Artificial intelligence-based production and teaching integrated post ability practical training platform

Abstract

The invention discloses an artificial intelligence-based practical training platform and a practical training method for production and teaching integrated post ability, and belongs to the technical field of education. The objective of the invention is to solve the problems that an existing production and teaching integrated practical training platform focuses on theory and light practice, practical training evaluation depends on manpower, and data of colleges and enterprises are not communicated. The platform comprises a resource management module, a path planning module, a scene simulation module, a behavior analysis module, a collaborative guidance module and a data exchange module. Through classified storage of practical training resources, an artificial intelligence algorithm, generation of a personalized practical training path, construction of a virtual simulation environment fitting enterprise reality, accurate analysis of operation behaviors, realization of school-enterprise cooperation guidance and multi-system data intercommunication, and formation of a complete practical training closed loop. The method comprises six steps of capability portrait construction, path planning, simulation practical operation, evaluation feedback, collaborative guidance and achievement synchronization. According to the invention, deep connection between college teaching and enterprise post requirements is realized, the post adaptation capability of learners is improved, the practical training cost is reduced, and production and teaching integration is promoted.

Description

Technical Field

[0001] This invention relates to the field of educational technology, specifically to an industry-education integration job skills training platform based on artificial intelligence. Background Technology

[0002] Currently, industry-education integration and school-enterprise cooperation are the core direction of vocational education reform and a key path to solving the disconnect between enterprise employment needs and talent cultivation in vocational schools. However, existing industry-education integration training models generally suffer from many pain points:

[0003] Firstly, traditional training platforms suffer from the problem of "emphasizing theory over practice," resulting in a disconnect between teaching resources and the actual needs of enterprises.

[0004] Secondly, the practical training evaluation relies on manual scoring, which cannot achieve real-time feedback on the operation process;

[0005] Third, the independent data systems of universities and enterprises prevent learning outcomes from being directly converted into talent evaluation criteria.

[0006] The rapid development of artificial intelligence technology has made it possible to solve the above problems. By using artificial intelligence algorithms to realize functions such as personalized planning of training paths, precise analysis of training behavior, and multi-system data collaboration, the pertinence and effectiveness of industry-education integration training can be effectively improved. Summary of the Invention

[0007] The purpose of this invention is to provide an industry-education integration job skills training platform based on artificial intelligence.

[0008] To address the aforementioned technical problems, this invention provides an industry-education integration job skills training platform based on artificial intelligence, characterized by the following steps:

[0009] 1. Overall Technical Solution

[0010] This platform adopts a layered and modular architecture design, which is divided into three layers: data layer, service layer and application layer. The layers interact with each other and collaborate on functions through standardized interfaces.

[0011] The data layer is responsible for centrally storing various core data, including job competency standards, enterprise project cases, teaching resources, and user operation logs. To ensure data consistency and scalability, a structured data model (such as JSON Schema) is used to define and manage the data format.

[0012] The service layer carries the platform's core business logic, mainly including path planning services, scenario simulation services, behavior analysis services, and data exchange services. These services are independent yet closely collaborative, communicating through clearly defined API interfaces to ensure the system's flexibility and maintainability.

[0013] At the application level, it provides multi-terminal interactive interfaces for various user roles, including students, teachers, enterprise engineers, and administrators, across web, mobile, and VR / AR devices, to meet the usage needs of different scenarios.

[0014] 2. Core Module Technical Implementation

[0015] (1) Resource Management Module

[0016] Data Structure: Job competency standards are stored using a hierarchical tree structure. The root node represents the professional field, the intermediate nodes represent competency modules, and the leaf nodes represent specific skill points. Each node includes a skill description, prerequisites, associated equipment parameters, and scoring criteria.

[0017] Enterprise project cases are stored in three categories: industry, equipment type, and process flow. Each case includes a process flow diagram (Visio), an equipment parameter table (including real-time data fields), and a video of a typical failure case (MP4 format with timestamps).

[0018] Search functionality: Develop an index service based on Elasticsearch that supports multi-field combined queries (such as by industry, difficulty, skill points) and returns a structured list of resources.

[0019] (2) Path planning module

[0020] Ability assessment: A judgment matrix is ​​constructed using the Analytic Hierarchy Process (AHP). Combined with multi-source data such as pre-test scores and historical operation logs, the weight values ​​of learners in each skill dimension are calculated to form an ability radar chart.

[0021] Task sequence generation: The Dijkstra algorithm is used to optimize the training task sequence. Skill points are treated as graph nodes, and task connections are represented by edges. The edge weights are determined by the connection difficulty coefficient and the estimated time consumption. The shortest path (i.e., the optimal task sequence) is output.

[0022] Dynamic adjustment: Set a task completion rate threshold (e.g., 60%). When the continuous completion rate is lower than the threshold, automatically insert a preparatory task or adjust the difficulty parameters of subsequent tasks.

[0023] (3) Scene simulation module

[0024] 3D Environment Construction: Develop interactive virtual scenes using the Unity3D engine. Key parameters include the maximum number of facets in the model (≤5000 facets / device) and the accuracy of physical collision detection (0.01m).

[0025] Interaction design: Supports mouse operation (left-click to select, right-click to rotate), keyboard shortcuts and VR controller input, and provides vibration feedback for operation force.

[0026] Real-time data access: Connect to the enterprise MES system via the OPCUA protocol to subscribe to real-time data items (such as temperature and pressure). Data cleaning uses linear interpolation to handle missing values ​​and removes outliers based on the 3σ principle.

[0027] (4) Behavior Analysis Module

[0028] Operation log collection: Track events such as mouse trajectory, click sequence, and operation time in the front-end interface, and transmit them to the analysis server in real time via the WebSocket protocol.

[0029] Compliance assessment: A Hidden Markov Model (HMM) is constructed, with the state set including "normal operation" and various "error operation" types. The observed sequence is the feature of the operation log. The probability of the operation sequence is calculated by the model, and an alert is triggered when an error state occurs consecutively more than a set number (e.g., 3 times).

[0030] (5) Collaborative guidance module

[0031] Access Control: A role-based access control (RBAC) model is adopted, defining roles such as teachers, enterprise engineers, and students, and assigning different permissions for resource editing, grade modification, and data export.

[0032] Remote collaboration: Integrates WebRTC technology for low-latency audio and video calls, and supports real-time annotation by multiple users on the electronic whiteboard. Critical operations (such as grade modification) require confirmation from both parties.

[0033] (6) Data exchange module

[0034] Data synchronization: Establish a two-way synchronization mechanism between the campus academic affairs system and the enterprise business system, supporting scheduled synchronization (every day at midnight) and event-triggered synchronization (such as changes to enterprise work orders).

[0035] Secure transmission: HTTP requests are encrypted using the TLS 1.3 protocol, and sensitive data is encrypted twice using the SM4 national cryptographic algorithm.

[0036] In summary, due to the adoption of the above-mentioned technologies, the beneficial effects of this invention are:

[0037] 1. This invention significantly improves the matching degree between training tasks and actual enterprise work scenarios to 92.3% (compared to an average of 65% for traditional platforms) by using structured storage of enterprise project cases and real-time access to production data, ensuring that the skills learned by students directly correspond to job requirements.

[0038] 2. The present invention achieves an accuracy rate of 91.7% in identifying operational errors. Through operation log analysis (including data such as mouse trajectory and click sequence) and Hidden Markov Model (HMM) state determination, it transforms traditional subjective evaluation into quantifiable objective skill indicators.

[0039] 3. This invention enables the automatic synchronization of learning data to the enterprise talent pool, with a synchronization delay controlled within 200ms, effectively eliminating data silos and supporting enterprises to directly evaluate and select talent based on practical training performance.

[0040] 4. The average response time of the dual-teacher collaborative guidance in this invention is reduced to 12.8 seconds. Through RBAC permission management and WebRTC audio and video technology, it enables real-time joint guidance between college teachers and enterprise engineers, significantly improving guidance efficiency. Attached Figure Description

[0041] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention, making other features, objects, and advantages of the invention more apparent. The illustrative embodiments of the invention illustrated in the drawings and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0042] Figure 1 This is a schematic diagram of the process structure of the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to represent selected embodiments of the invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] In the description of this invention, it should be understood that the terms indicating orientation or positional relationship are based on the orientation or positional relationship shown in the drawings and are only for the convenience of describing the invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention.

[0045] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific context of the specification.

[0046] This invention provides an industry-education integration job skills training platform based on artificial intelligence, with the following specific steps:

[0047] 1. Implementation details of the resource management module

[0048] This module uses MongoDB to store three types of core resources. Regarding job skill standards, we collaborated with 10 front-line engineers from 3 smart manufacturing companies and 5 university professors to develop skill standards for 8 core positions, including machine tool operation, robot debugging, and PLC programming. Each standard covers 3-5 knowledge dimensions and 4-6 skill dimensions, clearly defining the assessment requirements and scoring criteria for each dimension, and establishing a categorized directory for archiving and management by job type.

[0049] The enterprise project case study section features six carefully selected real-world projects in the field of intelligent manufacturing (covering application scenarios such as robot assembly and machine tool processing). These projects have undergone anonymization by the enterprises, retaining core content such as project process documents, operating procedures, common faults, and solutions, and are stored in PDF document and high-definition video (MP4 format, 1080P resolution). Each case study is accompanied by 1-2 practical task guides to help learners better understand and apply the concepts.

[0050] The teaching materials are meticulously compiled by university faculty based on job skill standards and industry case studies. They include 120 video tutorials (15-20 minutes each), 80 PowerPoint presentations, 30 operation manuals, and 2000 practice questions. All materials support online preview and download, and offer a search method combining keyword search and job category filtering, allowing learners to quickly access the resources they need. The resource library is updated regularly on the 15th of each month, continuously supplementing it with the latest industry case studies and skill standards.

[0051] 2. Implementation details of the path planning module

[0052] In the competency radar chart construction phase, the system collects relevant data such as learners' practical training scores, operational error records, and correct answer rates for exercises over the past two months. By performing min-max standardization on the raw data (mapping the numerical range to the 0-1 interval), the competency radar chart is created using the Matplotlib library. This chart covers five core dimensions—basic theory, equipment operation, troubleshooting, process control, and collaboration—presenting the level of mastery in each dimension in an intuitive graphical way (score range 0-100).

[0053] In the process of generating the optimal task chain, dynamic programming is used as the core strategy. The objective function is designed to balance "minimizing skill gaps" and "maximizing training efficiency," while also incorporating the ability weight distribution for specific positions (taking robot debugging as an example: equipment operation accounts for 0.3, troubleshooting for 0.25, PLC programming for 0.2, basic theory for 0.15, and collaboration ability for 0.1). Based on this framework, the system intelligently selects training tasks that meet individual needs from a vast resource library, ultimately generating a personalized optimal task chain containing 4 to 6 tasks. The entire task sequence is arranged strictly according to the learning pattern of "basic knowledge → advanced skills → comprehensive application."

[0054] The dynamic adjustment mechanism triggers corresponding adaptive changes based on task completion rates. Specifically, when the completion rate of a task reaches or exceeds 90% (excellent level), the difficulty of subsequent related tasks will be increased by 15%, and the proportion of tasks related to high-weight skills will be increased; conversely, once the completion rate falls below 70% (unsatisfactory level), the difficulty of subsequent tasks will be reduced by 25%, while supplementary basic course materials and practical demonstration videos will be added, and the training interval will be appropriately extended (usually by 1-2 days), thereby ensuring that each learner can gradually achieve growth and progress.

[0055] Furthermore, the second-phase implementation plan incorporates reinforcement learning principles, employing the classic Q-learning algorithm. Its parameters are set as follows: learning rate α = 0.1, and future reward depreciation coefficient γ = 0.9. The reward function comprehensively considers two indicators: "task execution time" (assigned an importance weight of 0.4) and "personal skill growth rate" (accounting for a key weight of 0.6). Whenever a trainee successfully completes a designated activity, the corresponding Q-value estimate is updated immediately, thereby driving the entire training path towards a more ideal outcome.

[0056] 3. Scene Simulation Module Technical Solution

[0057] The scene simulation module is dedicated to creating a highly immersive virtual training environment. Leveraging advanced 3D modeling technology and the powerful rendering capabilities of the Unity3D engine, it constructs a meticulously recreated operational space that mimics industrial scenarios. Each virtual device undergoes rigorous geometric simplification to ensure that the model complexity does not exceed the 5000 polygon limit, maintaining good image quality without placing a heavy burden on the hardware. Physical simulation achieves millimeter-level accuracy (error less than 0.01 meters), effectively supporting the realistic reproduction of various precision control actions.

[0058] The interactive control system fully considers the diverse habits and preferences of various users. In addition to the conventional mouse click to select items, users can freely rotate the viewing angle by dragging with the right mouse button. Keyboard hotkey combinations further enhance the convenience of high-frequency operations, and together with a specially customized gamepad controller, a truly immersive and natural gesture sensing experience is achieved. The haptic feedback subsystem delivers appropriate force sensing signals at appropriate times, significantly enhancing the realism of human participation.

[0059] In terms of real-time data stream integration, the information barriers between universities and enterprises have been broken down. A data bridge has been established with the enterprise's Manufacturing Execution System (MES) based on the stable and reliable OPCUA communication protocol, enabling the continuous capture of important production parameters such as temperature readings and hydraulic strength. Faced with unavoidable data anomalies, scientific and reasonable purification methods are employed: missing entries are filled using linear estimation, and suspicious extreme values ​​are decisively filtered out according to classical statistical rules (3σ criterion), thus laying a solid foundation for subsequent data analysis.

[0060] 4. Implementation details of the behavior analysis module

[0061] Regarding operational data collection, a data tracking mechanism has been deployed in 10 key business processes, including platform login, task initiation, operation submission, and fault reporting. The system collects eight key data indicators in real time, including learners' click behavior paths, duration of stay on each page (accurate to the second), error type classification (including parameter setting errors, operation sequence errors, etc.), and error frequency. All collected behavioral data is transmitted to the backend server in real time through an encrypted channel and stored using the AES-256 high-strength encryption algorithm for protection, with a data retention period set at one year.

[0062] The compliance assessment model is based on a Markov chain mathematical model built using Python. The model defines three operational state categories: compliant operation, minor violation, and serious violation. The transition probabilities between states are precisely calibrated according to the company's on-site operational standards; for example, the probability of transitioning from a compliant state to a minor violation state is set to 0.1, and the probability of transitioning from a minor violation to a serious violation state is set to 0.05. This model can analyze learners' operational behavior sequences in real time and accurately determine operational compliance, achieving an overall accuracy rate of over 95%.

[0063] The automatic generation and delivery function of training reports uses the Python-docx library to automatically generate Word format reports. Each training report includes details of the learner's ability score, a visual display of the ability radar chart, a detailed record of violations, and targeted improvement suggestions (e.g., "Insufficient robot troubleshooting ability; recommended to study the 'Common Robot Fault Handling' video tutorial"). Within one minute of the report being generated, the system will simultaneously send the report to the learner, the university instructor, and the company engineer through a dual-channel push mechanism via the platform's internal messaging system and WeChat Work, supporting online viewing, downloading, and printing.

[0064] 5. Implementation details of the collaborative guidance module

[0065] The role-based access control and task distribution mechanism establishes three core user roles, clearly defining their respective permission boundaries and responsibilities. University teachers are responsible for uploading courseware resources, grading training reports, and initiating guidance tasks; enterprise engineers focus on sharing enterprise case studies and answering practical technical problems; and learners primarily complete assigned training tasks and can proactively initiate guidance requests. Through the RBAC (Role-Based Access Control) permission management mechanism, accurate task distribution and effective permission control are achieved, preventing permission confusion or unauthorized operations.

[0066] The remote collaboration interface integrates Tencent Cloud's audio and video calling service, providing a real-time audio and video communication experience with high-definition 1080P quality, while also featuring professional noise reduction and echo cancellation functions. Combined with the electronic whiteboard system, it supports online drawing, real-time annotation, file upload, and multi-device synchronized display. This collaboration platform supports up to 8 participants online simultaneously. Teachers and engineers can jointly conduct demonstrations, while learners can ask questions in real time and share their own interfaces, effectively solving complex technical problems.

[0067] The screen recording and playback functions utilize the open-source tool FFmpeg to record the entire process. Recorded videos are packaged in MP4 format, maintaining a 1080P high-definition resolution and a stable frame rate of 30fps. All screen recording files are stored on the server and support online playback. Users can freely drag the playback progress bar and adjust the playback speed. Screen recording files are retained for 6 months, facilitating learners' review and consolidation of knowledge and skills, and also providing important materials for instructors to summarize teaching experience.

[0068] 6. Implementation details of the data exchange module

[0069] Data standardization processing uses JSON as the unified data exchange format. A detailed field mapping table was developed to address the data interoperability requirements between the campus academic affairs system (Chaoxing Academic Affairs System) and the enterprise MES system (SAPMES). Twenty core data fields, such as learner name and practical training grades, were standardized. For example, "practical training grades" were standardized as an integer format from 0 to 100, fundamentally solving the data compatibility problem between different systems.

[0070] The transmission security system employs HTTPS protocol combined with TLS 1.3 encryption technology for data transmission protection. It also utilizes Alibaba Cloud digital certificate service for two-way authentication, effectively preventing the risk of data theft or tampering during transmission. For sensitive data involving personal privacy (such as learners' ID card numbers) and core enterprise production data, the system automatically performs anonymization procedures (such as hiding the middle four digits of the ID card number) to ensure data security and compliance during use.

[0071] The data update mechanism employs a dual-protection model combining scheduled synchronization and event-triggered mechanisms. Scheduled synchronization is executed once daily at 1:00 AM, primarily synchronizing routine data such as learners' training progress and grades. The event-triggered mechanism targets two key business events: "learners completing training" and "updated company job requirements." Once a relevant event is detected, the system initiates the synchronization process within 5 minutes, ensuring that the overall data synchronization delay does not exceed 10 minutes, achieving near real-time data exchange between the school and the company.

Claims

1. An industry-education integration job skills training platform based on artificial intelligence, characterized in that: include: (1) Resource management module, used to classify and store job skill standards, enterprise project cases and teaching courseware; (2) Path planning module: Generates personalized training task sequence based on learners’ historical performance data; (3) Scene simulation module: Construct a 3D virtual work environment and connect it to the enterprise's real-time production interface; (4) Behavior analysis module: collects operation logs and assesses skill mastery. (5) Collaborative guidance module, supporting online joint teaching between college teachers and enterprise engineers; (6) Data exchange module, breaking down the data barriers between the campus academic affairs system and the enterprise business system.

2. The industry-education integration job skills training platform based on artificial intelligence according to claim 1, characterized in that, The path planning module is implemented in the following way: Construct learner competence radar charts to quantify their knowledge structure and skill gaps; The optimal task chain is generated by using a dynamic programming algorithm combined with job competency weights. Adjust the difficulty and resource allocation of subsequent tasks in real time based on the completion status.

3. The industry-education integration job skills training platform based on artificial intelligence according to claim 1, characterized in that, The scene simulation module includes: An interactive production line simulation environment was built using the Unity3D engine. Real-time operating data is obtained by connecting to the enterprise's MES system via the OPC protocol; It supports multi-terminal access from PCs, mobile devices, and VR devices.

4. The industry-education integration job skills training platform based on artificial intelligence according to claim 1, characterized in that, The behavior analysis module employs the following methods: Data collection techniques are used to track user clickstreams, dwell time, and error distribution. Establish an operational compliance determination model based on Markov chains; Automatically generate training reports containing improvement suggestions and push them to the user's device.

5. The industry-education integration job skills training platform based on artificial intelligence according to claim 1, characterized in that, The collaborative guidance module includes: A role-based task distribution mechanism; A remote collaboration interface integrating audio and video calls and an electronic whiteboard; It supports full screen recording and playback of the teaching process.

6. The industry-education integration job skills training platform based on artificial intelligence according to claim 1, characterized in that, The data exchange module is implemented in the following way: Establish a unified data platform and standardize the field formats of different systems; HTTPS protocol and digital certificates are used to ensure secure transmission; It supports two data update mechanisms: timed synchronization and event-triggered updates.

7. A method for industry-education integration in job skills training, characterized in that, Includes the following steps: ‌ Step S1: Construct learner competency profiles and identify skill gaps; Step S2: Generate a training path plan that matches the job requirements; Step S3: Complete the specified task in the simulation environment; Step S4: Generate a capability assessment report based on operational behavior data; Step S5: Provide guidance on difficult issues through a dual-teacher collaborative mechanism; Step S6: Synchronize the learning outcomes data to the enterprise talent pool.

8. The industry-education integration job skills training method according to claim 7, characterized in that, In step S2, a reinforcement learning algorithm is used, with the optimization goal of improving task completion efficiency and capability.

Images