Intelligent teaching plan generation system based on large model and knowledge base for young teachers

Abstract

The invention discloses an intelligent teaching plan generation system for young teachers, and relates to the technical field of teaching plan generation. A demand is obtained through multi-modal interaction, a hypergraph knowledge graph knowledge base is used for supporting, and a teaching plan containing emotional education is generated through fusion of a large language model and graph neural network processing; the teaching plan creation is comprehensively assisted by adjusting according to the child state and the teaching emotional tendency and by means of multi-dimensional evaluation feedback optimization. The teaching plan generation efficiency is greatly improved, and the teacher time is saved. A scientific and personalized teaching plan is provided, and emotion education is integrated; and the new dynamic state is updated and adapted in real time, feedback optimization is performed according to multi-dimensional evaluation, and the child teaching quality is comprehensively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of teaching plan generation systems, and in particular to an intelligent teaching plan generation system for preschool teachers based on a large model and a knowledge base. Background Art

[0002] In the field of early childhood education, lesson plan design is an important foundation for teaching. A high-quality lesson plan can effectively improve teaching quality and promote the all-round development of children. However, preschool teachers currently face many challenges in the process of lesson plan creation.

[0003] Traditional lesson plan generation relies primarily on preschool teachers' personal experience and manual compilation. This not only consumes considerable time and effort, but also results in inconsistent quality due to the limitations of their knowledge and experience. Some preschool teachers may lack systematic pedagogical theory and extensive practical teaching experience, resulting in unscientific and inappropriate lesson plans that fail to meet the diverse learning needs of children. Furthermore, manual lesson plan compilation is inefficient, making it difficult for preschool teachers to quickly generate appropriate lesson plans when faced with unexpected adjustments to teaching tasks or content updates.

[0004] With the development of information technology, a number of lesson plan generation tools have emerged. However, most of these tools are limited in functionality, offering only templates and fixed teaching content. They lack a deep understanding of early childhood education and personalized support. They are unable to flexibly adapt to the specific needs of preschool teachers, the characteristics of children, and the changing teaching environment. Furthermore, the knowledge bases used by these tools are not updated promptly, failing to keep pace with the latest research findings and changes in policies and regulations in the field of early childhood education. This results in the generated lesson plans lacking timeliness and innovation.

[0005] At the same time, existing lesson plan generation tools rarely consider the emotional development needs of young children. In early childhood education, emotional education is crucial for children's mental health and the development of their social skills. However, current tools fail to design appropriate emotional education goals and implementation strategies based on the teaching content and children's emotional characteristics, resulting in inadequate lesson plans in promoting children's emotional development.

[0006] Therefore, developing an intelligent lesson plan generation system for preschool teachers based on a large model and knowledge base is of great practical significance. This system can fully utilize the powerful language processing capabilities of the large model and the rich resources of the knowledge base to provide preschool teachers with an efficient, personalized, and scientific lesson plan generation solution, while also taking into account the emotional education needs of young children and promoting the improvement of early childhood education and teaching quality. Summary of the Invention

[0007] The present invention proposes an intelligent teaching plan generation system for preschool teachers based on a large model and a knowledge base to solve the problems mentioned in the above-mentioned prior art.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: an intelligent teaching plan generation system based on a large model and a knowledge base for preschool teachers, comprising the following modules:

[0009] Demand Acquisition Module: This module supports preschool teachers expressing their lesson plan generation needs through text input, voice description, and image upload. It also incorporates gesture recognition and eye tracking technologies. The system utilizes semantic understanding algorithms to not only extract teaching topics (T), teaching duration (D), and child age (A), but also analyzes preschool teachers' emotional tendencies (E), weighted through an attention mechanism.

[0010] Knowledge base module: integrates early childhood education resources, including teaching cases, teaching methods, and early childhood development knowledge; uses hypergraph knowledge graph technology to build a knowledge network, and uses quantum-inspired semantic similarity algorithms Retrieve knowledge related to demand in the knowledge space, where q i is the quantum state weighting factor, s i is the local similarity;

[0011] Large model processing module: This module uses a deep fusion of pre-trained large language models and graph neural networks to deeply integrate demand information with relevant knowledge in the knowledge base. Meta-learning techniques are used to fine-tune the large model, and the policy gradient algorithm used in reinforcement learning is combined to optimize the model's generation strategy. By introducing an exploration-exploitation balance mechanism, the module generates a lesson plan framework F that meets the teaching objectives and is innovative in children's cognitive level. Furthermore, the model adjusts the emotional tone of the lesson plan framework based on the teaching emotional tendency E.

[0012] Lesson plan generation module: Based on the lesson plan framework F output by the large model processing module and the specific content in the knowledge base, a complete lesson plan is generated. In addition to teaching objectives, teaching content, teaching methods, and teaching processes, the lesson plan also includes emotional education goals and implementation strategies. The system will reasonably allocate time for each teaching link based on the teaching time D and the children's attention concentration curve C(t). Through the time dynamic planning algorithm, Ensure that the teaching process is orderly and in line with the cognitive laws of young children; at the same time, it can generate corresponding emotional interaction links based on different teaching emotional tendencies E.

[0013] Furthermore, the following modules are also included:

[0014] Personalized recommendation module: This module builds personalized profiles of preschool teachers based on their historical usage records, teaching style preferences, and class information. In addition to using collaborative filtering algorithms and deep learning models, it also introduces an optimized recommendation mechanism based on genetic algorithms to search for recommended solutions in the solution space.

[0015] Evaluation and feedback module: Use evaluation indicators to evaluate the generated teaching plans, including teaching goal achievement G, content suitability S c , Method effectiveness E m , emotional education effect E e ; Determine the weight of each indicator by combining fuzzy hierarchical analysis method with evidence theory and calculate the comprehensive evaluation score where α i is the comprehensive weight.

[0016] Real-time update module: The module monitors the dynamics, policies and regulations, and research results in the field of early childhood education in real time, and uses information extraction technology in natural language processing to automatically extract updated content.

[0017] Furthermore, the demand acquisition module also supports docking with kindergarten hardware equipment, including wristbands and classroom sensors, to automatically obtain children's physiological status data.

[0018] Furthermore, the knowledge base module adopts quantum encrypted distributed storage and quantum parallel computing technology to improve the storage capacity and retrieval speed of knowledge.

[0019] Furthermore, the large model processing module adopts fusion technology to process image, audio and video data in addition to text information.

[0020] Furthermore, the teaching plan generation module supports Word, PDF, and PPT, and also supports the generation of interactive teaching plans, realizing dynamic display and interaction on the teaching terminal.

[0021] Furthermore, the evaluation and feedback module is integrated with the online teaching platform to collect children's feedback data during the teaching process in real time, including facial expressions, voice intonation, and body movements.

[0022] Furthermore, the personalized recommendation module adopts federated learning technology combined with homomorphic encryption, and uses kindergarten data for joint training while protecting the privacy of kindergarten teachers.

[0023] Furthermore, the real-time update module uses an incremental learning algorithm combined with online meta-learning to update and train only the parts of the knowledge base and the large model that have changed. This reduces computing resource consumption through an adaptive learning rate adjustment strategy.

[0024] Compared with the existing technology, the beneficial effects of the present invention are:

[0025] In terms of efficiency, the system significantly saves preschool teachers time and energy. Through a multimodal interactive interface, preschool teachers can conveniently input lesson plan requirements. The system uses advanced algorithms to quickly understand these requirements and generate a lesson plan framework. This system then integrates the knowledge base content to produce a complete lesson plan, significantly speeding up lesson plan generation. Compared to traditional manual writing methods, this system can shorten lesson plan generation time several times, freeing preschool teachers to devote more time to practical teaching and interaction with children.

[0026] In terms of quality, the system is highly scientific and personalized. The large model has been deeply fine-tuned, combined with a hypergraph knowledge graph and a quantum-inspired semantic similarity algorithm. This allows it to accurately retrieve and integrate knowledge from the knowledge base, generating lesson plans that align with teaching objectives and children's cognitive levels. Furthermore, based on the personalized profile of preschool teachers and the real-time status of children, the system can provide personalized lesson plan templates and teaching resources to meet the needs of different teaching scenarios. Furthermore, the system incorporates an emotional education dimension, enabling the design of corresponding emotional interaction sessions based on the emotional tendencies of teaching, promoting children's emotional development.

[0027] In terms of updating and adaptability, the real-time update module can promptly access the latest developments in early childhood education, updating and optimizing the knowledge base and large models. Incremental learning algorithms and adaptive strategies ensure that the system maintains the timeliness and innovation of generated lesson plans while consuming minimal computing resources. Furthermore, the system can be tailored to the geographical and cultural characteristics of different kindergartens, ensuring broad applicability.

[0028] In terms of evaluation and feedback, multi-dimensional assessment indicators and a dynamic adjustment mechanism accurately assess the quality of lesson plans and provide detailed feedback and suggestions. Combined with real-time feedback data from the online teaching platform, the system can continuously optimize evaluation indicators and weightings, improving the accuracy and effectiveness of assessments and helping preschool teachers continuously improve their lesson plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a schematic block diagram of an intelligent teaching plan generation system for preschool teachers based on a large model and knowledge base proposed by the present invention;

[0030] Figure 2 This is a schematic diagram comparing the efficiency of intelligent teaching plan generation based on a large model and a knowledge base for preschool teachers proposed by the present invention;

[0031] Figure 3 This is a schematic diagram showing the comparison of the degree of achievement of teaching objectives for an intelligent teaching plan generated for preschool teachers based on a large model and a knowledge base, as proposed by the present invention. DETAILED DESCRIPTION

[0032] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0033] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like to indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0034] In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the said features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. In addition, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. The present invention will be further described in detail below with reference to the accompanying drawings.

[0035] Reference Figure 1-3 : An intelligent teaching plan generation system for preschool teachers based on a large model and knowledge base, including the following modules:

[0036] The Requirements Acquisition Module features a multimodal interactive interface. In addition to supporting traditional methods such as text input, voice description, and image uploads for expressing lesson plan requirements, it also incorporates gesture recognition and eye tracking technology. Gesture recognition achieves millimeter-level accuracy, accurately identifying specific gestures directed by teachers. Eye tracking captures the teacher's gaze in real time, combining it with natural language processing to deeply analyze the input. The system utilizes enhanced semantic understanding algorithms to not only extract key information such as the teaching topic (T), teaching duration (D), and child age (A), but also analyzes the teacher's underlying emotional orientation (E). This weighted processing, achieved through an attention mechanism, significantly improves the accuracy of requirement comprehension. This module serves as the starting point for lesson plan generation, acting as a sensitive "sensor." The multimodal interactive interface provides teachers with diverse channels for expression, while gesture recognition and eye tracking act as precise "capture devices," discerning the teacher's subtle intentions. The enhanced semantic understanding algorithm is like an intelligent "analyzer". It not only extracts key teaching information, but also explores potential emotional tendencies, just like reading the teacher's inner thoughts. Through the weighted processing of the attention mechanism, it greatly improves the accuracy of understanding teachers' needs, laying a solid foundation for subsequent lesson plan generation.

[0037] Knowledge base module: It integrates a wealth of early childhood education resources, covering teaching cases, teaching methods, early childhood development characteristics and other knowledge. It uses hypergraph knowledge graph technology to build a knowledge network. Hypergraph can more flexibly represent multiple relationships. Nodes not only represent knowledge entities, but also reflect the attribute hierarchy of entities. Through quantum-inspired semantic similarity algorithm (where q i is the quantum state weighting factor, s i is local similarity), quickly and accurately retrieves knowledge related to needs in high-dimensional knowledge space, providing a solid foundation for lesson plan generation. At the same time, the knowledge base introduces the emotional knowledge dimension to record children's emotional feedback corresponding to different teaching contents. This module is a "resource treasure house" for lesson plan generation, integrating massive early childhood education resources. Hypergraph knowledge graph technology is like a sophisticated "weaver", building a complex and flexible knowledge network, clearly presenting the multiple relationships and attribute hierarchies between knowledge. The quantum-inspired semantic similarity algorithm is like an efficient "searcher", which can quickly locate relevant knowledge in high-dimensional knowledge space. The introduced emotional knowledge dimension records children's emotional feedback, which is like adding "emotional warmth" to the resource treasure house, making the retrieved knowledge more in line with teaching practice.

[0038] The large model processing module utilizes a deeply integrated pre-trained large language model and graph neural network architecture to deeply integrate demand information with relevant knowledge in the knowledge base. Meta-learning technology allows for rapid fine-tuning of the large model, enabling it to adapt to the diverse characteristics of early childhood education in a very short time. The model's generation strategy is optimized using the policy gradient algorithm from reinforcement learning. By incorporating an exploration-exploitation balance mechanism, the model generates innovative lesson plan frameworks F that align with teaching objectives and children's cognitive levels. Furthermore, the model can adjust the emotional tone of the lesson plan framework based on the teaching emotional orientation E. This module serves as the "intelligent brain" for lesson plan generation, deeply integrating the pre-trained large language model with the graph neural network architecture. It acts as a powerful "integrator" for deeply integrating demand and knowledge. Meta-learning technology acts as a flexible "adapter," enabling the model to quickly adapt to the specific characteristics of early childhood education. The policy gradient algorithm and exploration-exploitation balance mechanism act as intelligent "creators," generating innovative lesson plan frameworks that align with teaching objectives and children's cognitive abilities. They can also adjust the emotional tone based on emotional orientation, making the lesson plans more humanistic.

[0039] Lesson plan generation module: Based on the lesson plan framework F output by the large model processing module and the specific content in the knowledge base, a complete lesson plan is generated. In addition to teaching objectives, teaching content, teaching methods, and teaching processes, the lesson plan also includes emotional education goals and implementation strategies. The system will reasonably allocate time for each teaching link based on the teaching time D and the children's attention concentration curve C(t). Through the time dynamic planning algorithm, This ensures a cohesive and organized teaching process that aligns with children's cognitive patterns. Furthermore, it generates corresponding emotional interaction sessions based on different teaching emotional tendencies (E). Based on the framework output by the large model and the specific content of the knowledge base, complete lesson plans are meticulously crafted. In addition to conventional teaching elements, emotional education goals and strategies are incorporated, reflecting a focus on children's holistic development. Based on teaching duration and children's attention spans, a time-based dynamic planning algorithm is used, acting like a precise "planner," rationally planning teaching sessions to ensure an efficient and consistent learning process.

[0040] Evaluation and feedback module: Use multi-dimensional evaluation indicators to evaluate the generated teaching plans, such as teaching goal achievement G, content suitability S c , Method effectiveness E m , emotional education effect E e Etc. The weight of each indicator is determined by combining fuzzy analytic hierarchy process with evidence theory, and the comprehensive evaluation score is calculated. where α iBased on the evaluation results, detailed feedback is provided to preschool teachers, including recommendations for improvements to content and methods, as well as optimization of emotional education. The system also utilizes generative adversarial networks to verify the evaluation results, enhancing their reliability. Innovatively integrated with online teaching platforms, the system collects multimodal feedback from children, breaking away from traditional single-model evaluation methods. Dynamically adjust evaluations using affective computing and multimodal machine learning.

[0041] The present invention also includes:

[0042] Personalized recommendation module: The module constructs a multi-dimensional personalized portrait of preschool teachers based on their historical usage records, teaching style preferences, and preschool class characteristics. In addition to using collaborative filtering algorithms and deep learning models, it also introduces an optimization recommendation mechanism based on genetic algorithms to search for the best recommendation solution in the solution space. Utilizing knowledge transfer learning technology, the recommendation experience from different kindergartens and different teaching scenarios is transferred to recommend lesson plan templates and teaching resources that meet the personalized needs of preschool teachers. At the same time, it recommends emotional education materials that match the emotional tendencies of teaching, thereby improving the user experience and lesson plan generation efficiency of preschool teachers. Federated learning is combined with homomorphic encryption and differential privacy mechanisms to jointly train data from multiple kindergartens under privacy protection, breaking down data silos while ensuring privacy. It can make recommendations based on real-time teaching scenarios and emotional dynamics.

[0043] Real-time Update Module: This module monitors the latest developments, policies, regulations, research findings, and other information in the field of early childhood education in real time, automatically extracting key updates using information extraction techniques from natural language processing. An active learning strategy is used to update the content in the knowledge base, prioritizing knowledge with high uncertainty. Simultaneously, large models are continuously trained and optimized based on new data, and model distillation technology is introduced to migrate knowledge from large models to lightweight models. This reduces computing resource consumption while ensuring system performance, ensuring that the lesson plans generated by the system are timely and scientific. Incremental learning and online meta-learning are used to update only the changed parts, avoiding resource waste. The learning rate is adaptively adjusted to improve update efficiency. Targeted optimization can also be implemented based on the regional and cultural characteristics of different kindergartens.

[0044] In this invention, the demand acquisition module also supports integration with the kindergarten's intelligent hardware devices, such as smart bracelets and smart classroom sensors, to automatically acquire children's physiological status data, such as heart rate and attention concentration. This data is analyzed using a recurrent neural network in deep learning. Combined with the child's basic information and learning progress, this data provides a deeper understanding of the child's real-time status and needs, providing more accurate and real-time information for lesson plan generation.

[0045] In this invention, the knowledge base module utilizes quantum-encrypted distributed storage and quantum parallel computing technology to significantly increase knowledge storage capacity and retrieval speed. The principle of quantum entanglement is used to ensure the security and immutability of knowledge data, while quantum error-correcting code technology ensures the reliability of knowledge. Furthermore, a knowledge evolution mechanism is introduced to simulate the dynamic development of knowledge in educational practice, enabling the knowledge base to be adaptively updated and improved.

[0046] In this invention, the large model processing module utilizes multimodal fusion technology, enabling it to process multimodal data such as images, audio, and video in addition to text. By combining a cross-modal attention mechanism with a multimodal fusion strategy based on a generative adversarial network, it can better integrate information from different modalities and improve the quality of lesson plan framework generation. Furthermore, the model can further optimize the emotional design of the lesson plan framework based on emotional cues in the multimodal data.

[0047] In this invention, the lesson plan generation module supports lesson plan output in multiple formats, such as Word, PDF, and PPT. It also supports the generation of interactive lesson plans, enabling dynamic display and interaction on intelligent teaching terminals. It provides a visual editing interface and uses virtual reality and augmented reality technologies, allowing preschool teachers to immersively edit and modify the generated lesson plans, making it easy to adjust the presentation of teaching content and the emotional interaction aspects.

[0048] In this invention, the assessment and feedback module is integrated with the online teaching platform, enabling real-time collection of multimodal feedback data from children during the teaching process, such as facial expressions, voice intonation, and body movements. Through affective computing and multimodal machine learning algorithms, assessment indicators and weights are dynamically adjusted to improve the accuracy and effectiveness of the assessment. Furthermore, the system can generate emotional growth reports based on the feedback data, providing a reference for preschool teachers' subsequent teaching.

[0049] In this paper, the personalized recommendation module utilizes federated learning combined with homomorphic encryption to perform joint training using data from multiple kindergartens while protecting the privacy of preschool teachers. A differential privacy mechanism is introduced to protect the privacy of training data, improving the generalization and accuracy of the recommendation model. Furthermore, the recommendation module can dynamically make recommendations based on the preschool teachers' real-time teaching scenarios and emotional states.

[0050] In the present invention, the real-time update module uses an incremental learning algorithm combined with online meta-learning to update and train only the parts of the knowledge base and large model that have changed. Through an adaptive learning rate adjustment strategy, the consumption of computing resources is reduced, and the update efficiency of the system is improved. At the same time, targeted knowledge updates and model optimization can be carried out according to the regional and cultural characteristics of different kindergartens. Using incremental learning and online meta-learning, only the changed parts are updated, avoiding resource waste. The learning rate is adaptively adjusted to improve update efficiency. Targeted optimization can also be carried out based on the regional and cultural characteristics of different kindergartens.

[0051] Beneficial effect data characterization:

[0052]

[0053] Through the above specific implementation methods and data comparison, it can be seen that this system has significant advantages in teaching plan generation efficiency, quality, and support for children's emotional education, and can effectively improve the quality of early childhood education and teaching.

[0054] The above are only preferred specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solutions and inventive concepts of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. An intelligent teaching plan generation system for preschool teachers based on a large model and knowledge base, characterized by: include: Demand Acquisition Module: This module supports preschool teachers expressing their lesson plan generation needs through text input, voice description, and image upload. It also incorporates gesture recognition and eye tracking technologies. The system utilizes semantic understanding algorithms to not only extract teaching topics (T), teaching duration (D), and child age (A), but also analyzes preschool teachers' emotional tendencies (E), weighted through an attention mechanism. Knowledge base module: integrates early childhood education resources, including teaching cases, teaching methods, and early childhood development knowledge; uses hypergraph knowledge graph technology to build a knowledge network, and uses quantum-inspired semantic similarity algorithms Retrieve knowledge related to demand in the knowledge space, where q i is the quantum state weighting factor, s i is the local similarity; Large model processing module: This module uses a deep fusion pre-trained large language model and graph neural network architecture to deeply integrate demand information and relevant knowledge in the knowledge base; Meta-learning techniques are used to fine-tune the large model, combined with the policy gradient algorithm from reinforcement learning to optimize the model's generation strategy. By introducing an exploration-exploitation balance mechanism, a lesson plan framework F is generated that meets the teaching objectives and is innovative and responsive to children's cognitive level. Furthermore, the model adjusts the emotional tone of the lesson plan framework based on the teaching emotional tendency E. Lesson plan generation module: Based on the lesson plan framework F output by the large model processing module and the specific content in the knowledge base, a complete lesson plan is generated; in addition to teaching objectives, teaching content, teaching methods, and teaching processes, the lesson plan also includes emotional education goals and implementation strategies; the system will reasonably allocate the time for each teaching link based on the teaching time D and the children's attention concentration curve C(t), and through the time dynamic planning algorithm The teaching process is orderly and in line with the cognitive laws of young children; at the same time, corresponding emotional interaction links can be generated according to different teaching emotional tendencies E.

2. The intelligent teaching plan generation system based on large model and knowledge base for preschool teachers according to claim 1 is characterized in that: Also includes: Personalized recommendation module: This module builds a personalized profile of preschool teachers based on their historical usage records, teaching style preferences, and preschool class information; In addition to using collaborative filtering algorithms and deep learning models, an optimization recommendation mechanism based on genetic algorithms is also introduced to search for recommendation solutions in the solution space; Evaluation and feedback module: Use evaluation indicators to evaluate the generated teaching plans, including teaching goal achievement G, content suitability S c , Method effectiveness E m , emotional education effect E e ; Determine the weight of each indicator by combining fuzzy hierarchical analysis method with evidence theory and calculate the comprehensive evaluation score where α i is the comprehensive weight.

3. The intelligent teaching plan generation system based on large model and knowledge base for preschool teachers according to claim 1 is characterized in that: Also includes: Real-time update module: The module monitors the dynamics, policies and regulations, and research results in the field of early childhood education in real time, and uses information extraction technology in natural language processing to automatically extract updated content.

4. The intelligent teaching plan generation system based on large model and knowledge base for preschool teachers according to claim 1 is characterized in that: The demand acquisition module also supports docking with kindergarten hardware equipment, including wristbands and classroom sensors, to automatically obtain children's physiological status data.

5. The intelligent teaching plan generation system based on large model and knowledge base for preschool teachers according to claim 1 is characterized in that: The knowledge base module adopts quantum encrypted distributed storage and quantum parallel computing technology to improve the storage capacity and retrieval speed of knowledge.

6. The intelligent teaching plan generation system based on large model and knowledge base for preschool teachers according to claim 1 is characterized in that: The large model processing module adopts fusion technology to process image, audio and video data in addition to text information.

7. The intelligent teaching plan generation system based on large model and knowledge base for preschool teachers according to claim 1 is characterized in that: The teaching plan generation module supports Word, PDF, and PPT, and also supports the generation of interactive teaching plans, realizing dynamic display and interaction on the teaching terminal.

8. The intelligent teaching plan generation system based on large model and knowledge base for preschool teachers according to claim 2 is characterized in that: The evaluation and feedback module is integrated with the online teaching platform to collect children's feedback data during the teaching process in real time, including facial expressions, voice intonation, and body movements.

9. The intelligent teaching plan generation system based on large model and knowledge base for preschool teachers according to claim 2 is characterized in that: The personalized recommendation module adopts federated learning technology combined with homomorphic encryption, and is jointly trained using data from various kindergartens while protecting the privacy of kindergarten teachers.

10. The intelligent teaching plan generation system based on large model and knowledge base for preschool teachers according to claim 3 is characterized in that: The real-time update module adopts an incremental learning algorithm combined with online meta-learning to update and train only the parts of the knowledge base and the large model that have changed; and reduces the consumption of computing resources through an adaptive learning rate adjustment strategy.

Images