New energy operation and maintenance training resource automatic generation method and system

Abstract

The invention discloses a new energy operation and maintenance training resource automatic generation method and system, and the method comprises the steps: obtaining a new energy operation and maintenance original teaching material, and carrying out the storage and preprocessing, and obtaining a preprocessed material; performing character recognition and voice recognition processing according to the preprocessed material to obtain processed text data; performing data cleaning and standardization processing according to the text data, and storing the data in a database of a system to obtain a structured knowledge point set; and automatically generating a structured PPT courseware file and a matched teaching video file based on a preset training target according to the structured knowledge point set. According to the method, the multi-modal analysis and the AI generation technology are connected in series, an automatic assembly line from original data to standard video courseware is constructed, the mode of manually making courseware and recording videos at a low speed is changed, and the efficiency is greatly improved.

Description

Technical Field

[0001] This invention relates to a method and system for automatically generating training resources for new energy operation and maintenance, belonging to the fields of artificial intelligence and education technology, and relating to the new energy power industry. Background Technology

[0002] With the rapid development of the new energy industry, the scale of new energy assets such as photovoltaic power plants and wind farms is constantly expanding. Unlike the centralized management of traditional thermal power plants, new energy power plants are often characterized by wide geographical distribution and numerous scattered locations. In order to reduce operating costs, the industry currently widely adopts the business model of outsourcing operation and maintenance.

[0003] The existing new energy operation and maintenance training system mainly relies on the following methods for personnel skills and safety training: First, traditional offline teaching, where the owner or senior technicians organize on-site training and distribute paper or electronic technical manuals for personnel to read; second, general online learning platforms, where companies purchase general online learning systems, upload pre-recorded standard course videos, and employees log in with their accounts to watch and take exams; and third, manually creating multimedia courseware, where training instructors use PowerPoint and video editing software to manually create teaching courseware and videos.

[0004] However, in practical applications, the above model has the following problems when facing specific scenarios of new energy operation and maintenance:

[0005] 1. The efficiency of training resource generation is low, failing to meet the training needs of highly mobile personnel. Due to the high turnover rate of outsourced maintenance personnel, on-site safety production education and basic skills training have become high-frequency repetitive tasks. Faced with frequent personnel changes, the owner consumes a lot of energy in repetitive on-site education, resulting in low training efficiency.

[0006] 2. New energy power plants have accumulated a massive amount of technical documents, including equipment manuals, operation and maintenance manuals, and fault case libraries. This data is mostly in heterogeneous unstructured or semi-structured formats such as PDFs, images, and TXT files. Current technologies lack effective automated methods to deeply analyze this heterogeneous data and utilize it to generate teaching content.

[0007] 3. The teaching methods are monotonous and lack intelligent transformation mechanisms. Existing document management systems can only store and preview files, but cannot understand the document content; while existing video production software is powerful, it lacks knowledge comprehension capabilities. There is a disconnect between the two, making it impossible to automatically transform technical documents into intuitive audiovisual teaching resources.

[0008] In summary, existing enterprise training systems, especially those applied in the new energy power sector, urgently require a system capable of automatically parsing heterogeneous technical documents in various formats and, based on an understanding of the document content, automatically generating standardized, high-quality training materials and instructional videos. This would address the issues of high training costs, slow resource updates, and poor relevance in new energy operation and maintenance scenarios. Summary of the Invention

[0009] The purpose of this invention is to provide an automated method and system for generating training resources for new energy operation and maintenance, which can automatically and accurately convert messy, unstructured industrial field data into high-quality standardized courseware and teaching videos, greatly improving the efficiency and relevance of enterprise training.

[0010] To achieve the above objectives, the present invention is implemented using the following technical solution.

[0011] In a first aspect, the present invention provides a method for automatically generating training resources for new energy operation and maintenance, comprising:

[0012] Obtain original teaching materials for new energy operation and maintenance, store and preprocess them to obtain preprocessed materials;

[0013] Based on the preprocessed material, text recognition and speech recognition are performed to obtain processed text data.

[0014] Based on the text data, data cleaning and standardization are performed, and the data is stored in the system's database to obtain a structured knowledge point set.

[0015] Based on the structured knowledge point set and the preset training objectives, structured PPT courseware files and corresponding teaching video files are automatically generated.

[0016] Furthermore, multimodal resource data is collected and accessed in a unified manner, and original training materials in various formats uploaded by users are received and temporarily stored.

[0017] Furthermore, based on the original teaching materials for new energy operation and maintenance uploaded by users, the built-in parsing engine is used to preprocess the original materials.

[0018] The materials include PDF documents, TXT text, images, audio, and video files.

[0019] Furthermore, the preprocessing of the raw materials using the built-in parsing engine includes: receiving the raw teaching materials for new energy operation and maintenance uploaded by users through a pre-set data interface; firstly, reading the file's metadata and extension, and automatically identifying the file type through the file header signature; based on the identification result, automatically distributing the files to the corresponding preprocessing queues: document image queue or audio / video queue; performing basic preprocessing on the received files, namely, converting video files to MP4 format and text files to UTF-8 encoding, to provide standardized data input for subsequent parsing modules.

[0020] Furthermore, the text recognition processing is based on Baidu's PaddleOCR framework, which is used to perform in-depth analysis and key information extraction of the original training materials, and output structured knowledge points. Specifically, for images or PDF documents in the document image preprocessing queue, PaddleOCR's built-in PP-Structure layout analysis is used to perform region detection on the document images. Through image segmentation algorithms, the page is accurately divided into title area, body text area, table area and image area.

[0021] For the identified text areas and table areas, OCR text recognition technology is used to extract the text content. During this process, the system automatically records the coordinate position and hierarchical relationship of each text block in the original page, including first-level headings, second-level headings, etc., to preserve the logical structure of the document.

[0022] In the method of this invention, an OCR-based text recognition method is used to process the preprocessed material, which can effectively remove non-text elements from the document image preprocessing queue and provide independent image materials for subsequent PPT generation.

[0023] Furthermore, the speech recognition processing is based on OpenAI's Whisper model, used to process audio tracks in audio and video files, converting speech content into text, specifically including:

[0024] For the files in the video preprocessing queue, an Automatic Speech Recognition (ASR) model based on the Transformer architecture is used for processing. First, the audio track is separated and background noise, including wind noise and current noise, is filtered out to improve speech clarity. Then, speech recognition technology is used to convert the processed audio stream into a text sequence.

[0025] The method of this invention employs an ASR-based speech recognition processing method. During the text conversion process, text data with timestamps is automatically generated to ensure that the text content corresponds precisely to the timeline of the original audio, serving as a key synchronization benchmark for subsequent audiovisual synthesis.

[0026] Furthermore, the text data output from text recognition and speech recognition is cleaned and standardized to generate easy-to-use teaching resources. The cleaned text, the corresponding image material paths, and the timestamp information of the audio text are encapsulated into JSON objects and stored in the system's database to form a structured knowledge base that can be called by large language models.

[0027] Furthermore, it receives the structured knowledge point set output from text recognition and speech recognition processing, and automatically generates structured PPT courseware files and corresponding teaching video files based on preset training objectives. The courseware text and the corresponding teaching video files are obtained by the collaborative work of the content planning agent, the PPT generation agent, and the video synthesis agent.

[0028] Furthermore, the content planning intelligent agent, serving as the logical control center for courseware generation, is primarily equipped with a large language model fine-tuned using domain-specific data. This ensures that the generated content outline highly aligns with the professional context and logical requirements of industrial operations and maintenance. The construction and operation of the large language model engine include:

[0029] Based on a large language model pre-trained on a large general corpus, LoRA fine-tuning technology is used to efficiently fine-tune parameters using a corpus containing a large amount of knowledge in the industrial operation and maintenance field; and combined with training objectives, a course content outline and detailed script that conform to the laws of industrial training and teaching are automatically generated.

[0030] In the method of this invention, a content planning intelligent agent is used to enable a large language model to deeply master the professional terminology, knowledge system, safety standard logic and content organization method of specific fields such as photovoltaic operation and maintenance, thus becoming a domain expert model.

[0031] Furthermore, the PPT generation agent is connected to the content planning agent to automatically generate visual PPT courseware based on the content planning script. It includes a PPT template library, a layout engine, and a content rendering engine. The workflow is as follows:

[0032] Based on the course attributes in the course content planning script, a suitable PPT template is automatically matched from the PPT template library or specified by the user. The template predefines theme colors, fonts, master slide styles, and page layout.

[0033] The layout engine analyzes the course content planning script and dynamically creates corresponding PPT pages for each part of the content, such as cover page, table of contents page, chapter title page, content page, summary page, etc.

[0034] The content rendering engine is responsible for automatically filling the titles and body text in the script into the corresponding placeholders on the PPT page, while calling the corresponding image materials in the database to insert illustrations, and finally generating a complete PPT courseware file that can be directly used for playback and editing.

[0035] Furthermore, the video synthesis agent is connected to the PPT generation agent to convert static PPT courseware into audio-visual synchronized teaching videos. It includes a speech generator, a speech synthesis unit, and an audio-visual synthesis engine. The workflow is as follows:

[0036] The speech generator uses the text content and detailed scripts of each page in the PPT courseware to call a large language model for spoken language transcription and expansion, generating a natural and fluent voice-over speech script suitable for explanation.

[0037] The text-to-speech (TTS) unit converts the generated voice-over speech into a high-quality audio file.

[0038] The audio / video synthesis engine performs the following operations:

[0039] Based on the duration of the voice-over audio, the system automatically assigns an appropriate display duration to each page in the PPT, generates a timeline, and adds default or specified transition animations between PPT pages. The PPT page sequence is then synchronized and mixed with the synthesized voice-over audio track, and finally rendered and output as a standard video file.

[0040] The method of this invention utilizes the sequential collaboration of three intelligent agents—content planning agent, PPT generation agent, and video synthesis agent—to automatically and coherently transform knowledge content into clearly structured PPT courseware and accompanying teaching videos. In this process, the method not only emphasizes the natural connection and logical transition between sentences and automatically inserts necessary explanatory content, but also supports flexible adjustments to the timbre, speed, and tone of the voiceover to adapt to different teaching styles and scenario requirements.

[0041] Secondly, the present invention provides an automated generation system for new energy operation and maintenance training resources, comprising:

[0042] The data acquisition and preprocessing module is used to acquire original teaching materials for new energy operation and maintenance, store and preprocess them to obtain preprocessed materials.

[0043] The text data recognition module is used to perform text recognition and speech recognition processing based on the preprocessed material to obtain processed text data;

[0044] The data cleaning and processing module is used to clean and standardize the text data and store it in the system's database to obtain a set of structured knowledge points.

[0045] The automated document generation module is used to automatically generate structured PPT courseware files and corresponding teaching video files based on the structured knowledge point set and preset training objectives.

[0046] Furthermore, the system provided by this invention enables material uploading through a user interface, supports batch uploading and real-time preview, and is used to ensure the accuracy and completeness of data input.

[0047] Thirdly, the present invention provides a computer-readable storage medium having a computer program / instruction stored thereon, wherein when the computer program / instruction is executed by a processor, it implements the steps of the method for automatically generating new energy operation and maintenance training resources as described in any of the first aspects.

[0048] Fourthly, the present invention provides a computer device, comprising:

[0049] Memory, used to store computer programs / instructions;

[0050] A processor is configured to execute the computer program / instructions to implement the steps of the method for automatically generating new energy operation and maintenance training resources as described in any one of the first aspects.

[0051] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0052] This invention employs an automated method for generating training resources for new energy operation and maintenance. It unifies the storage and preprocessing of multimodal heterogeneous data to form standardized data input. Preprocessed document image queues and audio / video queues are then obtained, and OCR-based text recognition and ASR-based speech recognition are performed respectively. The processed text data is extracted, cleaned, and standardized to generate a structured knowledge point set, which is stored in the system's database. Based on preset training objectives, courseware files and accompanying teaching video files are automatically generated. This invention achieves an end-to-end automated content production pipeline through multi-agent collaboration and creatively links multimodal analysis with AI generation technology to construct an automated pipeline from "raw data" to "standardized video courseware," completely changing the reliance on slow manual courseware production and video recording, and greatly improving efficiency.

[0053] The method of this invention lowers the threshold and cost of knowledge transfer, transforms the experience of domain experts into digital courseware that can be replicated on a large scale, and reduces the over-reliance on the professional skills of courseware creators. Attached Figure Description

[0054] Figure 1 This is an overall flowchart of the method for automatically generating training resources for new energy operation and maintenance according to an embodiment of the present invention;

[0055] Figure 2 This is a schematic diagram of the system architecture of an automated generation system for new energy operation and maintenance training resources provided according to an embodiment of the present invention. Detailed Implementation

[0056] It should be noted that:

[0057] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0058] The term "and / or" simply describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0059] Example 1

[0060] like Figure 1 As shown in the figure, this embodiment introduces a method for automatically generating training resources for new energy operation and maintenance, including:

[0061] Obtain original teaching materials for new energy operation and maintenance, store and preprocess them to obtain preprocessed materials;

[0062] Based on the preprocessed material, text recognition and speech recognition are performed to obtain processed text data.

[0063] Based on the text data, data cleaning and standardization are performed, and the data is stored in the system's database to obtain a structured knowledge point set.

[0064] Based on the structured knowledge point set and the preset training objectives, structured PPT courseware files and corresponding teaching video files are automatically generated.

[0065] Furthermore, multimodal resource data is collected and accessed in a unified manner, and original training materials in various formats uploaded by users are received and temporarily stored.

[0066] Furthermore, based on the original teaching materials for new energy operation and maintenance uploaded by users, the built-in parsing engine is used to preprocess the original materials.

[0067] In this embodiment, multimodal resource data includes PDF documents, TXT text, images, audio, and video files.

[0068] Furthermore, the preprocessing of the raw materials using the built-in parsing engine includes: receiving the raw teaching materials for new energy operation and maintenance uploaded by users through a preset data interface; firstly, reading the file's metadata and extension; and automatically identifying the file type through File Signature; based on the identification result, automatically distributing the file to the corresponding preprocessing queue: document image queue or audio / video queue.

[0069] In this embodiment, the received files undergo basic preprocessing. Text files are transcoded and uniformly converted to UTF-8 encoding; video files are format transcoded and uniformly converted to mp4 format to provide standardized data input for the subsequent parsing module.

[0070] Furthermore, the text recognition processing is based on Baidu's PaddleOCR framework, which is used to perform in-depth analysis and key information extraction of the original training materials, and output structured knowledge points. Specifically, for images or PDF documents in the document image preprocessing queue, PaddleOCR's built-in PP-Structure layout analysis is used to perform region detection on the document images. Through image segmentation algorithms, the page is accurately divided into title area, body text area, table area and image area.

[0071] For the identified text area and table area, the text content is extracted using OCR text recognition technology. In this embodiment, the system automatically records the coordinate position and hierarchical relationship of each text block in the original page, including first-level headings and second-level headings, in order to preserve the logical structure of the document.

[0072] Furthermore, the speech recognition processing is based on OpenAI's Whisper model, used to process audio tracks in audio and video files, converting speech content into text, specifically including:

[0073] For the files in the video preprocessing queue, an ASR model based on the Transformer architecture is used for processing. In this embodiment, the audio track is first separated and background noise, including wind noise and current noise, is filtered out to improve speech clarity. Then, speech recognition technology is used to convert the processed audio stream into a text sequence.

[0074] Furthermore, the text data output from text recognition and speech recognition is cleaned and standardized to generate easy-to-use teaching resources. The cleaned text, the corresponding image material paths, and the timestamp information of the audio text are encapsulated into JSON objects and stored in the system's database to form a structured knowledge base that can be called by large language models.

[0075] Furthermore, it receives the structured knowledge point set output from text recognition and speech recognition processing, and automatically generates structured PPT courseware files and corresponding teaching video files based on preset training objectives. The courseware text and the corresponding teaching video files are obtained by the collaborative work of the content planning agent, the PPT generation agent, and the video synthesis agent.

[0076] Furthermore, the content planning intelligent agent, serving as the logical control center for courseware generation, is primarily equipped with a large language model fine-tuned using domain-specific data. This ensures that the generated content outline highly aligns with the professional context and logical requirements of industrial operations and maintenance. The construction and operation of the large language model engine include:

[0077] In this embodiment, based on the Qwen3-2507 basic model, LoRA parameter efficient fine-tuning technology is employed. Training is performed using a corpus containing extensive knowledge of industrial operations and maintenance, aligning the model's output distribution towards the target domain. This method reduces domain error through supervised loss, enabling small-scale customization of response style, tone, or functionality, and adjusting the target weight matrix. LoRA learns two low-rank matrices , In inference, the weights are updated with a lower rank, expressed as:

[0078]

[0079] in, , Scaling factor The original model weight matrix, This is the updated weight matrix.

[0080] In this embodiment, the fine-tuning training data comes from a domain corpus composed of textual materials such as the "Power Safety Work Regulations" and the "Photovoltaic Power Plant Operation and Maintenance Manual". The fine-tuning process enables the large language model to deeply master the professional terminology, knowledge system, safety standard logic and content organization methods of specific fields such as photovoltaic operation and maintenance, thus becoming a domain expert model.

[0081] Based on the training objectives, a course outline and detailed script that conform to the principles of industrial training are automatically generated. Compared with general models that rely solely on cue word constraints, the LoRA-tuned model in this embodiment internalizes domain logic at the parameter level, resulting in a qualitative improvement in the professionalism, accuracy, and process compliance of the generated content.

[0082] Furthermore, the PPT generation agent is connected to the content planning agent to automatically generate visual PPT courseware based on the content planning script. It includes a PPT template library, a layout engine, and a content rendering engine. The workflow is as follows:

[0083] Based on the course attributes in the course content planning script, a suitable PPT template is automatically matched from the PPT template library or specified by the user. This embodiment is designed based on the visual recognition requirements of the new energy operation and maintenance field, and defines the theme colors, fonts, master slide styles, and page layout for the PPT template:

[0084] The theme color scheme uses blue, representing clean energy and technology, supplemented by yellow and red for safety warnings and neutral gray. This color system is fixed in the template and automatically applied to the background, title bar, and text accent colors of all pages to ensure visual consistency and industry adaptability of the courseware.

[0085] The font settings include a pairing scheme for Chinese and Western fonts, where the Chinese font is a bold sans-serif font to ensure clear screen display; the Western font is Times New Roman; and the font size and weight of different levels of text are clearly defined in the template, with first-level headings using 44-point bold, second-level headings using 32-point bold, and body text using 20-point bold.

[0086] The master style defines the logical structure and basic layout of the courseware pages, including a variety of preset page types, such as cover page, table of contents page, chapter title page, content page and summary page. In this embodiment, each page type has a preset fixed placeholder layout.

[0087] The page layout refers to the spatial arrangement rules of placeholders on each page;

[0088] The layout engine parses the course content planning script and dynamically calls the corresponding page types in the master page according to the chapter structure in the script to generate a page sequence.

[0089] The content rendering engine is responsible for filling the text content in the script into the corresponding placeholders on each page, querying the structured database, calling the corresponding image materials in the database, inserting illustrations, and finally generating a complete PPT courseware file that can be directly used for playback and editing.

[0090] In this embodiment, the entire PPT generation process requires no manual intervention, achieving end-to-end conversion from structured data to visual courseware.

[0091] Furthermore, the video synthesis agent is connected to the PPT generation agent to convert static PPT courseware into audio-visual synchronized teaching videos. It includes a speech generator, a speech synthesis unit, and an audio-visual synthesis engine. The workflow is as follows:

[0092] The speech generator is based on the text content and detailed content script of each page in the PPT courseware. In this embodiment, the agent sends the text content and outline details of each page in the PPT to the fine-tuned Qwen3-2507 model and issues the instruction "to convert the following content into a conversational speech suitable for explanation".

[0093] TTS converts the generated voice-over speech into a high-quality audio file. The audio and video synthesis engine performs the following operations:

[0094] Based on the duration of the voice-over audio, the system automatically assigns an appropriate display duration to each page in the PPT, generates a timeline, and adds default or specified transition animations between PPT pages. The PPT page sequence is then synchronized and mixed with the synthesized voice-over audio track, and finally rendered and output as a standard video file.

[0095] In this embodiment, the final generated teaching videos and PPT courseware are automatically stored in the training resource library, the generation path is recorded, and an index relationship is established with the original PDF documents, video files and parsed structured knowledge points used. When the required training content is updated in the future, the system can quickly locate and prompt that the training video needs to be regenerated.

[0096] In summary, this method successfully transforms unstructured multimodal data into high-quality, standardized training content through a highly automated intelligent pipeline, effectively addressing pain points in corporate training.

[0097] Example 2

[0098] like Figure 2 As shown, based on the automated generation method for new energy operation and maintenance training resources described in Example 1, this example introduces an automated generation system for new energy operation and maintenance training resources, including:

[0099] The data acquisition and preprocessing module is used to acquire original teaching materials for new energy operation and maintenance, store and preprocess them to obtain preprocessed materials.

[0100] The text data recognition module is used to perform text recognition and speech recognition processing based on the preprocessed material to obtain processed text data;

[0101] The data cleaning and processing module is used to clean and standardize the text data and store it in the system's database to obtain a set of structured knowledge points.

[0102] The automated document generation module is used to automatically generate structured PPT courseware files and corresponding teaching video files based on the structured knowledge point set and preset training objectives.

[0103] Furthermore, in this embodiment, the system enables material uploading through a user interface, supporting batch uploading and real-time preview to ensure the accuracy and completeness of data input.

[0104] Example 3

[0105] Based on the automated generation method for new energy operation and maintenance training resources described in Embodiment 1, this embodiment introduces a computer-readable storage medium storing a computer program / instruction thereon. When the computer program / instruction is executed by a processor, it implements the steps of the automated generation method for new energy operation and maintenance training resources as described in any of Embodiment 1.

[0106] Example 4

[0107] Based on the automated generation method for new energy operation and maintenance training resources described in Example 1, this example provides a computer device, including:

[0108] Memory, used to store computer programs / instructions;

[0109] A processor is used to execute the computer program / instructions to implement the steps of the method for automatically generating new energy operation and maintenance training resources as described in any one of Embodiment 1.

[0110] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0111] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0112] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0113] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0114] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A method for automatically generating training resources for new energy operation and maintenance, characterized in that, include: Obtain original teaching materials for new energy operation and maintenance, store and preprocess them to obtain preprocessed materials; Based on the preprocessed material, text recognition and speech recognition are performed to obtain processed text data. Based on the text data, data cleaning and standardization are performed, and the data is stored in the system's database to obtain a structured knowledge point set. Based on the structured knowledge point set and the preset training objectives, structured PPT courseware files and accompanying teaching video files are automatically generated.

2. The method for automatically generating training resources for new energy operation and maintenance according to claim 1, characterized in that, The original teaching materials for new energy operation and maintenance include PDF documents, TXT text, images, audio and video files.

3. The method for automatically generating training resources for new energy operation and maintenance according to claim 1, characterized in that, The storage and preprocessing process includes: Read the metadata and file extension of teaching material files and automatically identify the file type; Based on the identified file type, the file is distributed to the document image preprocessing queue or the audio / video preprocessing queue; Perform file transcoding processing, converting text files to UTF-8 encoding and audio / video files to mp4 format.

4. The method for automatically generating training resources for new energy operation and maintenance according to claim 1, characterized in that, The text recognition is based on Baidu's PaddleOCR framework, used for deep analysis and key information extraction of the original training materials, including: Region detection is performed on the images and PDF documents in the document image preprocessing queue. Through image segmentation algorithm, the page is accurately divided into title area, body text area, table area and image area. Based on the identified text area and table area, the text content is extracted using OCR text recognition technology.

5. The method for automatically generating training resources for new energy operation and maintenance according to claim 1, characterized in that, The speech recognition processing is based on OpenAI's Whisper model, used to process audio tracks in audio and video files, converting speech content into text, including: The audio tracks of the files in the audio and video preprocessing queue are extracted. After background noise is filtered out and purified, the processed audio stream is converted into a text sequence using speech recognition technology.

6. The method for automatically generating training resources for new energy operation and maintenance according to claim 1, characterized in that, The standardization process is used to construct a structured knowledge base for use by large language models, including: The cleaned text, the corresponding image material paths, and the timestamp information of the audio text are encapsulated into a JSON object and stored in the system's database.

7. The method for automatically generating training resources for new energy operation and maintenance according to claim 1, characterized in that, The automatic generation of PPT presentation files and accompanying instructional video files is achieved collaboratively by a content planning agent, a PPT generation agent, and a video synthesis agent. The content planning agent, acting as the logical control center for courseware generation, uses a large language model fine-tuned with domain-specific data and combines it with training objectives to automatically generate a course content outline and detailed script that conforms to the laws of industrial training and teaching. The PPT generation agent is connected to the content planning agent and is used to automatically generate visual PPT courseware based on the course content planning script, including a PPT template library, layout engine and content rendering engine. The video synthesis agent is connected to the PPT generation agent and is used to convert the generated static PPT courseware into teaching videos with synchronized audio and video. It includes a speech generator, a speech synthesis unit, and an audio and video synthesis engine.

8. An automated resource generation system for new energy operation and maintenance training, characterized in that, include: The data acquisition and preprocessing module is used to acquire original teaching materials for new energy operation and maintenance, store and preprocess them to obtain preprocessed materials. The text data recognition module is used to perform text recognition and speech recognition processing based on the preprocessed material to obtain processed text data; The data cleaning and processing module is used to clean and standardize the text data and store it in the system's database to obtain a set of structured knowledge points. The automated document generation module is used to automatically generate structured PPT courseware files and accompanying teaching video files based on the structured knowledge point set and preset training objectives.

9. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instruction is executed by the processor, it implements the steps of the method for automatically generating new energy operation and maintenance training resources as described in any one of claims 1 to 7.

10. A computer device / equipment / system, characterized in that, include: Memory, used to store computer programs / instructions; A processor is configured to execute the computer program / instructions to implement the steps of the method for automatically generating new energy operation and maintenance training resources as described in any one of claims 1 to 7.

Images