledge graph to conduct a comprehensive review of the job ticket, including extracting the personnel information in the job ticket and comparing it with the job permissions in the database, and checking whether the target operation method and safety measures in the job ticket meet the dependencies and standard requirements in the power grid management knowledge graph. If all conditions are met, the job ticket is reviewed and approved, otherwise, the job ticket is not reviewed and the system issues a warning and provides modification suggestions until the job ticket meets all requirements. Through the above method, this multi-level and detailed review process effectively improves the safety and reliability of the operation, ensuring that each operation link meets the standards and specifications of power grid management.

[0095] Figure 6 A schematic diagram of an application scenario of a graphical automatic ticketing method based on NLP provided in an embodiment of the present application, that is, a schematic diagram of the system structure of the aforementioned method applied to a power grid operation ticket management system, such as Figure 6 As shown, the application scenario includes at least: at least two terminal devices and a server.

[0096] In response to the operator inputting the operation content related to this operation, the first terminal device sends a request for the operation ticket to the server. The server first uses NLP to process the operation content to obtain the key information corresponding to this operation, and then obtains the operation ticket based on the key information and the information processing model corresponding to the operation ticket type, and sends a request for the operation ticket review to the second terminal device. The second terminal device reviews the operation ticket according to the operation ticket review request, and after the review is passed, sends the review confirmation information to the server. The server determines the operation ticket as the target operation ticket based on the review confirmation information and sends it to the first terminal device.

[0097] The first terminal device and the second terminal device can be communication devices such as smart phones, tablet computers, local personal computers, etc. that have data processing capabilities such as network connection, user interface interaction, compatibility and scalability; the server can be a cloud server, an internal server, a hybrid server, etc. that has data processing capabilities such as computing, storage, high availability and fault tolerance.

[0098] The physical devices involved in the above description are exemplary in the figures and do not represent the only ones. This application does not specifically limit the specific form and type of the specific devices involved.

[0099] In this solution, the power grid operation ticket management system 600 includes: The first terminal device 601 is used for: Send a job ticket request to the server, which carries the job content related to this job entered by the operator.

[0100] Server 602, used for: According to the job ticket request, the natural language processing model NLP is used to process the job content to obtain the key information corresponding to this job; Input the key information into the generation model corresponding to the job ticket type to generate the job ticket.

[0101] The second terminal device 603 is used for: According to the work ticket review request, the work ticket is reviewed using the pre-acquired power grid management knowledge graph, which includes multiple work modes, multiple safety measures, dependencies between different work modes, and dependencies between safety measures and work modes; After the job ticket is reviewed and approved, the review confirmation information is sent to the server.

[0102] Server 602 is also used to: Sending a job ticket review request to the second terminal device, wherein the job ticket review request carries the job ticket; According to the audit confirmation information, the job ticket is determined as a target job ticket, and the target job ticket is sent to the first terminal device.

[0103] The power grid operation ticket management system provided in the embodiment of the present application is used to execute the technical solution in the aforementioned method embodiment, and its implementation principle and technical effect are similar and will not be repeated here.

[0104] In a specific implementation, the server 602 is specifically used for: Based on the key information, the pre-configured job ticket identification rules are used to determine the job ticket type for this job; The job ticket identification rules include: if the key information includes multiple job steps and the time corresponding to each job step, the job ticket type is determined to be an operation ticket, otherwise, the job ticket type is determined to be a work ticket; If the work ticket type is an operation ticket, the key information includes: operator, supervisor, operating equipment, multiple operation methods and operation sequence, operation time of each operation method and safety measures; If the work ticket type is a work ticket, key information includes: staff, reviewer, work location, work time, work methods involved in the work task, and safety measures for each work method; Among them, the operation mode includes the operation mode or the working mode.

[0105] In a specific implementation, the server 602 is specifically used for: Through NLP, the homework content is parsed to obtain the dependency relationship between the words in the homework content and the multiple entities in the homework ticket involved in the homework content; Extract information based on the dependency relationship between words and multiple entities to obtain key information; Input each operation mode and corresponding safety measures in the key information into the information processing model corresponding to the operation ticket type for processing, and output the standardized operation mode and corresponding safety measures corresponding to the operation ticket type; The standardized working method and corresponding safety measures, as well as other key information, are written into a preset template corresponding to the type of work ticket to obtain a work ticket.

[0106] In a specific implementation, the server 602 is specifically used for: For each job ticket type, an initial model is built based on the deep learning model; Acquire a training sample set corresponding to the work ticket type, the training sample set including: a plurality of samples, each sample including a first work mode and a description of safety measures input by a worker, and a manually annotated standardized second work mode and a description of safety measures corresponding to the first work mode and safety measures; Based on the training sample set, the initial model is trained to obtain the information processing model.

[0107] In a specific implementation, the server 602 is further configured to: In the process of training the initial model according to the training sample set, the model training parameters are optimized until the preset convergence function converges to obtain the information processing model.

[0108] In a specific implementation manner, the second terminal device 603 is specifically used for: Obtain personnel information from the job ticket, and determine whether each person in the personnel information has the corresponding job authority based on the job authority of each person stored in the database; Determine whether the target operation mode and target safety measures included in the operation ticket conform to the dependencies in the power grid management knowledge graph; If each person in the personnel information in the operation ticket has the corresponding operation authority, and the target operation mode and target safety measures are consistent with the dependency relationship in the power grid management knowledge graph, the operation ticket is approved; otherwise, the operation ticket is not approved; If the job ticket fails the review, the second terminal device returns an error message and / or modification suggestions to the server.

[0109] The server 602 sends the error information and / or modification suggestion to the first terminal device 601 .

[0110] The power grid operation ticket management system provided in any of the above embodiments is used to execute the technical solution in the above method embodiments, and its implementation principle and technical effect are similar and will not be repeated here.

[0111] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0112] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A graphical automatic ticketing method based on NLP, characterized in that: Applied to a power grid operation ticket management system, the power grid operation ticket management system includes a server, a first terminal device and a second terminal device, and the method includes: The first terminal device sends a job ticket generation request to the server, wherein the job ticket generation request carries the job content related to the current job input by the operator; The server processes the job content according to the job ticket creation request using a natural language processing model NLP to obtain key information corresponding to the job; The server obtains the job ticket according to the key information and the information processing model corresponding to the job ticket type; The server sends a job ticket review request to the second terminal device, wherein the job ticket review request carries the job ticket; The second terminal device reviews the operation ticket according to the operation ticket review request using a pre-acquired power grid management knowledge graph, wherein the power grid management knowledge graph includes multiple operation modes, multiple safety measures, dependencies between different operation modes, and dependencies between safety measures and operation modes; After the job ticket is reviewed and approved, the second terminal device sends review confirmation information to the server. The server determines the job ticket as a target job ticket based on the review confirmation information and sends the target job ticket to the first terminal device.

2. The method according to claim 1, characterized in that The method further comprises: According to the key information, the type of the job ticket for this job is determined by using a pre-configured job ticket identification rule; Among them, the job ticket identification rule includes: if the key information includes multiple job steps and the time corresponding to each job step, then the job ticket type is determined to be an operation ticket; otherwise, the job ticket type is determined to be a work ticket.

3. The method according to claim 1, characterized in that The natural language processing model NLP is used to process the job content to obtain key information corresponding to the job, including: By using NLP, the operation content is parsed to obtain the dependency relationship between the words in the operation content and the multiple entities in the operation ticket involved in the operation content; Information is extracted according to the dependency relationship between the words and the multiple entities to obtain the key information.

4. The method according to claim 3, characterized in that: If the type of the work ticket is an operation ticket, the key information includes: operator, supervisor, operating equipment, multiple operation modes and operation sequence, operation time of each operation mode and safety measures; If the type of the work ticket is a work ticket, the key information includes: staff, reviewer, work location, work time, work methods involved in the work task, and safety measures for each work method; Among them, the operation mode includes the operation mode or the working mode.

5. The method according to any one of claims 1 to 4, characterized in that: The server obtains the job ticket according to the key information and the information processing model corresponding to the job ticket type, including: The server inputs each operation mode and corresponding safety measures in the key information into the information processing model corresponding to the operation ticket type for processing, and outputs the standardized operation mode and corresponding safety measures corresponding to the operation ticket type; The server writes the standardized operation method and the corresponding safety measures, as well as other information in the key information, into a preset template corresponding to the operation ticket type to obtain the operation ticket.

6. The method according to claim 5, characterized in that The method further comprises: For each job ticket type, an initial model is built based on the deep learning model; Acquire a training sample set corresponding to the work ticket type, the training sample set comprising: a plurality of samples, each sample comprising a first work mode and a description of safety measures input by a worker, and a manually annotated standardized second work mode and a description of safety measures corresponding to the first work mode and safety measures; Based on the training sample set, the initial model is trained to obtain the information processing model.

7. The method according to claim 6, characterized in that The step of training the initial model based on the training sample set to obtain the information processing model includes: In the process of training the initial model according to the training sample set, the model training parameters are optimized until a preset convergence function converges to obtain the information processing model.

8. The method according to any one of claims 1 to 4, characterized in that: The use of the pre-acquired power grid management knowledge graph to review the operation ticket includes: Obtaining personnel information from the work ticket, and reviewing whether each person in the personnel information has the corresponding work authority according to the work authority of each person stored in the database; Determine whether the target operation mode and target safety measures included in the operation ticket conform to the dependency relationship in the power grid management knowledge graph; If each person in the personnel information in the operation ticket has the corresponding operation authority, and the target operation mode and the target safety measures both conform to the dependency relationship in the power grid management knowledge graph, the operation ticket is approved; Otherwise, the job ticket will not pass the review.

9. The method according to claim 7, characterized in that: The method further comprises: If the job ticket fails to pass the review, the second terminal device returns error information and / or modification suggestions to the server; The server sends the error information and / or the modification suggestion to the first terminal device.

10. A power grid operation ticket management system, characterized in that: include: A server, a first terminal device and a second terminal device, wherein: The first terminal device is used to send a job ticket generation request to the server, wherein the job ticket generation request carries the job content related to the current job input by the operator; The server is used to: According to the job ticket creation request, the job content is processed using a natural language processing model NLP to obtain key information corresponding to the job; Input the key information into a generation model corresponding to the job ticket type to generate a job ticket; Sending a job ticket review request to the second terminal device, wherein the job ticket review request carries the job ticket; The second terminal device is used for: According to the operation ticket review request, the operation ticket is reviewed using a pre-acquired power grid management knowledge graph, wherein the power grid management knowledge graph includes multiple operation modes, multiple safety measures, dependencies between different operation modes, and dependencies between safety measures and operation modes; After the job ticket is reviewed and approved, a review confirmation message is sent to the server; The server is further configured to determine the job ticket as a target job ticket based on the audit confirmation information, and send the target job ticket to the first terminal device.

Images