Automobile production teaching evaluation optimization system based on large model

Abstract

The invention specifically relates to the technical field of multi-modal large models, and discloses an automobile production teaching evaluation optimization system based on a large model, and the system comprises an industry knowledge obtaining module, a knowledge graph construction module, a teaching resource generation module, a capability data collection module, an intelligent capability evaluation module, and a resource matching cooperation module. An industry knowledge acquisition module acquires an automobile production and teaching fusion dynamic knowledge data set, a knowledge graph construction module constructs an automobile production and teaching knowledge graph, a teaching resource generation module generates teaching resources, and a capability data acquisition module acquires learning capability associated data of students. Student ability comprehensive scores are calculated through the intelligent ability evaluation module, personalized learning paths and learning materials are recommended for students through the resource matching cooperation module, talent lists are recommended for enterprises, automatic and high-frequency updating of teaching content is achieved, and intelligent scheduling of the whole process and intelligent upgrading of automobile production teaching evaluation are achieved. And the production and education docking efficiency is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of multimodal large models, and more specifically, to an automobile production-education-evaluation optimization system based on a large model. Background Art

[0002] With the increasing market penetration of intelligent connected vehicles, the automotive industry is developing towards intelligence and networking, and the knowledge and skills requirements for relevant talents are also increasing. The traditional industry-education-assessment model is difficult to meet the requirements of talent training. With the development of artificial intelligence technology, it provides a technical foundation for the development of the automotive industry-education-assessment optimization system. The advancement of digital education has promoted the deep integration of artificial intelligence technology and teaching practice, and improved the teaching quality and talent training effect.

[0003] The existing automobile production, education and evaluation optimization system includes a teaching management module, a production practice module, an evaluation feedback module, and a resource optimization and collaboration module. Through modular design, it achieves a deep integration of teaching, production and evaluation, effectively improving the quality of talent training and providing technical and skilled talents for enterprises.

[0004] However, in actual use, it still has some shortcomings. First, the teaching adaptability is insufficient. The teaching content and production practice cases of the existing automobile production-education-evaluation optimization system mainly rely on manual input, lacking an automated knowledge dynamic iteration mechanism, resulting in lagging knowledge updates and difficulty in matching the speed of technological change in the automotive industry. Second, the evaluation feedback dimension is limited. Existing automotive industry-education evaluation optimization systems rely heavily on standardized test questions and fixed indicators for scoring operational processes. These evaluation dimensions focus on knowledge retention and basic operations, lacking in-depth analysis of complex capabilities. Furthermore, the lack of depth in analysis makes it difficult to achieve accurate talent capability assessment. Third, resource coordination efficiency is low. The existing automobile production-education-evaluation optimization system lacks the ability to intelligently match and dynamically schedule the needs of teachers, students, and enterprises. Resource coordination relies on manual coordination, resulting in low resource matching efficiency and high error rate. It is impossible to provide differentiated resources based on students' personal knowledge reserves and learning progress, and it is difficult to meet the dynamic needs of teachers, students, and enterprises. Summary of the Invention

[0005] In view of this, an embodiment of the present invention provides an automobile industry-education evaluation optimization system based on a large model, which obtains a dynamic knowledge data set of automobile industry-education integration through an industry knowledge acquisition module, constructs an automobile industry-education knowledge graph through a knowledge graph construction module, obtains student learning ability related data through an ability data collection module, calculates the student's comprehensive ability score, recommends personalized learning paths and learning materials to students, and recommends a talent list to enterprises, effectively solving the problems of insufficient teaching adaptability, single evaluation feedback dimension and low resource collaboration efficiency raised in the background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solutions: a large-scale model-based automobile production-education-evaluation optimization system, comprising an operation database, an industry knowledge acquisition module, a knowledge graph construction module, a teaching resource generation module, a capability data collection module, an intelligent capability assessment module, and a resource matching and collaboration module: The operation database includes all the data information of the automobile production, education and evaluation optimization system, and collects the data information output by each module in real time; The industry knowledge acquisition module is used to deploy targeted crawler tools to collect and clean multi-dimensional automotive industry data and obtain a dynamic knowledge dataset for automotive industry-education integration. The knowledge graph construction module is used to generate knowledge nodes based on the dynamic knowledge dataset of automotive industry-education integration, build the automotive industry-education knowledge graph, and perform intelligent reasoning; The teaching resource generation module is used to automatically generate teaching resources based on the knowledge nodes in the automotive industry-education knowledge graph, including teaching courseware, virtual training scripts, and dynamic question banks; Ability data collection module, used to obtain students' learning ability related data in real time, provides multimodal interaction functions, and supports manual input of multimodal data; The intelligent ability assessment module is used to build an intelligent ability assessment model, calculate the comprehensive score of students' abilities, and output the diagnosis results of students' abilities; The resource matching and collaboration module is used to match resources based on student ability diagnosis results and the automotive industry-education knowledge graph, achieving precise matching of learning resources and paths with corporate talent needs.

[0007] Technical effects and advantages of the present invention: The present invention deploys a directional crawler tool through the industry knowledge acquisition module, which can collect multi-source data such as automobile enterprise technology white papers, industry standard updates, and maintenance case libraries in real time, obtain a dynamic knowledge dataset for automobile industry-education integration, and construct a dynamic knowledge graph through the knowledge graph construction module to achieve automation and high-frequency updates of teaching content. The present invention uses the multimodal interaction function in the ability data acquisition module to collect students' fine-grained data such as operation timing, fault analysis ideas, and spatial cognition in real time. Through the intelligent ability evaluation module, an intelligent ability evaluation model is constructed to obtain comprehensive student ability scores and student ability diagnosis results, thereby constructing a full-dimensional evaluation system to accurately locate shortcomings. The present invention uses a resource matching collaboration module to automatically recommend personalized learning paths based on the node association between student ability diagnosis results and dynamic knowledge graphs, and uses a large model to disassemble corporate job requirements and recommend talent lists, realizing intelligent scheduling of the entire process, greatly improving the efficiency of industry-education docking, and realizing the intelligent upgrade of automobile industry-education evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 It is a schematic diagram of the overall structure of the present invention.

[0009] Figure 2 Schematic diagram of the steps for generating teaching resources of the present invention.

[0010] Figure 3 Schematic diagram of the steps for determining the comprehensive ability level of students according to the present invention. DETAILED DESCRIPTION

[0011] 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.

[0012] As attached Figure 1 The automobile production-education-evaluation optimization system shown in the figure is based on a large model and includes an operation database, an industry knowledge acquisition module, a knowledge graph construction module, a teaching resource generation module, a capability data collection module, an intelligent capability evaluation module, and a resource matching and collaboration module.

[0013] The operation database includes all data information of the automobile production, education and evaluation optimization system, and collects data information output by each module in real time.

[0014] The specific implementation of the present invention includes the following contents: Industry knowledge acquisition module: Deploys targeted crawler tools to collect and clean multi-dimensional automotive industry data and obtain dynamic knowledge datasets for automotive industry-education integration; Furthermore, the steps for obtaining the automotive industry-education integration dynamic knowledge dataset are as follows: S1.1: Use the deployed targeted crawler tools to collect multi-dimensional automotive industry data, including enterprise technology data, industry standard data, academic research data, production practice data, talent demand data, and teaching practice data; In this embodiment, it should be specifically noted that the deployed directional crawler tool should have the ability to filter large model languages, collect multi-dimensional automotive industry data according to the needs of automotive production, education and evaluation scenarios, and support multi-format data capture of text, charts and videos.

[0015] It should be specifically noted that the enterprise technical data in the multi-dimensional automotive industry data includes technical white papers from OEMs such as Tesla and BYD, technical documents for new car releases, and public cases of after-sales maintenance systems, such as the 4680 battery structure parameters, the smart cockpit domain controller function list, and the diagnostic process of ESP failures in a certain model; industry standard data includes the automotive standards database of the Ministry of Industry and Information Technology, the road vehicle safety standards of the International Organization for Standardization, and skill certification specifications issued by industry associations; academic research data includes core journal papers and technical achievements of automotive engineering laboratories of universities; production practice data includes production process documents, workshop equipment operating manuals, and supply chain management cases of automobile manufacturers; talent demand data includes automotive job requirements on mainstream recruitment platforms and job descriptions released by corporate HR; teaching resource data includes automotive professional course outlines of vocational colleges and industry training courseware.

[0016] S1.2: Filter multi-dimensional automotive industry data through the semantic recognition capabilities of the large model, initiate the intelligent cleaning process, integrate and obtain the dynamic knowledge dataset of automotive industry-education integration, perform multi-dimensional classification and storage, and mark the update time with a timestamp.

[0017] In this embodiment, it should be specifically explained that the intelligent cleaning process refers to the use of the text similarity calculation function of the large model to identify and eliminate duplicate data, such as eliminating the same technical parameter table from different sources, automatically correcting erroneous information through pre-trained industry terminology library, and converting and processing unstructured data. It uses OCR technology to identify the text content in handwritten maintenance notes pictures, extracts the text content in technical videos through speech-to-text technology, and uniformly converts all data into a structured format; marking the update time with a timestamp helps the system to call the latest industry knowledge in real time, providing strong data support for subsequent knowledge graph construction and teaching resource generation.

[0018] Knowledge graph construction module: Generates knowledge nodes based on the dynamic knowledge dataset of automotive industry-education integration, builds the automotive industry-education knowledge graph, and performs intelligent reasoning; Furthermore, the generation of knowledge nodes requires pre-training a named entity recognition model dedicated to the automotive field through a large model. The dynamic knowledge dataset of automotive industry-education integration is imported into the named entity recognition model dedicated to the automotive field to extract core entities and generate knowledge nodes. Knowledge nodes specifically include technology entities, fault entities, process entities, standard entities, job entities, and teaching entities; In this embodiment, it should be specifically noted that the knowledge nodes all contain attribute labels and timestamps. For example, the attributes of the "800V high-voltage platform" are "compatible vehicle model: Zeekr 001" and the core component "silicon carbide inverter". The original source in the data set is associated with the timestamp to ensure the traceability of the knowledge node.

[0019] The construction of the automotive industry-education knowledge graph requires the ability to extract relationships based on large models, automatically identify logical relationships between entities, build a multi-level knowledge network, and adopt a dynamic update mechanism; In this embodiment, it should be specifically explained that the logical associations between entities include subordinate relationships, causal relationships, adaptation relationships, standard associations, and teaching associations. Subordinate relationships include, for example, the city NOA function is subordinate to the autonomous driving L2+ function, causal relationships include, for example, battery cell aging leads to reduced mileage, adaptation relationships include, for example, the high-voltage electrician certificate is adapted to the new energy maintenance position, standard associations include, for example, the battery pack waterproof level must comply with the IP67 standard, and teaching associations include, for example, high-voltage safety operation is associated with virtual training project 5.

[0020] It should be specifically explained that the dynamic update mechanism means that when new data is passed into the industry knowledge acquisition module, the large model automatically determines the relationship between the new data and the existing nodes, and updates the automobile industry-education knowledge graph structure in real time to ensure the real-time nature of the knowledge network.

[0021] Intelligent reasoning includes fault diagnosis reasoning, teaching resource association reasoning, and job capability matching reasoning.

[0022] In this embodiment, it should be specifically explained that fault diagnosis reasoning refers to the output of a fault, the system automatically associates knowledge nodes, and deduces possible causes and troubleshooting paths; teaching resource association reasoning refers to when students learn teaching data, the system automatically recommends course-related courseware and practical training projects through the logical association between knowledge nodes; job capability matching reasoning refers to when the company's needs are input, the system disassembles the knowledge nodes and compares them with the student's capability profile to generate a matching score; in the process of intelligent reasoning, the large model in the system will combine the case data in the dynamic knowledge data set of automotive industry-education integration to optimize the reasoning weight and improve the accuracy and practicality of the conclusion.

[0023] Teaching resource generation module: Automatically generates teaching resources based on knowledge nodes in the automotive industry-education knowledge graph, including teaching courseware, virtual training scripts, and dynamic question banks; Further, such as Figure 2 As shown in the figure, the steps for generating teaching resources are as follows: S2.1: Generate teaching courseware through logical integration and visual arrangement of large models based on the technical entities, standard entities, and teaching entities of knowledge nodes in the automotive industry-education knowledge graph; In this embodiment, it should be specifically explained that the generation of teaching courseware needs to automatically construct the courseware structure based on the subordinate relationship of knowledge nodes to ensure a clear knowledge context, automatically embed charts, parameter comparison tables and dynamic links related to the automotive industry-education knowledge graph, and the big model performs semantic adaptation on the inserted content and dynamically adjusts the content depth according to the difficulty label of the teaching entity node.

[0024] S2.2: Generate interactive virtual training scripts based on the process entities and fault entities of the knowledge nodes in the automotive industry-education knowledge graph and production practice data; In this embodiment, it should be specifically explained that the generation of virtual training scripts needs to be based on the operating specifications in the process entity nodes, and the entity operations must be broken down into multiple steps. Each step should include 3D model annotations, force feedback parameters, and risk enhancement; combined with the causal relationship of the fault entity, fault variables are randomly implanted in the basic operation script, and the interaction granularity of the script is automatically adjusted according to the hardware parameters of the training equipment to ensure the consistency between the virtual operation and the physical equipment.

[0025] S2.3: Based on the logical association of knowledge nodes in the automotive industry-education knowledge graph, a dynamic question bank covering multi-dimensional assessment needs is generated. The dynamic question bank is automatically updated as the automotive industry-education knowledge graph is iterated.

[0026] In this embodiment, it should be specifically explained that the dynamic question bank includes multiple-choice questions, practical questions and fault analysis questions. The multiple-choice questions generate options around technical entity nodes, the practical questions are combined with the operating specifications of the process entity, and students are required to answer according to the steps. The fault analysis questions are based on the cause-effect relationship of the fault entity and require the association of corresponding detection tools.

[0027] It should be specifically noted that when the dynamic question bank is generated, each question needs to be automatically associated with the corresponding knowledge node in the automotive industry-education knowledge graph, and the difficulty coefficient is marked according to the node level. The system can automatically compile papers according to the rules to meet the assessment needs of different stages; when the automotive industry-education knowledge graph is updated, the question bank should automatically generate corresponding new questions and eliminate outdated questions to ensure that the assessment content in the dynamic question bank is synchronized with industry technology.

[0028] Ability data collection module: acquires students' learning ability-related data in real time, provides multimodal interaction functions, and supports manual input of multimodal data; Furthermore, the data related to students' learning ability include theoretical knowledge mastery data, practical operation ability data and thinking and analysis ability data. The theoretical knowledge mastery data specifically include the accuracy rate of answering questions R1, the coverage of knowledge points D1 and the proportion of learning time R2. The practical operation ability data specifically include the operation standardization D2, the troubleshooting efficiency R3 and the equipment operation proficiency D3. The thinking ability data specifically include the logical reasoning matching degree D4 and the question quality score R4.

[0029] In this embodiment, it should be specifically explained that the acquisition of student learning ability-related data requires the use of a database trigger mechanism. Each time a question is completed, the answer record table is automatically updated. The accuracy rate is calculated through SQL association query, and the knowledge point coverage is calculated by combining the label matching of the knowledge nodes in the automotive industry-education knowledge graph; JavaScript is used to count the loading and leaving events of the interface, and the length of stay is calculated by the difference in timestamps. The knowledge nodes in the automotive industry-education knowledge graph are classified and stored in the learning behavior log table according to the difficulty labels to obtain the proportion of learning time; The time series comparison method is used to match the student's operation steps with the preset standard steps, calculate the edit distance, and convert it into the operation standardization degree. The status data is obtained through the API interface of the training equipment. The operation time is recorded with a timer, and the troubleshooting efficiency is calculated using the ratio method. The type of tool selected by the student and the number of repeated operations are recorded based on the collision detection algorithm to calculate the equipment operation proficiency. The entity extraction and relationship extraction capabilities of the large model are used to parse student input content, generate triples, and calculate the cosine similarity with the standard triples in the automotive industry-education knowledge graph to obtain the logical reasoning matching degree; the BERT model is used to classify the text and extract keywords, and the bag-of-words model is used to calculate the overlap rate with standard knowledge points to obtain the question quality score.

[0030] Furthermore, multimodal interaction functions include voice interaction, handwritten formula and sketch input, 3D model annotation, gesture command interaction, and document upload interaction.

[0031] In this embodiment, it should be specifically explained that voice interaction requires the ability data acquisition module to integrate the voice recognition and semantic understanding engine driven by the large model, and the system automatically converts the voice into text and associates the corresponding knowledge points; handwritten formulas and sketch input require a handwriting board tool to be provided in the interactive interface, and the handwritten content is converted into a vector diagram through image recognition technology, and the accuracy is judged in combination with the large model; 3D model annotation means that in the virtual training simulation, students can annotate the 3D car model, and the system automatically records the standard position and content, and records the interaction efficiency and accuracy; gesture command interaction supports gesture operation recognition in virtual and real simulation training; document upload interaction means supporting students to upload self-organized learning materials, and the system extracts key information through OCR and text analysis, and associates it with the automotive industry-education knowledge graph to judge the student's knowledge integration ability.

[0032] Intelligent ability assessment module: builds an intelligent ability assessment model, calculates the comprehensive score of students' abilities, and outputs the diagnosis results of students' abilities; Furthermore, the intelligent ability assessment model is used to calculate the comprehensive score of students' abilities based on the imported student learning ability related data. The calculation steps of the comprehensive score of students' abilities are as follows: S3.1: Set a time window, obtain the student learning ability related data X within the time window, import the intelligent ability assessment model, and use the min-max normalization formula to obtain the standardized score X of the student learning ability related data. b ; In this embodiment, it should be specifically noted that the time window can be 10 days. The student learning ability-related data within the last 10 days is obtained. The collection frequency of the student learning ability-related data is once a day. The average value of the student learning ability-related data within 10 days is calculated and used as the input value of the intelligent ability model. The min-max normalization formula is expressed as: , where X min Indicates the minimum value of the student learning ability correlation data, X max It represents the maximum value of the data related to students' learning ability. By standardizing the scoring of the original data, the dimensional differences between different data types are eliminated and the original data related to students' learning ability are converted into a standard score of 0-100.

[0033] S3.2: Standardized score R of correct answer rate, knowledge point coverage and learning time ratio 1b 、D 1b and R 2b Substituting into the formula: , The theoretical knowledge mastery score L1 is calculated, where a1, a2, and a3 represent the weight coefficients of the correct answer rate, knowledge point coverage, and learning time ratio, respectively, and the sum is 1; the standardized scores of operation standardization, troubleshooting efficiency, and equipment operation proficiency D are calculated. 2b 、R 3b and D 3b Substituting into the formula: , The practical operation ability score L2 is calculated, where b1, b2 and b3 represent the weight coefficients of operation standardization, troubleshooting efficiency and equipment operation proficiency respectively, and the sum is 1; the standardized score D of logical reasoning matching and question quality score is calculated. 4b and R 4b Substituting into the formula: , The thinking ability score L3 is calculated, where c1 and c2 represent the weight coefficients of logical reasoning matching and question quality score, respectively, and the sum is 1; In this embodiment, it should be specifically explained that in the calculation of the theoretical knowledge mastery score, the logarithmic function is used to amplify the impact of high accuracy on the score, and the exponential function is used to strengthen the impact of coverage completeness, and a1, a2 and a3 can be 50%, 30% and 20%; in the calculation of the practical operation ability score, the exponential function is used to highlight the importance of standardized operation, and the logarithmic function is used to correct the adverse impact of troubleshooting efficiency, and b1, b2 and b3 can be 40%, 30% and 30%; in the calculation of the thinking ability score, the logarithmic function is used to amplify the advantage of high matching degree, and the linear weighting combined with the exponential function is used to strengthen the value of high-order questioning, and c1 and c2 can be 60% and 40%; among them, 1.35, 1.2 and 1.2 are used as correction coefficients to avoid the situation where the calculation results of the theoretical knowledge mastery score, practical operation ability score and thinking ability score are greater than 100.

[0034] S3.3: The theoretical knowledge mastery score, practical operation ability score and thinking ability score are weighted to obtain the student's comprehensive ability score L.

[0035] In this embodiment, it should be specifically explained that to obtain the comprehensive score of student ability, the theoretical knowledge mastery score, practical operation ability score and thinking ability score need to be introduced into the formula L=ω1L1+ω2L2+ω3L3 for calculation, where ω1, ω2 and ω3 represent the weight coefficients of the theoretical knowledge mastery score, practical operation ability score and thinking ability score respectively, and the sum is 1, which can be 40%, 35% and 25%. By incorporating logarithmic exponents to amplify high-quality indicators and exponential functions to strengthen key abilities, the calculated comprehensive score of student ability is more in line with student ability.

[0036] Furthermore, the results of student ability diagnosis specifically include the student's comprehensive ability level, ability portrait, key ability shortcoming positioning and improvement suggestions, among which the student's comprehensive ability level is determined based on the student's comprehensive ability score.

[0037] In this embodiment, it should be specifically explained that the comprehensive ability level of students includes the first level of student ability, the second level of student ability, the third level of student ability, the fourth level of student ability and the fifth level of student ability. Figure 3 As shown in Figure 2, the steps for determining students’ comprehensive ability levels are as follows: S4.1: When L is between 90 and 100 points, it is at the first level of student ability, indicating an excellent level, comprehensive knowledge coverage, and the ability to independently solve complex technical problems; S4.2: When L is between 80 and 89 points, it is at the second level of student ability, indicating that the student is at a good level, has a slight weakness in a single dimension, and can skillfully complete routine technical tasks; S4.3: When L is between 60 and 79 points, it is at the third level of student ability, indicating that the student is at a qualified level, but has some obvious shortcomings. The student can complete basic tasks but the efficiency is low. S4.4: When L is between 40 and 59 points, it is at the fourth level of student ability, indicating that the student is at a level that needs improvement. The comprehensive ability score does not meet the standard and the student is unable to complete basic tasks independently. S4.5: When L is between 0 and 39 points, it is at the fifth level of student ability, indicating a weak level, extremely low comprehensive ability score, and lack of basic safety operation awareness and technical knowledge.

[0038] It should be specifically explained that the capability portrait includes theoretical knowledge dimension, practical operation dimension and thinking analysis dimension, which can be visualized through radar charts or bar charts; the positioning of key capability shortcomings needs to be combined with the automotive industry-education knowledge map to accurately locate weak links, including knowledge level, skill level and thinking level; improvement suggestions need to automatically generate executable improvement paths based on the positioning of key capability shortcomings to provide students with personalized growth guidance.

[0039] Resource matching and collaboration module: Matches resources based on student ability diagnosis results and the automotive industry-education knowledge graph to achieve precise matching of learning resources and paths with corporate talent needs.

[0040] Furthermore, the resource matching collaboration module links the student ability diagnosis results with the automotive industry-education knowledge graph, and uses intelligent matching algorithms to accurately match learning resources and paths with corporate talent needs, recommending learning paths and learning materials to students and pushing talent lists to companies.

[0041] In this embodiment, it should be specifically noted that the recommendation of learning paths needs to be based on the association between students' ability shortcomings and the knowledge nodes of the automotive industry-education knowledge graph, generating a step-by-step learning path divided into the basic stage, the reinforcement stage, and the application stage. A dynamic adjustment mechanism is also established to track students' path completion status in real time and adjust the recommended content in real time according to students' learning progress. Recommended learning materials should be based on the phased goals of the learning path and the knowledge node attributes of the automotive industry-education knowledge graph, matching multiple types of highly adaptable learning materials, including theoretical knowledge materials, practical operation materials, and thinking training materials. The push of talent lists requires receiving corporate job requirements, breaking them down into core capability indicators through a large model, and mapping them to the knowledge nodes of the automotive industry-education knowledge graph. The matching degree between the comprehensive student ability score and the job requirements is calculated, and a talent list is generated from high to low according to the matching degree. The core strengths of the students are marked, and the company's recruitment results and job adaptation evaluation are fed back to the system. The matching algorithm weight is optimized through reinforcement learning to continuously improve the accuracy of talent recommendations.

[0042] Secondly: The drawings of the embodiments disclosed in the present invention only involve structures related to the embodiments disclosed in the present invention. Other structures may refer to conventional designs. The same embodiment and different embodiments of the present invention may be combined with each other without conflict. Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A large-scale model-based automobile production-education-evaluation optimization system, characterized by: It includes an operation database, an industry knowledge acquisition module, a knowledge graph construction module, a teaching resource generation module, a capability data collection module, an intelligent capability assessment module, and a resource matching and collaboration module: The operation database includes all data information of the automobile production, education and evaluation optimization system, and collects data information output by each module in real time; The industry knowledge acquisition module is used to deploy a targeted crawler tool to collect and clean multi-dimensional automotive industry data and obtain a dynamic knowledge dataset for automotive industry-education integration; The knowledge graph construction module is used to generate knowledge nodes based on the automotive industry-education integration dynamic knowledge dataset, construct the automotive industry-education knowledge graph, and perform intelligent reasoning; The teaching resource generation module is used to automatically generate teaching resources based on the knowledge nodes in the automobile production and education knowledge graph, including teaching courseware, virtual training scripts and dynamic question banks; The capability data acquisition module is used to obtain students' learning capability-related data in real time, provide multimodal interaction functions, and support manual input of multimodal data; The intelligent ability assessment module is used to build an intelligent ability assessment model, calculate the comprehensive score of students' abilities, and output the students' ability diagnosis results; The resource matching collaboration module is used to match resources based on student ability diagnosis results and the automotive industry-education knowledge graph, thereby achieving accurate matching of learning resources and paths with corporate talent needs.

2. The large-scale model-based automobile production, education, and evaluation optimization system according to claim 1, characterized in that: The steps for obtaining the automotive industry-education integration dynamic knowledge dataset are as follows: S1.1: Use the deployed targeted crawler tools to collect multi-dimensional automotive industry data, including enterprise technology data, industry standard data, academic research data, production practice data, talent demand data, and teaching practice data; S1.2: Filter multi-dimensional automotive industry data through the semantic recognition capabilities of the large model, initiate the intelligent cleaning process, integrate and obtain the dynamic knowledge dataset of automotive industry-education integration, perform multi-dimensional classification and storage, and mark the update time with a timestamp.

3. The large-scale model-based automobile production, education, and evaluation optimization system according to claim 1, characterized in that: The generation of the knowledge nodes requires pre-training a named entity recognition model dedicated to the automotive field through a large model, importing the automotive industry-education integration dynamic knowledge dataset into the named entity recognition model dedicated to the automotive field, extracting core entities, and generating knowledge nodes; Knowledge nodes specifically include technology entities, fault entities, process entities, standard entities, job entities, and teaching entities; The construction of the automotive industry-education knowledge graph requires the ability to extract relationships based on large models, automatically identify logical relationships between entities, build a multi-level knowledge network, and adopt a dynamic update mechanism; Intelligent reasoning includes fault diagnosis reasoning, teaching resource association reasoning, and job capability matching reasoning.

4. The large-scale model-based automobile production, education, and evaluation optimization system according to claim 1, characterized in that: The steps for generating the teaching resources are as follows: S2.1: Generate teaching courseware through logical integration and visual arrangement of large models based on the technical entities, standard entities, and teaching entities of knowledge nodes in the automotive industry-education knowledge graph; S2.2: Generate interactive virtual training scripts based on the process entities and fault entities of the knowledge nodes in the automotive industry-education knowledge graph and production practice data; S2.3: Based on the logical association of knowledge nodes in the automotive industry-education knowledge graph, a dynamic question bank covering multi-dimensional assessment needs is generated. The dynamic question bank is automatically updated as the automotive industry-education knowledge graph is iterated.

5. The large-scale model-based automobile production, education, and evaluation optimization system according to claim 1, characterized in that: The student learning ability related data includes theoretical knowledge mastery data, practical operation ability data and thinking and analysis ability data. The theoretical knowledge mastery data specifically includes the correct answer rate R1, knowledge point coverage D1 and learning time ratio R2. The practical operation ability data specifically includes operation standardization D2, troubleshooting efficiency R3 and equipment operation proficiency D3. The thinking ability data specifically includes logical reasoning matching D4 and question quality score R 4。 6. The large-scale model-based automobile production, education, and evaluation optimization system according to claim 1, characterized in that: The multimodal interaction functions include voice interaction, handwritten formula and sketch input, 3D model annotation, gesture command interaction and document upload interaction.

7. The large-scale model-based automobile production-education-evaluation optimization system according to claim 1, characterized in that: The intelligent ability assessment model is used to calculate the comprehensive score of students' abilities based on the imported student learning ability related data. The calculation steps of the comprehensive score of students' abilities are as follows: S3.1: Set a time window, obtain the student learning ability related data X within the time window, import the intelligent ability assessment model, and use the min-max normalization formula to obtain the standardized score X of the student learning ability related data. b ; S3.2: Standardized score R of correct answer rate, knowledge point coverage and learning time ratio 1b 、D 1b and R 2b Substituting into the formula: ， The theoretical knowledge mastery score L1 is calculated, where a1, a2, and a3 represent the weight coefficients of the correct answer rate, knowledge point coverage, and learning time ratio, respectively, and the sum is 1; the standardized scores of operation standardization, troubleshooting efficiency, and equipment operation proficiency D are calculated. 2b 、R 3b and D 3b Substituting into the formula: ， The practical operation ability score L2 is calculated, where b1, b2 and b3 represent the weight coefficients of operation standardization, troubleshooting efficiency and equipment operation proficiency respectively, and the sum is 1; the standardized score D of logical reasoning matching and question quality score is calculated. 4b and R 4b Substituting into the formula: ， The thinking ability score L3 is calculated, where c1 and c2 represent the weight coefficients of logical reasoning matching and question quality score, respectively, and the sum is 1; S3.3: The theoretical knowledge mastery score, practical operation ability score and thinking ability score are weighted to obtain the student's comprehensive ability score L.

8. The large-scale model-based automobile production, education, and evaluation optimization system according to claim 1, characterized in that: The student ability diagnosis results specifically include the student's comprehensive ability level, ability profile, key ability shortcoming positioning and improvement suggestions, among which the student's comprehensive ability level is determined based on the student's comprehensive ability score.

9. The large-scale model-based automobile production, education, and evaluation optimization system according to claim 1, characterized in that: The resource matching and collaboration module associates the student ability diagnosis results with the automotive industry-education knowledge graph, and uses an intelligent matching algorithm to achieve accurate matching of learning resources and paths with corporate talent needs, recommending learning paths and learning materials to students and pushing talent lists to companies.

Images