Intelligent processing method, system and equipment for power grid regulation and control standard file and medium

Abstract

The invention discloses a power grid regulation and control standard file intelligent processing method, system and device and a medium, and the method comprises the steps: analyzing a power grid regulation and control standard file through a natural language processing model, and extracting a target file parameter, so as to generate preliminary structured text data; unstructured content in the scanned power grid regulation and control standard file is converted into structured supplementary content; performing cross-modal alignment on the structured supplementary content and the preliminary structured text data to generate a structured task list, and binding each task unit in the task list with corresponding file information in the power grid regulation and control standard file; analyzing co-occurrence frequency and dependency relationship among tasks in the task list by using a label graph neural network, dynamically generating a corresponding task label, and counting multi-dimensional data of the same label based on the task label; and comparing semantic differences between the power grid regulation and control standard files of different versions through a multi-modal comparison algorithm, generating a visual difference report for performing abnormity reminding, and labeling corresponding revision suggestions.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, system, device and medium for intelligent processing of power grid regulation standard files. Background Art

[0002] With the rapid development of the power grid, the increasing scale of renewable energy integration, and the accelerating advancement of power grid technology, the number of grid regulations and standard documents has increased, and these documents are frequently updated, resulting in a massive accumulation of standard documents. In the field of power grid regulation, standard documents carry critical information such as dispatch instructions, technical regulations, and equipment parameters. Their efficient parsing and management directly impact the reliability of power grid operations and the efficiency of emergency response. Currently, the processing of such documents mostly relies on general document management systems. These systems can generally perform text extraction and simple classification, but they have significant limitations in practical applications.

[0003] Existing technologies are insufficiently capable of parsing unstructured content. Technical parameters between table content and text paragraphs often become disconnected due to semantic conflicts, requiring manual intervention and correction, resulting in inefficient structuring. Furthermore, generated task lists are mostly static data, lacking the ability to dynamically bind to file versions and business rules. Dependencies and progress deviations between tasks cannot be mapped to scheduling decisions in real time, resulting in a formalized task tracking process that struggles to support precise business collaboration.

[0004] Traditional version management methods rely heavily on static tags and manual comparison, making it difficult to automatically identify differences in technical standards between different versions of a document. Key changes in revision records must be manually reviewed one by one, making it easy to miss technical conflicts and unable to generate actionable revision suggestions. Summary of the Invention

[0005] The embodiments of the present application provide a method, system, device and medium for intelligent processing of power grid control standard files to solve the above-mentioned technical problems.

[0006] On the one hand, an embodiment of the present application provides a method for intelligently processing a power grid regulation standard file, including:

[0007] Parsing the grid regulation standard document using a natural language processing model, extracting target file parameters to generate preliminary structured text data, and converting the scanned unstructured content in the grid regulation standard document into structured supplementary content; the unstructured content includes embedded scanned tables and handwritten annotations;

[0008] Cross-modally aligning the structured supplementary content with the preliminary structured text data to generate a structured task list, and binding each task unit in the task list to corresponding file information in the power grid regulation standard file; the file information includes a file version number and revision time;

[0009] Utilizing a label graph neural network to analyze the co-occurrence frequency and dependency relationships between tasks in the task list, dynamically generating corresponding task labels, and statistically analyzing multi-dimensional data with the same labels based on the task labels;

[0010] Through the multimodal comparison algorithm, the semantic differences between different versions of power grid regulation standard documents are compared, and a visual difference report is generated to provide abnormal reminders and mark the corresponding revision suggestions.

[0011] In one implementation of the present application, the target file parameters include paragraph titles, entity information, and technical indicator parameters;

[0012] Cross-modally aligning the structured supplementary content with the preliminary structured text data to generate a structured task list, specifically including:

[0013] Performing a topological structure analysis on the row and column data in the scan table to identify a corresponding relationship between a table header and a data unit in the scan table;

[0014] Mapping the row and column data with technical indicator parameters in the preliminary structured text data by calculating semantic similarity to determine conflicting content between the structured supplementary content and the preliminary structured text data, and marking the conflicting content;

[0015] Using a pre-trained deep learning model, the conflicting content is disambiguated and the corresponding disambiguation path is recorded for version comparison.

[0016] The entities in the entity information are associated with the preset power grid dispatching business rules to generate a corresponding task list; the task list includes the task name, responsible unit, completion time and key indicator threshold.

[0017] In one implementation of the present application, a label graph neural network is used to analyze the co-occurrence frequency and dependency relationship between tasks in the task list and dynamically generate corresponding task labels, specifically including:

[0018] When the difference between the task completion time and the current system time is less than the preset threshold, the corresponding work task is marked with a deadline tag and a countdown reminder is pushed to the mobile terminal;

[0019] When the difference between the actual value of a task's key technical indicator and the threshold exceeds the preset tolerance range, the corresponding work task will be marked with an indicator abnormality label, and the historical handling plan and revision record of similar tasks will be linked;

[0020] Based on the user's manual operations in the visual interface, a manual intervention tag is added to the corresponding work task, and the corresponding operation log and impact analysis report are recorded; the manual operation includes adjusting the task progress or responsible unit.

[0021] In one implementation of the present application, after analyzing the co-occurrence frequencies and dependencies between tasks in the task list using a label graph neural network and dynamically generating corresponding task labels, the method further includes:

[0022] Based on the user's operation trigger on the task tag, feedback data of the tag operation is collected, and based on the feedback data, the tag data confirmed by the user and the tag data modified or deleted are extracted;

[0023] The label data confirmed by the user is used as a positive sample, and the modified or deleted label data is used as a negative sample to construct a training set;

[0024] By using a contrastive learning algorithm, the semantic differences between the structured task list and the user operation are calculated to adjust the entity recognition priority of the natural language processing model;

[0025] Multimodal feature consistency verification is performed on the unlabeled task data, pseudo labels are generated for the unlabeled task data that pass the verification, and the unlabeled task data corresponding to the pseudo labels are added to the training set to expand the training set and optimize the natural language processing model's ability to parse power professional terminology.

[0026] In one implementation of this application, a multimodal comparison algorithm is used to compare the semantic differences between different versions of power grid regulation standard documents, generate a visual difference report to provide abnormal reminders, and annotate the corresponding revision suggestions, specifically including:

[0027] Parse the revision records of different versions of grid regulation standard documents to extract the changed paragraph content and technical indicator parameters; the changes include addition, deletion or modification;

[0028] Using a semantic similarity algorithm, we calculate the differences between the changed paragraphs, generate a corresponding difference matrix, and mark any technical standard conflict points.

[0029] Mapping the difference matrix to a visual interface to display the difference analysis results in a highlighted comparison format; the difference analysis results include the impact scope of the version change and compatibility recommendations;

[0030] A visual difference report is generated based on the difference analysis results, abnormal reminders are given for version differences, and corresponding revision suggestions are marked in the visual difference report; the revision suggestions include unified suggestions for power professional terminology and linkage adjustment strategies for technical indicator parameters.

[0031] In one implementation of the present application, the present invention further includes:

[0032] Based on the test requirement trigger, the system extracts frequently accessed file tags and task types from the user's learning records. Based on the attention mechanism, it extracts the procedural clauses associated with the current task from the grid regulation standard file corresponding to the file tags to generate the current test paper. The question types in the current test paper include single-choice questions, multiple-choice questions, true-or-false questions, and fill-in-the-blank questions.

[0033] During the exam, the user's answer track is monitored in real time, and the test questions that are potential weak points are marked by combining the answer track and the answer time of each question;

[0034] Identify the user's answers in the submitted examination papers, determine whether the user's answers are correct based on the extracted rules and regulations, and score the user's answers;

[0035] The user's overall score is ranked, and the exam questions that are easy to make mistakes are counted to form a wrong question set, so as to adjust the difficulty distribution of the questions in the exam paper according to the easy to make mistakes in the wrong question set.

[0036] In one implementation of the present application, after analyzing the co-occurrence frequencies and dependencies between tasks in the task list using a label graph neural network and dynamically generating corresponding task labels, the method further includes:

[0037] When a user deletes a task unit, it is automatically marked as discarded, the task unit is unbound from the file information, and the reason for deletion is recorded in the preset feedback database;

[0038] When a user edits a task unit, the system monitors changes in key fields in the modified content, marks the corresponding task unit with a pending review tag, and generates a revision comparison snapshot.

[0039] When the task progress deviates from the preset timeline, a deviation warning is triggered, and the corresponding deviation situation is associated with the administrator's mobile notification;

[0040] Based on the comparison between the task completion rate and the indicator threshold, overdue tasks are marked as abnormal and the priority reallocation logic is activated.

[0041] On the other hand, the embodiment of the present application further provides a system for intelligent processing of power grid regulation standard files, the device comprising: a file parsing module, a task list generation module, a task label generation module, and a difference comparison module;

[0042] The file parsing module is configured to parse the grid regulation standard file using a natural language processing model, extract target file parameters to generate preliminary structured text data, and convert unstructured content in the scanned grid regulation standard file into structured supplementary content; the unstructured content includes embedded scanned tables and handwritten annotations;

[0043] The task list generation module is configured to perform cross-modal alignment between the structured supplementary content and the preliminary structured text data to generate a structured task list, and bind each task unit in the task list to corresponding file information in the power grid regulation standard file; the file information includes a file version number and revision time;

[0044] The task label generation module is used to analyze the co-occurrence frequency and dependency relationship between tasks in the task list using a label graph neural network, dynamically generate corresponding task labels, and based on the task labels, count multi-dimensional data with the same label;

[0045] The difference comparison module is used to compare the semantic differences between different versions of power grid regulation standard documents through a multimodal comparison algorithm, generate a visual difference report to provide abnormal reminders, and mark corresponding revision suggestions.

[0046] On the other hand, an embodiment of the present application further provides a device for intelligently processing power grid regulation standard files, the device comprising:

[0047] at least one processor;

[0048] and, a memory communicatively coupled to the at least one processor;

[0049] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned method for intelligent processing of power grid regulation standard files.

[0050] On the other hand, an embodiment of the present application further provides a non-volatile computer storage medium storing computer executable instructions, which, when executed, implements the above-mentioned method for intelligent processing of power grid regulation standard files.

[0051] The embodiments of the present application provide a method, system, device, and medium for intelligently processing power grid regulation standard files, which have at least the following beneficial effects:

[0052] The natural language processing model can be used to parse preliminary structured text data in standard documents for power grid regulation, automatically identify structured data in scanned tables and handwritten annotations, and eliminate semantic conflicts between text paragraphs and technical indicator parameters through cross-modal alignment technology, reducing the need for manual corrections and improving the efficiency and accuracy of file parsing. Task units are dynamically bound to file version numbers and revision times, and the dependency relationship between tasks is analyzed using a label graph neural network to achieve real-time updates and multi-dimensional statistics of task status, thereby enhancing the refined management capabilities of scheduling decisions. Multimodal comparison algorithms are used to automatically detect differences in technical standards between different versions of files, generate highlighted comparison visual reports and revision suggestions, avoid manual verification and omission of key change points, and improve the integrity and timeliness of version management. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0054] Figure 1 A flowchart of a method for intelligently processing power grid control standard files provided in an embodiment of the present application;

[0055] Figure 2 A schematic diagram of the internal structure of a system for intelligently processing power grid control standard files provided in an embodiment of the present application;

[0056] Figure 3 A schematic diagram of the internal structure of a power grid regulation standard file intelligent processing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0057] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0058] The following describes in detail the technical solutions provided by various embodiments of the present application in conjunction with the accompanying drawings.

[0059] Figure 1 A flowchart of a method for intelligently processing power grid regulation standard files provided in an embodiment of the present application.

[0060] The analysis method involved in the embodiments of the present application can be implemented by a terminal device or a server, and the present application does not impose any special restrictions on this. For ease of understanding and description, the following embodiments are described in detail using a server as an example.

[0061] It should be noted that the server can be a single device or a system composed of multiple devices, that is, a distributed server, and this application does not make any specific restrictions on this.

[0062] like Figure 1 As shown, an embodiment of the present application provides a method for intelligently processing a power grid control standard file, including:

[0063] Step 101: parse the grid regulation standard file through a natural language processing model, extract target file parameters to generate preliminary structured text data, and convert the unstructured content in the scanned grid regulation standard file into structured supplementary content.

[0064] It should be noted that the unstructured content in the embodiment of the present application includes embedded scanned forms and handwritten annotation content.

[0065] In one embodiment of the present application, the target file parameters include paragraph titles, entity information, and technical indicator parameters. For example, a natural language processing model is first used to perform semantic segmentation on the text paragraphs in the input power grid regulation standard document. This model identifies paragraph titles such as relay protection regulations and load adjustment requirements, extracts entity information such as equipment numbers (e.g., #3 main transformer) and dispatch instructions (e.g., load-limited operation), and extracts technical indicator parameters from the document (e.g., voltage fluctuation range ±5%), thereby generating preliminary structured text data.

[0066] It is understandable that the unstructured content in the scanned power grid regulation standard file needs to be specially processed. Specifically, the unstructured content in the power grid regulation standard file, such as the tables and handwritten annotations embedded in the file, is parsed by improved optical character recognition technology (OCR). First, the table is topologically analyzed to identify the hierarchical relationship between the table header and the data unit. For example, in the line load limit table, the data of the column of the line name and the column of the maximum load rate are automatically associated to ensure that the row and column correspondence is accurate. For handwritten annotations, they are converted into standard text by tracking the stroke trajectory and matching the context semantics. For example, the handwritten annotation #3 main transformer overload is parsed into structured supplementary content and associated with the dispatching rules of the main transformer equipment.

[0067] In one embodiment of this application, access and storage of electronic versions of standard documents for power grid regulation are implemented. An archive metadata repository for standard documents for power grid regulation is constructed, and configuration management is performed on the storage location, access method, and relationships of various files. This includes operations such as adding, deleting, modifying, and querying the server user and password, file path, read and write permissions, read time, and access mode (FTP, SFTP), and includes file server connectivity testing capabilities.

[0068] Step 102: perform cross-modal alignment on the structured supplementary content and the preliminary structured text data to generate a structured task list, and bind each task unit in the task list to the corresponding file information in the power grid control standard file.

[0069] It should be noted that the file information in the embodiment of the present application includes the file version number and revision time.

[0070] In one embodiment of the present application, the core of cross-modal alignment is to achieve data fusion through semantic similarity calculation. Exemplarily, the system maps the technical indicator parameters in the scanned table with the description in the text paragraph, and marks the semantic conflict points. The technical indicator parameters are such as line load rate ≤85%, and the description in the text paragraph is such as load rate control requirements. If a conflict is detected, such as the load rate of a line in the table is 90%, the pre-trained deep learning model is used for disambiguation. Specifically, the model gives priority to the indicator parameters of the latest version of the file, and records the disambiguation path for comparison with subsequent versions.

[0071] As you can understand, each task unit in the generated task list includes the task name, responsible unit, completion time, and key indicator thresholds. Task names may include "Line Load Adjustment" and responsible units may include "Provincial Dispatch Operation and Maintenance Department." Task units are dynamically linked to file version numbers and revision times to ensure task traceability. For example, when a file is updated from V1.2 to V1.3, the system automatically links the old and new task lists, noting the scope of the change, such as adjusting the completion time from December 31 to December 15.

[0072] Step 103: Use the label graph neural network to analyze the co-occurrence frequency and dependency relationship between tasks in the task list, dynamically generate corresponding task labels, and based on the task labels, count the multi-dimensional data with the same label.

[0073] In one embodiment of the present application, a Label Graph Neural Network (Label GNN) is used to construct a task association graph. It should be noted that task labels include types such as approaching deadline, indicator abnormality, and manual intervention. For example, when the difference between the task completion time and the system time is less than a preset threshold, such as when there are only three days left until the deadline, the system automatically marks the approaching deadline label and triggers a countdown reminder on the mobile terminal. At the same time, the dependency between tasks is represented by the edge weights of the graph nodes. For example, the dependency between tasks requires that equipment maintenance be started after the power outage application is completed.

[0074] Specifically, if the difference between the actual value of a mission's key metric and the threshold exceeds a preset tolerance, such as a load factor of 90%, the system labels the metric as abnormal and links historical resolution plans for similar tasks, such as the 2022 line capacity expansion record. Furthermore, when users manually adjust task progress or responsible units in the visual interface, the system adds a manual intervention tag and generates an operation log and impact analysis report, such as if a change in responsible unit causes delays in associated tasks. Stacked bar charts display multi-dimensional data statistics based on tags, such as the number of overdue tasks by unit, to support scheduling decision optimization.

[0075] In one embodiment of the present application, after dynamically generating task labels, the system further iteratively optimizes the model through user operation feedback to improve the parsing accuracy of power professional terms. It should be noted that user feedback data includes the user's confirmation, modification or deletion operations on task labels. For example, when the user clicks the confirmation button in the visual interface, the system automatically records the label as valid data. If the user manually modifies the label content, such as changing the indicator abnormality to require review, or deleting a label, it is marked as a negative sample that needs to be corrected.

[0076] Understandably, the collection of feedback data relies on a front-end event monitoring mechanism. Specifically, the system captures user operations through an API interface, extracting the operation type, operation time, and associated task unit ID. Operation types include confirmation, modification, and deletion. For example, if a user changes the indicator anomaly label of a line load adjustment task to manual intervention, the system records the original label, the modified label, and the operation context, such as a comparison of the indicator values ​​before and after the modification.

[0077] The system uses user-verified labeled data as positive samples and modified or deleted labeled data as negative samples to construct a training set. For example, positive samples are used to strengthen the model's memory of correct parsing rules, while negative samples are used to correct the model's tendency to misjudge.

[0078] Through the contrastive learning algorithm, the system calculates the semantic differences between the structured task list and the user operations. It should be noted that semantic differences include entity recognition deviations and context association errors. Entity recognition deviations, such as incorrectly parsing #3 main transformer as #3 transformer, and context association errors, such as incorrectly associating load rate ≤ 85% with non-related tasks. Exemplarily, the contrastive learning algorithm calculates the cosine similarity between the task list and the user operation to determine the degree of match with the user operation, and adjusts the entity recognition priority of the natural language processing model accordingly. For example, if the user modifies the main transformer-related tags multiple times, the recognition weight of the main transformer equipment number in the natural language processing model will be increased to reduce the probability of confusion between similar terms.

[0079] It is understandable that unlabeled data refers to task units that have not yet been covered by user operation feedback. When performing multimodal feature consistency verification on unlabeled task data, first compare the text description in the task unit with the technical indicators of the associated table data. If the semantics are consistent, such as the load rate in the text ≤ 85% matches the table data, then mark it as credible data. Then, compare the label distribution of the current task with that of historical similar tasks. If it conforms to the typical pattern, such as equipment maintenance is often associated with manual intervention labels, it is considered to have passed the consistency. It is also necessary to check whether the task unit conflicts with the technical standards of the latest version of the file. If there is no conflict, it passes the verification.

[0080] For unlabeled data that passes verification, the system automatically generates pseudo-labels. For example, if an unlabeled task, relay protection parameter calibration, matches a historical task pattern and contains no version conflicts, it is automatically labeled as routine maintenance. This pseudo-labeled data is added to the training set to increase sample diversity. Through semi-supervised learning, the model gradually improves its ability to parse specialized power terminology, such as differential protection settings and reclosing time, reducing reliance on manual labeling.

[0081] In one embodiment of the present application, after generating the task label, the system further implements dynamic monitoring and priority adjustment through user operations and task status changes. When a user deletes a task unit such as line load adjustment, the system marks the deleted task with an abandoned label and displays it in grayscale in the visual interface to avoid restarting due to erroneous operation. At the same time, the association between the task unit and the original file version number and revision time is released to ensure the purity of the task list. The reasons for user deletion, such as task duplication or outdated indicators, are collected through pop-up forms or voice input and stored in a preset feedback database, which is stored in the feedback database according to task type.

[0082] When the user edits the task unit, such as modifying the responsible unit or completion time for the main transformer overload disposal, the system monitors the changes in key fields in real time. Specifically, the key fields include the task name, responsible unit, completion time and technical indicator thresholds. For example, when the user changes the responsible unit from the Provincial Dispatching Operation and Maintenance Department to the Local Dispatching Maintenance Group, the edited task unit is marked with a pending review label and highlighted in orange in the visual interface. Compare the task content before and after editing, generate a difference report, such as changes in the responsible unit and a delay of 2 days in the completion time, and push the revised snapshot to the relevant administrator's mobile terminal to initiate a cross-departmental collaborative review mechanism.

[0083] When the progress of a task deviates from the preset timeline, such as the delayed start of an equipment maintenance task, the system triggers a deviation warning. It is understandable that the preset timeline is dynamically generated based on task dependencies and historical execution data. The actual progress of the task is compared with the planned timeline in real time, and the deviation rate is calculated, such as the percentage of delayed days in the total construction period. If the deviation rate exceeds the fault tolerance threshold, the task is labeled as delayed and marked with a red area in the visual interface. The deviation details are pushed to the administrator's mobile app, such as the reason for the delay: spare parts have not arrived, the impact chain: associated with 5 subtasks, and one-click jump to the corresponding emergency response plan is also supported.

[0084] The system automatically handles overdue tasks based on the comparison between the task completion rate and the indicator threshold. It should be noted that the priority reallocation logic is dynamically adjusted based on the task criticality score. When a task is overdue and the indicator is not met, such as the load rate is still 90%, the system marks it as a serious anomaly and highlights it in the stacked bar chart. Based on the scope of task impact, such as the impact on the power grid security level, the task list is reordered and high-priority tasks, such as main transformer overload disposal, are placed at the top. The resource scheduling interface is automatically triggered to allocate additional manpower or equipment to high-priority tasks, such as adding maintenance teams.

[0085] As you can understand, the priority scoring model incorporates task labels, historical processing efficiency, and real-time grid operation data. For example, if a line load adjustment task is nearing a critical load at the associated substation, the system will automatically increase its priority and shorten the review cycle.

[0086] Step 104: Using a multimodal comparison algorithm, compare the semantic differences between different versions of the power grid regulation standard documents, generate a visual difference report to provide abnormal reminders, and mark the corresponding revision suggestions.

[0087] In one embodiment of the present application, a multimodal comparison algorithm is used to detect version differences. It should be noted that the differences between different versions of files include newly added, deleted, or modified paragraph contents and technical indicators. For example, the system parses the revision records of different versions of files, extracts the changed content, such as adjusting the action time threshold from 0.5 seconds to 0.3 seconds in versions V1.2 and V1.3, and calculates the paragraph differences between different versions of files through a semantic similarity algorithm to generate a difference matrix.

[0088] As you can understand, the difference matrix is ​​mapped to a visual interface, highlighting the impact of the change, such as a threshold adjustment affecting five related tasks. Specifically, the system automatically identifies technical standard conflicts, such as conflicting voltage level definitions between the old and new versions, and generates revision suggestions, such as synchronously updating inspection cycle parameters. When a conflict in a key indicator is detected, the system sends an exception alert to the responsible personnel and links to emergency response plans, such as the main transformer overload handling process.

[0089] In one embodiment of the present application, the exam requirement triggering conditions include an administrator proactively initiating an exam request or the system automatically recommending an exam based on learning progress. For example, when a user completes a task label learning phase, such as relay protection procedures, the system automatically triggers the exam process to test the user's knowledge level at that phase.

[0090] It is understandable that the test paper is generated based on the frequently accessed file tags and task types in the user's learning history. Specifically, the system extracts high-frequency tags from the user's historical data, such as main transformer overload handling and line load adjustment, and uses the attention mechanism to filter the procedural clauses associated with the extracted high-frequency tags. For example, if the user frequently accesses files related to line load, the system will prioritize extracting clauses such as load rate control requirements and overload response procedures from the power grid dispatching operating procedures, thereby generating a test paper containing single-choice questions, multiple-choice questions, true-or-false questions, and fill-in-the-blank questions.

[0091] The system monitors the user's answering trajectory in real time during the exam. It should be noted that the answering trajectory includes the answering time, number of revisions and option switching frequency of each question. For example, if the user's stay time on the differential protection constant setting question is significantly higher than the average, and the answer is revised many times, the system will mark the question as a potential weak knowledge point. Specifically, the system evaluates the answering behavior through a timing analysis model. For example, in the fill-in-the-blank question "The main transformer overload handling time limit is ____ minutes", if the user repeatedly enters different values, such as 5 minutes and 10 minutes, the system will combine the answering time and modification record to determine that the knowledge point is not well mastered, and will link it to the original paragraph of the main transformer operating procedures in the background for subsequent reinforcement learning.

[0092] The system automatically grades submitted exam papers by identifying user responses and combining them with the correct answers to selected questions and their difficulty weights. After the exam, the system generates an overall score ranking and compiles a statistical list of frequently failed questions based on error rates. For example, questions related to overload response time have an error rate of 70%, creating a collection of incorrect questions.

[0093] Then, the system also dynamically optimizes the difficulty distribution of questions in subsequent test papers based on the wrong question set data. Specifically, the high-frequency wrong questions in the wrong question set will be classified as high-difficulty or easy-to-confuse types. For example, if the error rate of the differential protection constant calculation question remains high in multiple exams, the system will increase the frequency of similar questions in subsequent test papers and adjust the confusingness of the options, such as adding similar numerical interference items. It is understandable that the difficulty adjustment also takes into account the overall performance of the user group. For example, if most users score low on the technical standard difference analysis question, the system will automatically reduce the initial difficulty weight of similar questions and associate them with the comparison case in the visual interface to assist users in understanding.

[0094] The above is an embodiment of the method proposed in this application. Based on the same inventive concept, this application embodiment also provides an intelligent processing system for power grid control standard files, the structure of which is as follows: Figure 2 shown.

[0095] Figure 2 This is a schematic diagram of the internal structure of a power grid control standard file intelligent processing system provided in an embodiment of the present application. Figure 2 As shown, the system includes: a file parsing module 201 , a task list generating module 202 , a task label generating module 203 , and a difference comparing module 204 .

[0096] In one embodiment of the present application, the file parsing module 201 is configured to parse the power grid regulation standard file using a natural language processing model, extract target file parameters to generate preliminary structured text data, and convert unstructured content in the scanned power grid regulation standard file into structured supplementary content; the unstructured content includes embedded scanned tables and handwritten annotations;

[0097] A task list generation module 202 is configured to perform cross-modal alignment of the structured supplementary content with the preliminary structured text data to generate a structured task list, and to bind each task unit in the task list to the corresponding file information in the power grid regulation standard file; the file information includes the file version number and revision time;

[0098] The task label generation module 203 is used to analyze the co-occurrence frequency and dependency relationship between tasks in the task list using the label graph neural network, dynamically generate corresponding task labels, and count multi-dimensional data with the same label based on the task labels;

[0099] The difference comparison module 204 is used to compare the semantic differences between different versions of the power grid regulation standard documents through a multimodal comparison algorithm, generate a visual difference report to provide abnormal reminders, and mark corresponding revision suggestions.

[0100] Based on the same inventive concept, the present application also provides a power grid control standard file intelligent processing device, the structure of which is as follows: Figure 3 shown.

[0101] Figure 3 This is a schematic diagram of the internal structure of a power grid control standard file intelligent processing device provided in an embodiment of the present application. Figure 3 As shown, the equipment includes:

[0102] at least one processor;

[0103] and, a memory communicatively coupled to the at least one processor;

[0104] The memory stores instructions that can be executed by at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to:

[0105] The system uses a natural language processing model to parse the grid regulation standard document, extract target file parameters, and generate preliminary structured text data. It also converts the unstructured content in the scanned grid regulation standard document into structured supplementary content. The unstructured content includes embedded scanned tables and handwritten annotations.

[0106] Cross-modally align the structured supplementary content with the preliminary structured text data to generate a structured task list, and bind each task unit in the task list to the corresponding file information in the power grid regulation standard file; the file information includes the file version number and revision time;

[0107] Use label graph neural networks to analyze the co-occurrence frequency and dependency relationships between tasks in the task list, dynamically generate corresponding task labels, and based on the task labels, count multi-dimensional data with the same label;

[0108] Through the multimodal comparison algorithm, the semantic differences between different versions of power grid regulation standard documents are compared, and a visual difference report is generated to provide abnormal reminders and mark the corresponding revision suggestions.

[0109] The present application also provides a non-volatile computer storage medium storing computer-executable instructions. When the computer-executable instructions are executed, they can:

[0110] The system uses a natural language processing model to parse the grid regulation standard document, extract target file parameters, and generate preliminary structured text data. It also converts the unstructured content in the scanned grid regulation standard document into structured supplementary content. The unstructured content includes embedded scanned tables and handwritten annotations.

[0111] Cross-modally align the structured supplementary content with the preliminary structured text data to generate a structured task list, and bind each task unit in the task list to the corresponding file information in the power grid regulation standard file; the file information includes the file version number and revision time;

[0112] Use label graph neural networks to analyze the co-occurrence frequency and dependency relationships between tasks in the task list, dynamically generate corresponding task labels, and based on the task labels, count multi-dimensional data with the same label;

[0113] Through the multimodal comparison algorithm, the semantic differences between different versions of power grid regulation standard documents are compared, and a visual difference report is generated to provide abnormal reminders and mark the corresponding revision suggestions.

[0114] The various embodiments in this application are described in a progressive manner. Similar portions between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the device and medium embodiments are generally similar to the method embodiments, so their descriptions are relatively simple. For relevant portions, refer to the descriptions of the method embodiments.

[0115] The devices and media provided in the embodiments of the present application correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects to their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0116] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0117] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0118] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0119] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0120] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0121] Memory may include non-permanent storage in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0122] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can be implemented using any method or technology for information storage. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change RAM (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media, such as modulated data signals and carrier waves.

[0123] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0124] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A method for intelligently processing power grid control standard files, characterized in that: The method comprises: Parsing the grid regulation standard document using a natural language processing model, extracting target file parameters to generate preliminary structured text data, and converting the scanned unstructured content in the grid regulation standard document into structured supplementary content; the unstructured content includes embedded scanned tables and handwritten annotations; Cross-modally aligning the structured supplementary content with the preliminary structured text data to generate a structured task list, and binding each task unit in the task list to corresponding file information in the power grid regulation standard file; the file information includes a file version number and revision time; Utilizing a label graph neural network to analyze the co-occurrence frequency and dependency relationships between tasks in the task list, dynamically generating corresponding task labels, and statistically analyzing multi-dimensional data with the same labels based on the task labels; Through the multimodal comparison algorithm, the semantic differences between different versions of power grid regulation standard documents are compared, and a visual difference report is generated to provide abnormal reminders and mark the corresponding revision suggestions.

2. The method for intelligently processing power grid control standard files according to claim 1, characterized in that: The target file parameters include paragraph title, entity information and technical indicator parameters; Cross-modally aligning the structured supplementary content with the preliminary structured text data to generate a structured task list, specifically including: Performing a topological structure analysis on the row and column data in the scan table to identify a corresponding relationship between a table header and a data unit in the scan table; Mapping the row and column data with technical indicator parameters in the preliminary structured text data by calculating semantic similarity to determine conflicting content between the structured supplementary content and the preliminary structured text data, and marking the conflicting content; Using a pre-trained deep learning model, the conflicting content is disambiguated and the corresponding disambiguation path is recorded for version comparison. The entities in the entity information are associated with the preset power grid dispatching business rules to generate a corresponding task list; the task list includes the task name, responsible unit, completion time and key indicator threshold.

3. The method for intelligently processing power grid control standard files according to claim 1, characterized in that: The label graph neural network is used to analyze the co-occurrence frequency and dependency relationship between tasks in the task list and dynamically generate corresponding task labels, specifically including: When the difference between the task completion time and the current system time is less than the preset threshold, the corresponding work task is marked with a deadline tag and a countdown reminder is pushed to the mobile terminal; When the difference between the actual value of a task's key technical indicator and the threshold exceeds the preset tolerance range, the corresponding work task will be marked with an indicator abnormality label, and the historical handling plan and revision record of similar tasks will be linked; Based on the user's manual operations in the visual interface, a manual intervention tag is added to the corresponding work task, and the corresponding operation log and impact analysis report are recorded; the manual operation includes adjusting the task progress or responsible unit.

4. The method for intelligently processing power grid control standard files according to claim 1, characterized in that: After analyzing the co-occurrence frequencies and dependencies between tasks in the task list using a label graph neural network and dynamically generating corresponding task labels, the method further includes: Based on the user's operation trigger on the task tag, feedback data of the tag operation is collected, and based on the feedback data, the tag data confirmed by the user and the tag data modified or deleted are extracted; The label data confirmed by the user is used as a positive sample, and the modified or deleted label data is used as a negative sample to construct a training set; By using a contrastive learning algorithm, the semantic differences between the structured task list and the user operation are calculated to adjust the entity recognition priority of the natural language processing model; Multimodal feature consistency verification is performed on the unlabeled task data, pseudo labels are generated for the unlabeled task data that pass the verification, and the unlabeled task data corresponding to the pseudo labels are added to the training set to expand the training set and optimize the natural language processing model's ability to parse power professional terminology.

5. The method for intelligently processing power grid control standard files according to claim 1, characterized in that: By using a multimodal comparison algorithm, the semantic differences between different versions of power grid regulation standard documents are compared, and a visual difference report is generated to provide abnormality reminders and mark the corresponding revision suggestions, including: Parse the revision records of different versions of grid regulation standard documents to extract the changed paragraph content and technical indicator parameters; the changes include addition, deletion or modification; Using a semantic similarity algorithm, we calculate the differences between the changed paragraphs, generate a corresponding difference matrix, and mark any technical standard conflict points. Mapping the difference matrix to a visual interface to display the difference analysis results in a highlighted comparison format; the difference analysis results include the impact scope of the version change and compatibility recommendations; A visual difference report is generated based on the difference analysis results, abnormal reminders are given for version differences, and corresponding revision suggestions are marked in the visual difference report; the revision suggestions include unified suggestions for power professional terminology and linkage adjustment strategies for technical indicator parameters.

6. The method for intelligently processing power grid control standard files according to claim 1, characterized in that: The method further comprises: Based on the test requirement trigger, the system extracts frequently accessed file tags and task types from the user's learning records. Based on the attention mechanism, it extracts the procedural clauses associated with the current task from the grid regulation standard file corresponding to the file tags to generate the current test paper. The question types in the current test paper include single-choice questions, multiple-choice questions, true-or-false questions, and fill-in-the-blank questions. During the exam, the user's answer track is monitored in real time, and the test questions that are potential weak points are marked by combining the answer track and the answer time of each question; Identify the user's answers in the submitted examination papers, determine whether the user's answers are correct based on the extracted rules and regulations, and score the user's answers; The user's overall score is ranked, and the exam questions that are easy to make mistakes are counted to form a wrong question set, so as to adjust the difficulty distribution of the questions in the exam paper according to the easy to make mistakes in the wrong question set.

7. The method for intelligently processing power grid control standard files according to claim 1, characterized in that: After analyzing the co-occurrence frequencies and dependencies between tasks in the task list using a label graph neural network and dynamically generating corresponding task labels, the method further includes: When a user deletes a task unit, it is automatically marked as discarded, the task unit is unbound from the file information, and the reason for deletion is recorded in the preset feedback database; When a user edits a task unit, the system monitors changes in key fields in the modified content, marks the corresponding task unit with a pending review tag, and generates a revision comparison snapshot. When the task progress deviates from the preset timeline, a deviation warning is triggered, and the corresponding deviation situation is associated with the administrator's mobile notification; Based on the comparison between the task completion rate and the indicator threshold, overdue tasks are marked as abnormal and the priority reallocation logic is activated.

8. An intelligent processing system for power grid control standard files, characterized in that: The system includes: a file parsing module, a task list generation module, a task label generation module, and a difference comparison module; The file parsing module is configured to parse the grid regulation standard file using a natural language processing model, extract target file parameters to generate preliminary structured text data, and convert unstructured content in the scanned grid regulation standard file into structured supplementary content; the unstructured content includes embedded scanned tables and handwritten annotations; The task list generation module is configured to perform cross-modal alignment between the structured supplementary content and the preliminary structured text data to generate a structured task list, and bind each task unit in the task list to corresponding file information in the power grid regulation standard file; the file information includes a file version number and revision time; The task label generation module is used to analyze the co-occurrence frequency and dependency relationship between tasks in the task list using a label graph neural network, dynamically generate corresponding task labels, and based on the task labels, count multi-dimensional data with the same label; The difference comparison module is used to compare the semantic differences between different versions of power grid regulation standard documents through a multimodal comparison algorithm, generate a visual difference report to provide abnormal reminders, and mark corresponding revision suggestions.

9. An intelligent processing device for power grid control standard files, characterized in that: The device comprises: at least one processor; and, a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the intelligent processing method of power grid regulation standard files as described in any one of claims 1-8.

10. A non-volatile computer storage medium storing computer executable instructions, characterized in that: When the computer-executable instructions are executed, a method for intelligently processing a power grid regulation standard file according to any one of claims 1 to 8 is implemented.

Images