Electric power engineering archive knowledge service method and system based on AI

Abstract

The invention provides an AI-based electric power engineering archive knowledge service method and system in the technical field of electric power engineering archive management. The method comprises the following steps: S1, obtaining a historical archive text to construct a data set; s2, training the archive knowledge extraction model based on the data set; mining an archiving rule from historical archiving information collected from the data set through a big data technology; s3, acquiring a to-be-copied electric power engineering archive file, and performing copy identification; s4, identifying an electric power project file text from the electric power project file; s5, inputting the electric power engineering file text into the file knowledge extraction model to obtain file knowledge so as to construct a file knowledge graph; s6, extracting archiving information from the electric power engineering archive file, generating an archiving reminding notification based on the archiving information and an archiving rule, and executing archiving operation; and S7, recording the archiving characteristics in the archiving process to carry out compliance check. The method has the advantage that the quality and efficiency of electric power engineering file management are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power engineering archive management, and in particular to an AI-based electric power engineering archive knowledge service method and system. Background Art

[0002] Power engineering projects often involve a large number of power engineering archives (archival files), such as design drawings, construction record files, approval documents, equipment information, maintenance logs, etc. The amount of data in these power engineering archives is huge and varied. During the construction of power engineering projects, these power engineering archives must be managed synchronously to ensure the construction quality of the power engineering projects and facilitate traceability in the later stage.

[0003] The management of power engineering archives traditionally relies on manual collection, classification, cataloging, storage, borrowing, and entry of paper-based power engineering archives into archive management systems. The degree of electronicization is not high, and there are the following disadvantages:

[0004] 1. Manual operation leads to inefficient management of power engineering archives and is prone to management omissions; 2. The query of power engineering archives relies on manual browsing or simple electronic search, which is difficult to meet the needs of file users for fast and accurate query and related archive reference and evidence; 3. It is difficult to determine the filing time, responsible person and storage method of each department, and the timeliness and integrity of power engineering archives are easily affected; 4. The lack of unified filing standards can easily lead to incomplete archived files, non-standard filing content, unclear classification, confusing numbering, and archive errors and omissions; 5. Archive management systems often only have basic storage and query functions, lack the ability of knowledge mining and intelligent analysis, and cannot extract valuable information and knowledge from massive electronic engineering archives.

[0005] Therefore, how to provide an AI-based power engineering archive knowledge service method and system to improve the quality and efficiency of power engineering archive management has become a technical problem that needs to be solved urgently. Summary of the invention

[0006] The technical problem to be solved by the present invention is to provide an AI-based electric power engineering archive knowledge service method and system to improve the quality and efficiency of electric power engineering archive management.

[0007] In a first aspect, the present invention provides an AI-based power engineering archive knowledge service method, comprising the following steps:

[0008] Step S1, obtaining a large number of historical archive texts of power projects, and constructing a data set after preprocessing and annotating each of the historical archive texts;

[0009] Step S2: creating an archival knowledge extraction model for identifying entities, attributes and relationships based on a neural network, and training the archival knowledge extraction model based on the data set; collecting historical archival information from the data set, mining archiving rules from each of the historical archival information through big data technology, and setting a corresponding archiving template based on the archiving rules;

[0010] Step S3: obtaining the electric power engineering archive files to be managed through the ETL tool, and performing copy identification on the electric power engineering archive files through the YOLO model;

[0011] Step S4, identifying the electric power engineering archive text from the electric power engineering archive file by using OCR technology, and preprocessing the electric power engineering archive text;

[0012] Step S5, inputting the preprocessed electric power engineering archive text into an archive knowledge extraction model to obtain archive knowledge including entities, attributes and relationships, and constructing an archive knowledge graph based on each of the archive knowledge;

[0013] Step S6: extracting archiving information from the electric power engineering archive file, and generating an archiving reminder notification carrying a recommended archiving template based on the archiving information and the archiving rule to perform an archiving operation;

[0014] Step S7: Record the archiving characteristics during the archiving process, and perform an archiving compliance check based on the archiving characteristics;

[0015] Step S8: The search statement input based on natural language technology is converted through generative AI technology to generate a search instruction, and the power engineering archive files are searched and traced based on the search instruction and the archive knowledge graph.

[0016] Furthermore, the step S1 is specifically as follows:

[0017] Acquire a large number of historical archive texts of power engineering projects, perform preprocessing on each of the historical archive texts, including at least data cleaning, format conversion and data integration, annotate entities, attributes and relationships on each of the preprocessed historical archive texts based on the model layer, and construct a data set based on each of the annotated historical archive texts;

[0018] The step S2 is specifically as follows:

[0019] Creating a file knowledge extraction model for identifying entities, attributes and relationships based on a neural network, dividing the data set into a training set and a verification set based on a preset ratio, training the file knowledge extraction model based on the training set until a preset convergence condition is met, and then verifying the trained file knowledge extraction model based on the verification set to determine whether the extraction accuracy is greater than a preset accuracy threshold, if so, terminating the training; if not, expanding the training set to continue the training;

[0020] Historical archiving information including at least archiving time, content type, project type and project stage is collected from the data set, archiving rules are mined from each of the historical archiving information through big data technology, and corresponding archiving templates are set based on the archiving rules.

[0021] Furthermore, the step S3 is specifically as follows:

[0022] Acquire the electric power engineering archive files to be managed through the ETL tool, identify the image features of the electric power engineering archive files including at least clarity, color reproduction, contrast, seal size and seal color through the YOLO model, and perform photocopy identification on the electric power engineering archive files based on the image features;

[0023] The step S4 is specifically as follows:

[0024] The electric power engineering archive text is identified from the electric power engineering archive file by using OCR technology, and the electric power engineering archive text is preprocessed including at least data cleaning, format conversion and data integration.

[0025] Furthermore, the step S6 is specifically as follows:

[0026] Archiving information including at least archiving time, content type, project type and project stage is extracted from the power engineering archive file, and an archiving reminder notification is generated based on the archiving information and archiving rules. The archiving reminder notification is displayed on a display screen, or the archiving reminder notification is automatically pushed to a pre-associated management terminal to perform the archiving operation.

[0027] Furthermore, the step S7 is specifically as follows:

[0028] During the archiving process, the archiving features including at least the file number style, title specification and archive template are recorded. The pre-trained large model is used to perform an archiving compliance check based on the archiving features, archiving rules and preset archiving rules.

[0029] The archiving rules include at least archiving entries, archiving directories, file number styles, title specifications and archive templates; the archiving entries include at least case number, file number, file name, case name, file type and file description; the archiving directory is the storage directory of power engineering archive files.

[0030] In a second aspect, the present invention provides an AI-based power engineering archive knowledge service system, comprising the following modules:

[0031] A data set construction module is used to obtain a large number of historical archive texts of power projects, and to construct a data set after preprocessing and annotating each of the historical archive texts;

[0032] A model training module is used to create an archival knowledge extraction model for identifying entities, attributes and relationships based on a neural network, and train the archival knowledge extraction model based on the data set; collect historical archival information from the data set, mine archiving rules from each of the historical archiving information through big data technology, and set corresponding archiving templates based on the archiving rules;

[0033] A copy identification module is used to obtain the electric power engineering archive files to be managed through the ETL tool, and to perform copy identification on the electric power engineering archive files through the YOLO model;

[0034] An archive text recognition module, used to recognize the electric power engineering archive text from the electric power engineering archive file by using OCR technology, and pre-process the electric power engineering archive text;

[0035] A knowledge graph construction module is used to input the preprocessed electric power engineering archive text into an archive knowledge extraction model to obtain archive knowledge including entities, attributes and relationships, and to construct an archive knowledge graph based on each of the archive knowledge;

[0036] An archiving reminder module, used to extract archiving information from the electric power engineering archive file, and generate an archiving reminder notification carrying a recommended archiving template based on the archiving information and the archiving rule, so as to perform an archiving operation;

[0037] A compliance check module, used for recording archiving characteristics during the archiving process and performing archiving compliance check based on the archiving characteristics;

[0038] The knowledge question and answer module is used to convert the search statements input based on natural language technology through generative AI technology, generate search instructions, and search and trace the power engineering archive files based on the search instructions and the archive knowledge graph.

[0039] Furthermore, the data set construction module is specifically used for:

[0040] Acquire a large number of historical archive texts of power engineering projects, perform preprocessing on each of the historical archive texts, including at least data cleaning, format conversion and data integration, annotate entities, attributes and relationships on each of the preprocessed historical archive texts based on the model layer, and construct a data set based on each of the annotated historical archive texts;

[0041] The model training module is specifically used for:

[0042] Creating a file knowledge extraction model for identifying entities, attributes and relationships based on a neural network, dividing the data set into a training set and a verification set based on a preset ratio, training the file knowledge extraction model based on the training set until a preset convergence condition is met, and then verifying the trained file knowledge extraction model based on the verification set to determine whether the extraction accuracy is greater than a preset accuracy threshold, if so, terminating the training; if not, expanding the training set to continue the training;

[0043] Historical archiving information including at least archiving time, content type, project type and project stage is collected from the data set, archiving rules are mined from each of the historical archiving information through big data technology, and corresponding archiving templates are set based on the archiving rules.

[0044] Furthermore, the copy identification module is specifically used for:

[0045] Acquire the electric power engineering archive files to be managed through the ETL tool, identify the image features of the electric power engineering archive files including at least clarity, color reproduction, contrast, seal size and seal color through the YOLO model, and perform photocopy identification on the electric power engineering archive files based on the image features;

[0046] The archive text recognition module is specifically used for:

[0047] The electric power engineering archive text is identified from the electric power engineering archive file by using OCR technology, and the electric power engineering archive text is preprocessed including at least data cleaning, format conversion and data integration.

[0048] Furthermore, the archiving reminder module is specifically used for:

[0049] Archiving information including at least archiving time, content type, project type and project stage is extracted from the power engineering archive file, and an archiving reminder notification is generated based on the archiving information and archiving rules. The archiving reminder notification is displayed on a display screen, or the archiving reminder notification is automatically pushed to a pre-associated management terminal to perform the archiving operation.

[0050] Furthermore, the compliance checking module is specifically used to:

[0051] During the archiving process, the archiving features including at least the file number style, title specification and archive template are recorded. The pre-trained large model is used to perform an archiving compliance check based on the archiving features, archiving rules and preset archiving rules.

[0052] The archiving rules include at least archiving entries, archiving directories, file number styles, title specifications and archive templates; the archiving entries include at least case number, file number, file name, case name, file type and file description; the archiving directory is the storage directory of power engineering archive files.

[0053] The advantages of the present invention are:

[0054] A data set is constructed by acquiring a large amount of historical archival texts of electric power projects and performing preprocessing and annotation. The created archival knowledge extraction model is trained based on the data set, and historical archival information is collected from the data set. Archiving rules are mined from each historical archival information through big data technology, and corresponding archiving templates are set based on the archiving rules. Then, the electric power engineering archival files to be managed are acquired through ETL tools and photocopies are identified. The electric power engineering archival texts are identified and preprocessed from the electric power engineering archival files through OCR technology. The preprocessed electric power engineering archival texts are input into the archival knowledge extraction model to obtain archival knowledge including entities, attributes and relationships, and an archival knowledge graph is constructed based on each archival knowledge. Then, archival information is extracted from the electric power engineering archival files, and an archiving reminder notification carrying a recommended archiving template is generated based on the archiving information and archiving rules to execute the archiving operation. The archiving features are recorded during the archiving process to perform archiving compliance checks. Finally, the retrieval statements input based on natural language technology are converted through generative AI technology to generate retrieval instructions. Based on the retrieval instructions and archiving rules, the archiving template is extracted and the archiving instructions are generated. The archive knowledge graph is used to retrieve and trace the power engineering archive files; that is, the power engineering archive files are automatically obtained through the ETL tool for photocopy identification, the power engineering archive text is automatically identified and preprocessed through the OCR technology, and the archive knowledge is automatically extracted from the power engineering archive text through the pre-trained archive knowledge extraction model to construct an archive knowledge graph for subsequent archive retrieval, and the archiving rules mined by big data technology are used to generate archiving reminders. The archiving compliance check is performed through archiving features, archiving rules, and archiving rules. No manual operation is required, avoiding omissions caused by manual operation, and also avoiding problems such as incomplete archived files, irregular archived content, unclear classification, confusing numbering, and archive omissions. The full electronicization facilitates rapid retrieval and traceability in the later stage. Combined with the archive knowledge graph, the correlation between power engineering archive files can be quickly discovered, and valuable information and knowledge can be fully extracted from massive power engineering archive files, which not only effectively improves the utilization value of power engineering archive files, but also effectively reduces the difficulty of knowledge acquisition, and ultimately greatly improves the quality and efficiency of power engineering archive management. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The present invention will be further described below in conjunction with embodiments with reference to the accompanying drawings.

[0056] Figure 1 It is a flow chart of an AI-based electric power engineering archive knowledge service method of the present invention.

[0057] Figure 2 It is a structural schematic diagram of an AI-based electric power engineering archive knowledge service system of the present invention. DETAILED DESCRIPTION

[0058] The technical solution in the embodiments of the present application has the following overall idea: automatically obtain power engineering archive files through ETL tools for photocopy identification, automatically identify power engineering archive texts through OCR technology and perform preprocessing, automatically extract archive knowledge from power engineering archive texts through pre-trained archive knowledge extraction models to construct archive knowledge graphs for subsequent archive retrieval, generate archive reminder notifications based on archive rules mined through big data technology, perform archive compliance checks based on archive features, archive rules, and archive rules without manual operation to ensure the standardization of archiving, and the entire process is electronic to facilitate rapid retrieval and traceability in the later stage. Combined with the archive knowledge graph, the correlation between power engineering archive files can be quickly discovered to improve the quality and efficiency of power engineering archive management.

[0059] Please refer to Figure 1 to Figure 2 As shown, a preferred embodiment of an AI-based power engineering archive knowledge service method of the present invention includes the following steps:

[0060] Step S1, obtaining a large number of historical archive texts of power projects, and constructing a data set after preprocessing and annotating each of the historical archive texts;

[0061] Step S2: creating an archival knowledge extraction model for identifying entities, attributes and relationships based on a neural network, and training the archival knowledge extraction model based on the data set; collecting historical archival information from the data set, mining archiving rules from each of the historical archival information through big data technology, and setting a corresponding archiving template based on the archiving rules;

[0062] Step S3, obtaining the electric power engineering archive files to be managed through the ETL tool, and performing copy identification on the electric power engineering archive files through the YOLO model; that is, only performing the archiving operation on the original electric power engineering archive files;

[0063] Step S4, identifying the electric power engineering archive text from the electric power engineering archive file by using OCR technology, and preprocessing the electric power engineering archive text;

[0064] Step S5, inputting the preprocessed electric power engineering archive text into an archive knowledge extraction model to obtain archive knowledge including entities, attributes and relationships, and constructing an archive knowledge graph based on each of the archive knowledge;

[0065] Step S6: extracting archiving information from the electric power engineering archive file, and generating an archiving reminder notification carrying a recommended archiving template based on the archiving information and the archiving rule to perform an archiving operation;

[0066] Step S7: Record the archiving characteristics during the archiving process, and perform an archiving compliance check based on the archiving characteristics;

[0067] Step S8: The search statement input based on natural language technology is converted through generative AI technology to generate a search instruction, and the power engineering archive files are searched and traced based on the search instruction and the archive knowledge graph.

[0068] The present invention constructs an integrated power engineering archive management system for archival knowledge collection, identification, extraction, analysis, fusion, processing and application of power engineering archive documents, integrating generative knowledge question and answer, archiving reminder, archiving template matching, photocopy identification and other functions to ensure the timeliness, standardization and integrity of archiving.

[0069] The step S1 is specifically as follows:

[0070] Acquire a large number of historical archive texts of power engineering projects, perform preprocessing on each of the historical archive texts, including at least data cleaning, format conversion and data integration, annotate entities, attributes and relationships on each of the preprocessed historical archive texts based on the model layer, and construct a data set based on each of the annotated historical archive texts;

[0071] The model layer is an important part of the knowledge graph, which defines the structure and rules of the knowledge graph and ensures the reliability and consistency of the knowledge graph. It mainly includes an ontology library and other rules for managing the connections between objects such as entities, relationships and attributes in the knowledge graph.

[0072] The step S2 is specifically as follows:

[0073] Creating a file knowledge extraction model for identifying entities, attributes and relationships based on a neural network, dividing the data set into a training set and a verification set based on a preset ratio, training the file knowledge extraction model based on the training set until a preset convergence condition is met, and then verifying the trained file knowledge extraction model based on the verification set to determine whether the extraction accuracy is greater than a preset accuracy threshold, if so, terminating the training; if not, expanding the training set to continue the training;

[0074] Historical archiving information including at least archiving time, content type, project type and project stage is collected from the data set, archiving rules are mined from each of the historical archiving information through big data technology, and corresponding archiving templates are set based on the archiving rules.

[0075] The step S3 is specifically as follows:

[0076] Acquire the electric power engineering archive files to be managed through the ETL tool, identify the image features of the electric power engineering archive files including at least clarity, color reproduction, contrast, seal size and seal color through the YOLO model, and perform photocopy identification on the electric power engineering archive files based on the image features;

[0077] The step S4 is specifically as follows:

[0078] The electric power engineering archive text is identified from the electric power engineering archive file by using OCR technology, and the electric power engineering archive text is preprocessed including at least data cleaning, format conversion and data integration.

[0079] The step S5 is specifically as follows:

[0080] The preprocessed electric power engineering archive text is input into the archive knowledge extraction model in real time to obtain the archive knowledge including entities, attributes and relationships, and an archive knowledge graph is constructed based on the pattern layer of each archive knowledge. The archive knowledge graph is continuously updated and iterated based on the new electric power engineering archive file to ensure the timeliness and accuracy of the archive knowledge graph.

[0081] The step S6 is specifically as follows:

[0082] Extract archiving information including at least archiving time, content type, project type and project stage from the electric power engineering archive text corresponding to the electric power engineering archive file, generate an archiving reminder notification based on the archiving information and archiving rules, display the archiving reminder notification through a display screen, or automatically push the archiving reminder notification to a pre-associated management terminal to perform an archiving operation. During the archiving process, an archiving log including at least archiving time, file source, responsible person, storage method, approval information, issuance information and modification information is recorded in real time.

[0083] The step S7 is specifically as follows:

[0084] During the archiving process, the archiving features including at least the file number style, title specification and archive template are recorded. The pre-trained large model is used to perform an archiving compliance check based on the archiving features, archiving rules and preset archiving rules.

[0085] The archiving rules include at least archiving entries, archiving directories, file number styles, title specifications and archive templates; the archiving entries include at least case number, file number, file name, case name, file type and file description; the archiving directory is the storage directory of power engineering archive files.

[0086] The step S8 is specifically as follows:

[0087] Generative AI technology is used to convert search statements input based on natural language technology to generate structured search instructions. Archived power engineering archive files are retrieved and traced based on the search instructions and archive knowledge graphs, and the search results are displayed in real time through a visual interface.

[0088] The generative AI technology can be selected as GPT, that is, through natural language interaction, questions about the early, middle and late stages of power engineering (power engineering archive files) can be asked, such as early guidance, engineering reference, project planning, construction difficulties, etc.

[0089] A preferred embodiment of an AI-based power engineering archive knowledge service system of the present invention includes the following modules:

[0090] A data set construction module is used to obtain a large number of historical archive texts of power projects, and to construct a data set after preprocessing and annotating each of the historical archive texts;

[0091] A model training module is used to create an archival knowledge extraction model for identifying entities, attributes and relationships based on a neural network, and train the archival knowledge extraction model based on the data set; collect historical archival information from the data set, mine archiving rules from each of the historical archiving information through big data technology, and set corresponding archiving templates based on the archiving rules;

[0092] A copy identification module is used to obtain the electric power engineering archive files to be managed through the ETL tool, and to perform copy identification on the electric power engineering archive files through the YOLO model; that is, only the original electric power engineering archive files are archived;

[0093] An archive text recognition module, used to recognize the electric power engineering archive text from the electric power engineering archive file by using OCR technology, and pre-process the electric power engineering archive text;

[0094] A knowledge graph construction module is used to input the preprocessed electric power engineering archive text into an archive knowledge extraction model to obtain archive knowledge including entities, attributes and relationships, and to construct an archive knowledge graph based on each of the archive knowledge;

[0095] An archiving reminder module, used to extract archiving information from the electric power engineering archive file, and generate an archiving reminder notification carrying a recommended archiving template based on the archiving information and the archiving rule, so as to perform an archiving operation;

[0096] A compliance check module, used for recording archiving characteristics during the archiving process and performing archiving compliance check based on the archiving characteristics;

[0097] The knowledge question and answer module is used to convert the search statements input based on natural language technology through generative AI technology, generate search instructions, and search and trace the power engineering archive files based on the search instructions and the archive knowledge graph.

[0098] The present invention constructs an integrated power engineering archive management system for archival knowledge collection, identification, extraction, analysis, fusion, processing and application of power engineering archive documents, integrating generative knowledge question and answer, archiving reminder, archiving template matching, photocopy identification and other functions to ensure the timeliness, standardization and integrity of archiving.

[0099] The data set construction module is specifically used for:

[0100] Acquire a large number of historical archive texts of power engineering projects, perform preprocessing on each of the historical archive texts, including at least data cleaning, format conversion and data integration, annotate entities, attributes and relationships on each of the preprocessed historical archive texts based on the model layer, and construct a data set based on each of the annotated historical archive texts;

[0101] The model layer is an important part of the knowledge graph, which defines the structure and rules of the knowledge graph and ensures the reliability and consistency of the knowledge graph. It mainly includes an ontology library and other rules for managing the connections between objects such as entities, relationships and attributes in the knowledge graph.

[0102] The model training module is specifically used for:

[0103] Creating a file knowledge extraction model for identifying entities, attributes and relationships based on a neural network, dividing the data set into a training set and a verification set based on a preset ratio, training the file knowledge extraction model based on the training set until a preset convergence condition is met, and then verifying the trained file knowledge extraction model based on the verification set to determine whether the extraction accuracy is greater than a preset accuracy threshold, if so, terminating the training; if not, expanding the training set to continue the training;

[0104] Historical archiving information including at least archiving time, content type, project type and project stage is collected from the data set, archiving rules are mined from each of the historical archiving information through big data technology, and corresponding archiving templates are set based on the archiving rules.

[0105] The copy identification module is specifically used for:

[0106] Acquire the electric power engineering archive files to be managed through the ETL tool, identify the image features of the electric power engineering archive files including at least clarity, color reproduction, contrast, seal size and seal color through the YOLO model, and perform photocopy identification on the electric power engineering archive files based on the image features;

[0107] The archive text recognition module is specifically used for:

[0108] The electric power engineering archive text is identified from the electric power engineering archive file by using OCR technology, and the electric power engineering archive text is preprocessed including at least data cleaning, format conversion and data integration.

[0109] The knowledge graph construction module is specifically:

[0110] The preprocessed electric power engineering archive text is input into the archive knowledge extraction model in real time to obtain the archive knowledge including entities, attributes and relationships, and an archive knowledge graph is constructed based on the pattern layer of each archive knowledge. The archive knowledge graph is continuously updated and iterated based on the new electric power engineering archive file to ensure the timeliness and accuracy of the archive knowledge graph.

[0111] The archiving reminder module is specifically used for:

[0112] Extract archiving information including at least archiving time, content type, project type and project stage from the electric power engineering archive text corresponding to the electric power engineering archive file, generate an archiving reminder notification based on the archiving information and archiving rules, display the archiving reminder notification through a display screen, or automatically push the archiving reminder notification to a pre-associated management terminal to perform an archiving operation. During the archiving process, an archiving log including at least archiving time, file source, responsible person, storage method, approval information, issuance information and modification information is recorded in real time.

[0113] The compliance checking module is specifically used for:

[0114] During the archiving process, the archiving features including at least the file number style, title specification and archive template are recorded. The pre-trained large model is used to perform an archiving compliance check based on the archiving features, archiving rules and preset archiving rules.

[0115] The archiving rules include at least archiving entries, archiving directories, file number styles, title specifications and archive templates; the archiving entries include at least case number, file number, file name, case name, file type and file description; the archiving directory is the storage directory of power engineering archive files.

[0116] The knowledge question answering module is specifically used for:

[0117] Generative AI technology is used to convert search statements input based on natural language technology to generate structured search instructions. Archived power engineering archive files are retrieved and traced based on the search instructions and archive knowledge graphs, and the search results are displayed in real time through a visual interface.

[0118] The generative AI technology can be selected as GPT, that is, through natural language interaction, questions about the early, middle and late stages of power engineering (power engineering archive files) can be asked, such as early guidance, engineering reference, project planning, construction difficulties, etc.

[0119] In summary, the advantages of the present invention are:

[0120] A data set is constructed by acquiring a large amount of historical archival texts of electric power projects and performing preprocessing and annotation. The created archival knowledge extraction model is trained based on the data set, and historical archival information is collected from the data set. Archiving rules are mined from each historical archival information through big data technology, and corresponding archiving templates are set based on the archiving rules. Then, the electric power engineering archival files to be managed are acquired through ETL tools and photocopies are identified. The electric power engineering archival texts are identified and preprocessed from the electric power engineering archival files through OCR technology. The preprocessed electric power engineering archival texts are input into the archival knowledge extraction model to obtain archival knowledge including entities, attributes and relationships, and an archival knowledge graph is constructed based on each archival knowledge. Then, archival information is extracted from the electric power engineering archival files, and an archiving reminder notification carrying a recommended archiving template is generated based on the archiving information and archiving rules to execute the archiving operation. The archiving features are recorded during the archiving process to perform archiving compliance checks. Finally, the retrieval statements input based on natural language technology are converted through generative AI technology to generate retrieval instructions. Based on the retrieval instructions and archiving rules, the archiving template is extracted and the archiving instructions are generated. The archive knowledge graph is used to retrieve and trace the power engineering archive files; that is, the power engineering archive files are automatically obtained through the ETL tool for photocopy identification, the power engineering archive text is automatically identified and preprocessed through the OCR technology, and the archive knowledge is automatically extracted from the power engineering archive text through the pre-trained archive knowledge extraction model to construct an archive knowledge graph for subsequent archive retrieval, and the archiving rules mined by big data technology are used to generate archiving reminders. The archiving compliance check is performed through archiving features, archiving rules, and archiving rules. No manual operation is required, avoiding omissions caused by manual operation, and also avoiding problems such as incomplete archived files, irregular archived content, unclear classification, confusing numbering, and archive omissions. The full electronicization facilitates rapid retrieval and traceability in the later stage. Combined with the archive knowledge graph, the correlation between power engineering archive files can be quickly discovered, and valuable information and knowledge can be fully extracted from massive power engineering archive files, which not only effectively improves the utilization value of power engineering archive files, but also effectively reduces the difficulty of knowledge acquisition, and ultimately greatly improves the quality and efficiency of power engineering archive management.

[0121] Although the specific implementation modes of the present invention are described above, those skilled in the art should understand that the specific implementation modes described are only illustrative and are not intended to limit the scope of the present invention. Equivalent modifications and changes made by those skilled in the art in accordance with the spirit of the present invention should be included in the scope of protection of the claims of the present invention.

Claims

1. An AI-based power engineering archive knowledge service method, characterized by: The steps include: Step S1, obtaining a large number of historical archive texts of power projects, and constructing a data set after preprocessing and annotating each of the historical archive texts; Step S2: creating an archival knowledge extraction model for identifying entities, attributes and relationships based on a neural network, and training the archival knowledge extraction model based on the data set; collecting historical archival information from the data set, mining archiving rules from each of the historical archival information through big data technology, and setting a corresponding archiving template based on the archiving rules; Step S3, obtaining the electric power engineering archive files to be managed through the ETL tool, and performing copy identification on the electric power engineering archive files through the YOLO model; Step S4, identifying the electric power engineering archive text from the electric power engineering archive file by using OCR technology, and preprocessing the electric power engineering archive text; Step S5, inputting the preprocessed electric power engineering archive text into an archive knowledge extraction model to obtain archive knowledge including entities, attributes and relationships, and constructing an archive knowledge graph based on each of the archive knowledge; Step S6: extracting archiving information from the electric power engineering archive file, and generating an archiving reminder notification carrying a recommended archiving template based on the archiving information and the archiving rule to perform an archiving operation; Step S7: Record the archiving characteristics during the archiving process, and perform an archiving compliance check based on the archiving characteristics; Step S8: The search statement input based on natural language technology is converted through generative AI technology to generate a search instruction, and the power engineering archive files are searched and traced based on the search instruction and the archive knowledge graph.

2. The AI-based power engineering archive knowledge service method according to claim 1, characterized in that: The step S1 is specifically as follows: Acquire a large number of historical archive texts of power engineering projects, perform preprocessing on each of the historical archive texts, including at least data cleaning, format conversion and data integration, annotate entities, attributes and relationships on each of the preprocessed historical archive texts based on the model layer, and construct a data set based on each of the annotated historical archive texts; The step S2 is specifically as follows: Creating a file knowledge extraction model for identifying entities, attributes and relationships based on a neural network, dividing the data set into a training set and a verification set based on a preset ratio, training the file knowledge extraction model based on the training set until a preset convergence condition is met, and then verifying the trained file knowledge extraction model based on the verification set to determine whether the extraction accuracy is greater than a preset accuracy threshold, if so, terminating the training; if not, expanding the training set to continue the training; Historical archiving information including at least archiving time, content type, project type and project stage is collected from the data set, archiving rules are mined from each of the historical archiving information through big data technology, and corresponding archiving templates are set based on the archiving rules.

3. The AI-based power engineering archive knowledge service method according to claim 1, characterized in that: The step S3 is specifically as follows: Acquire the electric power engineering archive files to be managed through the ETL tool, identify the image features of the electric power engineering archive files including at least clarity, color reproduction, contrast, seal size and seal color through the YOLO model, and perform photocopy identification on the electric power engineering archive files based on the image features; The step S4 is specifically as follows: The electric power engineering archive text is identified from the electric power engineering archive file by using OCR technology, and the electric power engineering archive text is preprocessed including at least data cleaning, format conversion and data integration.

4. The AI-based power engineering archive knowledge service method according to claim 1, characterized in that: The step S6 is specifically as follows: Archiving information including at least archiving time, content type, project type and project stage is extracted from the power engineering archive file, and an archiving reminder notification is generated based on the archiving information and archiving rules. The archiving reminder notification is displayed on a display screen, or the archiving reminder notification is automatically pushed to a pre-associated management terminal to perform the archiving operation.

5. The AI-based power engineering archive knowledge service method according to claim 1, characterized in that: The step S7 is specifically as follows: During the archiving process, the archiving features including at least the file number style, title specification and archive template are recorded. The pre-trained large model is used to perform an archiving compliance check based on the archiving features, archiving rules and preset archiving rules. The archiving rules include at least archiving entries, archiving directories, file number styles, title specifications and archive templates; the archiving entries include at least case number, file number, file name, case name, file type and file description; the archiving directory is the storage directory of power engineering archive files.

6. An AI-based power engineering archive knowledge service system, characterized by: Includes the following modules: A data set construction module is used to obtain a large number of historical archive texts of power projects, and to construct a data set after preprocessing and annotating each of the historical archive texts; A model training module is used to create an archival knowledge extraction model for identifying entities, attributes and relationships based on a neural network, and train the archival knowledge extraction model based on the data set; collect historical archival information from the data set, mine archiving rules from each of the historical archiving information through big data technology, and set corresponding archiving templates based on the archiving rules; A copy identification module is used to obtain the electric power engineering archive files to be managed through the ETL tool, and to perform copy identification on the electric power engineering archive files through the YOLO model; An archive text recognition module, used to recognize the electric power engineering archive text from the electric power engineering archive file by using OCR technology, and pre-process the electric power engineering archive text; A knowledge graph construction module is used to input the preprocessed electric power engineering archive text into an archive knowledge extraction model to obtain archive knowledge including entities, attributes and relationships, and to construct an archive knowledge graph based on each of the archive knowledge; An archiving reminder module, used to extract archiving information from the electric power engineering archive file, and generate an archiving reminder notification carrying a recommended archiving template based on the archiving information and the archiving rule, so as to perform an archiving operation; A compliance check module, used for recording archiving characteristics during the archiving process and performing archiving compliance check based on the archiving characteristics; The knowledge question and answer module is used to convert the search statements input based on natural language technology through generative AI technology, generate search instructions, and search and trace the power engineering archive files based on the search instructions and the archive knowledge graph.

7. The AI-based power engineering archive knowledge service system according to claim 6, characterized in that: The data set construction module is specifically used for: Acquire a large number of historical archive texts of power engineering projects, perform preprocessing on each of the historical archive texts, including at least data cleaning, format conversion and data integration, annotate entities, attributes and relationships on each of the preprocessed historical archive texts based on the model layer, and construct a data set based on each of the annotated historical archive texts; The model training module is specifically used for: Creating a file knowledge extraction model for identifying entities, attributes and relationships based on a neural network, dividing the data set into a training set and a verification set based on a preset ratio, training the file knowledge extraction model based on the training set until a preset convergence condition is met, and then verifying the trained file knowledge extraction model based on the verification set to determine whether the extraction accuracy is greater than a preset accuracy threshold, if so, terminating the training; if not, expanding the training set to continue the training; Historical archiving information including at least archiving time, content type, project type and project stage is collected from the data set, archiving rules are mined from each of the historical archiving information through big data technology, and corresponding archiving templates are set based on the archiving rules.

8. The AI-based power engineering archive knowledge service system according to claim 6, characterized in that: The copy identification module is specifically used for: Acquire the electric power engineering archive files to be managed through the ETL tool, identify the image features of the electric power engineering archive files including at least clarity, color reproduction, contrast, seal size and seal color through the YOLO model, and perform photocopy identification on the electric power engineering archive files based on the image features; The archive text recognition module is specifically used for: The electric power engineering archive text is identified from the electric power engineering archive file by using OCR technology, and the electric power engineering archive text is preprocessed including at least data cleaning, format conversion and data integration.

9. The AI-based power engineering archive knowledge service system according to claim 6, characterized in that: The archiving reminder module is specifically used for: Archiving information including at least archiving time, content type, project type and project stage is extracted from the power engineering archive file, and an archiving reminder notification is generated based on the archiving information and archiving rules. The archiving reminder notification is displayed on a display screen, or the archiving reminder notification is automatically pushed to a pre-associated management terminal to perform the archiving operation.

10. The AI-based power engineering archive knowledge service system according to claim 6, characterized in that: The compliance checking module is specifically used for: During the archiving process, the archiving features including at least the file number style, title specification and archive template are recorded. The pre-trained large model is used to perform an archiving compliance check based on the archiving features, archiving rules and preset archiving rules. The archiving rules include at least archiving entries, archiving directories, file number styles, title specifications and archive templates; the archiving entries include at least case number, file number, file name, case name, file type and file description; the archiving directory is the storage directory of power engineering archive files.

Images