High-speed railway four-electricity system intelligent teaching system and method based on AI

Abstract

The invention relates to the technical field of railway maintenance education, in particular to a high-speed railway four-electricity system intelligent teaching system and method based on AI, and the method comprises the steps: a teaching data obtaining module collects teaching process data, student learning tracks and industry fault data; the virtual simulation modeling module is fused with a BIM model and a technical specification text to construct a four-electric system twinborn body; the AI intelligent analysis engine constructs student ability portraits through a machine learning algorithm, generates a personalized learning path and intelligently answers questions; the virtual-real teaching interaction module builds a digital-intelligent practical training scene, carries out three-dimensional visual error correction and remote guidance through real-time motion recognition, and outputs practical training data. The cross-professional fault simulation module carries out cross-professional fault linkage simulation deduction through an association rule mining algorithm and outputs a deduction result; and the teaching effect evaluation module establishes a binary evaluation system combining process and practical operation skill assessment, and outputs teaching effect data. Therefore, the problems of fixed teaching strategy, low precision and the like in the prior art are solved.

Description

Technical Field

[0001] This invention relates to the field of railway maintenance education technology, specifically to an AI-based intelligent teaching system and method for the four electrical systems of high-speed railways. Background Technology

[0002] As railway networks advance towards intelligence and high speed, higher standards are being set for the operation and maintenance training of the four electrical systems (electric system, signal transmission, communication, and electronic control system) in terms of real-time status perception, dynamic skill adaptation, handling of complex scenarios, and efficient capability enhancement. This requires real-time linkage between teaching content and on-site working conditions to quickly respond to diverse fault scenarios; it needs broad teaching adaptability, covering the entire skill spectrum from basic principles to complex fault troubleshooting; it needs to ensure the stability of long-term training effects and guarantee the long-term retention of knowledge and skills; and it needs to improve the utilization efficiency of teaching resources to achieve personalized capability development. Therefore, there is an urgent need for an adaptive teaching system that can perceive trainees' learning status in real time, dynamically optimize teaching content delivery, and quickly adapt to complex railway scenarios to ensure the efficient implementation of the four electrical systems operation and maintenance training in diverse railway environments.

[0003] However, traditional railway electrical systems training has inherent flaws: rigid teaching strategies rely solely on a unified syllabus and fixed practical projects, failing to integrate multi-dimensional information such as trainee ability data, on-site load characteristics, and environmental parameters. This makes it difficult to adapt to the complex and ever-changing needs of railway operation and maintenance, resulting in low teaching accuracy and insufficient skill transfer efficiency. Furthermore, the teaching hardware has low fidelity in replicating simulated scenarios such as strong electromagnetic interference, wide temperature fluctuations, and continuous vibrations in railways. Equipment performance drift and signal transmission delays significantly reduce the realism of practical training, affecting trainees' perception and response capabilities to on-site conditions. The teaching resource management model is inefficient; passive knowledge transmission is energy-intensive, and active skills training lacks dynamic adaptability, leading to wasted teaching resources and exacerbating trainee fatigue. Moreover, the lack of redundant teaching path design means that a single teaching module failure can cause training interruptions, with an average annual outage duration far exceeding the stringent requirements of the railway system for training maintenance personnel. Overall, the technology faces multiple challenges: rigid teaching strategies, low teaching accuracy, and insufficient utilization of teaching resources. Summary of the Invention

[0004] This application provides an AI-based intelligent teaching system and method for the four electrical systems of high-speed railways, in order to solve the problems of rigid teaching strategies and low accuracy in existing technologies.

[0005] The first aspect of this application provides an AI-based intelligent teaching system for the four electrical systems of a high-speed railway, comprising: a teaching data acquisition module, a virtual simulation modeling module, an AI intelligent analysis engine, a virtual-real teaching interaction module, a cross-disciplinary fault simulation module, and a teaching effectiveness evaluation module; wherein, the teaching data acquisition module is used to collect teaching process data, student learning trajectories, and real-world industry fault data; the virtual simulation modeling module is used to construct a digital twin of the four electrical systems of a high-speed railway based on the teaching process data, student learning trajectories, and real-world industry fault data, integrating BIM models and railway technical specification text data; the AI ​​intelligent analysis engine is used to analyze teaching data based on the digital twin and construct a student... The system comprises three modules: a competency profile, a personalized learning path, and intelligent Q&A; a virtual-real teaching interaction module, which builds personalized digital training scenarios based on the student's competency profile and personalized learning path, conducts VR immersive equipment disassembly and assembly training, performs 3D visualization error correction and AR remote dynamic teaching guidance through real-time motion recognition, and outputs student training data; a cross-disciplinary fault simulation module, which simulates and dynamically extrapolates cross-disciplinary fault linkages based on the student training data using association rule mining algorithms, and outputs fault extrapolation results; and a teaching effectiveness evaluation module, which establishes a binary evaluation system combining process assessment and practical skills assessment based on the fault extrapolation results and student training data, and quantitatively outputs teaching effectiveness data.

[0006] Preferably, the teaching data acquisition module includes a teaching process data collection unit, a learning trajectory tracking unit, and a fault data integration unit. The teaching process data collection unit is used to collect real-time classroom interaction data, practical training operation behavior data, and theoretical assessment answer data. The learning trajectory tracking unit is used to record students' learning time, knowledge point mastery progress, frequency of repeated practical training operations, and types of incorrect operations. The fault data integration unit is used to summarize real fault cases, fault diagnosis processes, and maintenance and handling data from railway power supply, signaling, communication, and traction power supply specialties.

[0007] Preferably, the virtual simulation modeling module includes a twin construction unit and a specification fusion unit. The twin construction unit is used to reconstruct the equipment layout, pipeline routing, and electrical connection relationships of the high-speed railway's four electrical systems based on the BIM model, and integrates teaching process data, student learning trajectories, and real industry fault data to construct a high-precision digital twin. The specification fusion unit is used to extract equipment parameter standards, construction process requirements, and fault handling criteria from railway technical specification texts and embed the constraints of the digital twin.

[0008] Preferably, the AI ​​intelligent analysis engine includes a competency profile building unit, a learning path generation unit, and an intelligent Q&A unit. The competency profile building unit uses decision tree algorithms and neural network models to analyze students' learning trajectories and practical training data, quantifying weaknesses in knowledge points and shortcomings in practical skills to construct a student competency profile. The learning path generation unit dynamically generates personalized learning paths based on the student competency profile and industry job requirements, including knowledge point completion, targeted practical training, and fault simulation exercises. The intelligent Q&A unit uses a railway professional knowledge base and natural language processing algorithms to respond to students' theoretical questions and practical difficulties, providing accurate Q&A explanations and recommended reference materials.

[0009] Preferably, the virtual-real teaching interaction module includes a personalized training scenario building unit, a VR immersive training unit, a motion recognition and error correction unit, and an AR remote guidance unit. The personalized training scenario building unit generates suitable training scenarios for the disassembly, assembly, and maintenance of the four-electric system based on student ability profiles and learning paths. The VR immersive training unit provides a realistic virtual disassembly and assembly environment for equipment, allowing for component disassembly and assembly training using virtual hand models. The motion recognition unit captures student training actions in real time using a skeletal key point recognition algorithm, comparing them with standard operating procedures for 3D visualization and error correction. The AR remote guidance unit provides real-time voice guidance, operation step prompts, and virtual annotations via AR remote guidance based on error correction results and learning progress, recording student training data including operation accuracy, completion time, and error frequency.

[0010] Preferably, the cross-disciplinary fault simulation module includes a fault correlation analysis unit, a linkage simulation unit, and a simulation result output unit. The fault correlation analysis unit is used to identify the coupling relationship between faults of different disciplines based on trainee training data and through association rule mining algorithms. The linkage simulation unit is used to simulate the transmission path and impact range of cross-disciplinary faults, and to conduct dynamic simulation and emergency response simulation. The simulation result output unit is used to output the fault evolution process, key handling nodes, and fault simulation results for evaluating the rationality of the solution.

[0011] Preferably, the teaching effectiveness evaluation module includes a process assessment unit, a practical skills assessment unit, and an evaluation integration unit. The process assessment unit quantifies the student's mastery of knowledge points, learning attitude, and self-learning ability based on the student's learning trajectory, classroom interaction data, and stage-by-stage practical training performance. The practical skills assessment unit evaluates equipment operation proficiency, fault diagnosis accuracy, and cross-disciplinary collaborative handling capabilities based on fault simulation results and practical training operation data. The evaluation integration unit integrates the process assessment and practical skills assessment results to generate quantitative teaching effectiveness data including skill attainment rate, ability improvement, and job suitability.

[0012] The second aspect of this application provides an AI-based intelligent teaching method for the four electrical systems of a high-speed railway, comprising: acquiring teaching process data, student learning trajectories, and real-world industry fault data; based on the teaching process data, student learning trajectories, and real-world industry fault data, integrating BIM models and railway technical specification text data to construct a digital twin of the four electrical systems of a high-speed railway; constructing student competency profiles through machine learning algorithms; generating personalized learning paths and providing intelligent Q&A based on the student competency profiles; building personalized digital training scenarios based on the student competency profiles and the personalized learning paths; conducting VR immersive equipment disassembly and assembly training; performing 3D visualization error correction and AR remote dynamic teaching guidance through real-time motion recognition; outputting student training data; conducting cross-professional fault linkage simulation and dynamic deduction through association rule mining algorithms; and outputting fault deduction results; and establishing a binary evaluation system combining process assessment and practical skills assessment based on the fault deduction results and the student training data, quantitatively outputting teaching effectiveness data.

[0013] A third aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the program to implement an AI-based intelligent teaching method for the four electrical systems of a high-speed railway as described in the above embodiments.

[0014] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement an AI-based intelligent teaching method for the four electrical systems of a high-speed railway as described in the above embodiments.

[0015] Therefore, this application has the following beneficial effects: This application's embodiments collect multi-source teaching and fault data through a teaching data acquisition module, and construct a digital twin of the high-speed railway's four electrical systems (electric system, communication system, and electronic control system) using a virtual simulation modeling module, providing high-precision, scenario-based digital support for teaching. A personalized digital training scenario is built using a virtual-real teaching interaction module, and cross-disciplinary fault simulation is implemented using a cross-disciplinary fault linkage simulation module, effectively improving the adaptability of training content to complex on-site conditions. An AI intelligent analysis engine constructs student competency profiles and generates personalized learning paths, combined with a binary quantitative assessment system for teaching effectiveness evaluation, accurately identifying and optimizing learning weaknesses. This solves the problems of traditional teaching scenarios being singular and lacking cross-disciplinary fault drills, while improving teaching relevance and skill conversion efficiency. It can significantly shorten the training cycle for maintenance personnel, enhance their ability to handle complex cross-disciplinary faults, and provide a solid talent guarantee for the safe and stable operation of the high-speed railway's four electrical systems. Thus, it solves the problems of rigid teaching strategies and low accuracy in existing technologies.

[0016] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0017] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a schematic diagram of the structure of an AI-based intelligent teaching system for the four electrical systems of a high-speed railway, according to an embodiment of this application. Figure 2 This is a schematic diagram of a teaching data acquisition module according to an embodiment of this application; Figure 3 This is a schematic diagram of a virtual simulation modeling module provided according to an embodiment of this application; Figure 4 This is a schematic diagram of an AI intelligent analysis engine provided according to an embodiment of this application; Figure 5 This is a schematic diagram of a virtual-real teaching interaction module provided according to an embodiment of this application; Figure 6 This is a schematic diagram of a cross-disciplinary fault simulation module provided according to an embodiment of this application; Figure 7 This is a schematic diagram of a teaching effectiveness evaluation module provided according to an embodiment of this application; Figure 8 This is a flowchart of an AI-based intelligent teaching system for the four electrical systems of a high-speed railway, according to an embodiment of this application. Figure 9 A flowchart illustrating an AI-based intelligent teaching method for the four electrical systems of a high-speed railway, according to an embodiment of this application; Figure 10 This is a schematic diagram of an AI-based intelligent teaching method for the four electrical systems of a high-speed railway, according to an embodiment of this application. Figure 11 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] The following describes an AI-based intelligent teaching system and method for the four electrical systems of a high-speed railway, referring to the accompanying drawings. Addressing the problem of rigid teaching strategies mentioned in the background art, this application provides an AI-based intelligent teaching system for the four electrical systems of a high-speed railway. In this system, a teaching data acquisition module collects multi-source teaching and fault data, and a virtual simulation modeling module constructs a digital twin of the four electrical systems of a high-speed railway, providing high-precision, scenario-based digital support for teaching. A virtual-real teaching interaction module builds personalized digital training scenarios, and a cross-professional fault simulation module enables cross-professional fault linkage simulation, effectively improving the adaptability of training content to complex on-site conditions. An AI intelligent analysis engine constructs student competency profiles and generates personalized learning paths, combined with a binary quantitative assessment system for teaching effectiveness evaluation, accurately identifying and optimizing learning weaknesses. This solves the problems of single traditional teaching scenarios and lack of cross-professional fault drills, while improving teaching relevance and skill conversion efficiency. It can significantly shorten the training cycle for maintenance personnel, enhance their ability to handle complex cross-professional faults, and provide a solid talent guarantee for the safe and stable operation of the four electrical systems of a high-speed railway. Thus, it solves the problems of rigid teaching strategies and low accuracy in the prior art.

[0020] Figure 1 This is a schematic diagram of the structure of an AI-based intelligent teaching system for the four electrical systems of a high-speed railway, provided as an embodiment of this application.

[0021] This application provides an AI-based intelligent teaching system for the four electrical systems of a high-speed railway. The system 10 includes: The module includes: 100 modules for acquiring teaching data, 200 modules for virtual simulation modeling, 300 modules for AI intelligent analysis engine, 400 modules for virtual and real teaching interaction, 500 modules for cross-disciplinary fault simulation, and 600 modules for teaching effectiveness evaluation.

[0022] The system comprises the following modules: a teaching data acquisition module 100, which collects teaching process data, student learning trajectories, and real-world industry fault data; a virtual simulation modeling module 200, which, based on the teaching process data, student learning trajectories, and real-world industry fault data, integrates BIM models and railway technical specification text data to construct a digital twin of the high-speed railway's four electrical systems; an AI intelligent analysis engine 300, which analyzes teaching data based on the digital twin, constructs student competency profiles through machine learning algorithms, generates personalized learning paths, and provides intelligent Q&A; a virtual-real teaching interaction module 400, which, based on the student competency profiles and personalized learning paths, builds personalized digital training scenarios, conducts VR immersive equipment disassembly and assembly training, performs 3D visualization error correction and AR remote dynamic teaching guidance through real-time action recognition, and outputs student training data; a cross-disciplinary fault simulation module 500, based on the student training data, performs cross-disciplinary fault linkage simulation and dynamic deduction through association rule mining algorithms, and outputs fault deduction results; and a teaching effectiveness evaluation module 600, based on the fault deduction results and student training data, establishes a binary evaluation system combining process assessment and practical skills assessment, and quantitatively outputs teaching effectiveness data.

[0023] It is understood that in this embodiment, the teaching data acquisition module collects multi-source teaching and fault data, and combines it with the virtual simulation modeling module to construct a digital twin of the high-speed railway's four electrical systems, providing high-precision and scenario-based digital support for teaching. The virtual-real teaching interaction module builds personalized digital training scenarios, and the cross-professional fault simulation module enables cross-professional fault linkage simulation, effectively improving the adaptability of training content to complex on-site conditions. The AI ​​intelligent analysis engine constructs student competency profiles and generates personalized learning paths, combined with the binary quantitative assessment system of the teaching effectiveness evaluation module, accurately identifying and optimizing learning weaknesses. This solves the problems of traditional single teaching scenarios and lack of cross-professional fault drills, while improving the relevance of teaching and the efficiency of skill transformation. It can significantly shorten the training cycle for maintenance personnel, enhance their ability to handle complex cross-professional faults, and provide a solid talent guarantee for the safe and stable operation of the high-speed railway's four electrical systems. Thus, it solves the problems of rigid teaching strategies and low accuracy in existing technologies.

[0024] In this embodiment of the application, the teaching data acquisition module 100 includes: Figure 2 As shown, there are three units: a teaching process data acquisition unit, a learning trajectory tracking unit, and a fault data integration unit.

[0025] The teaching process data acquisition unit is used to collect classroom interaction data, practical training operation behavior data, and theoretical assessment answer data in real time; the learning trajectory tracking unit is used to record students' learning time, knowledge point mastery progress, frequency of repeated practical training operations, and error operation types; and the fault data integration unit is used to summarize real fault cases, fault diagnosis processes, and maintenance and handling data of railway power supply, signaling, communication, and traction power supply.

[0026] It is understood that the embodiments of this application, through the teaching process data acquisition unit, collect real-time data on classroom interaction, practical training behavior, and theoretical assessment answers, so as to keep abreast of the dynamics of the teaching process and provide support for the preliminary assessment of teaching effectiveness; the learning trajectory tracking unit accurately records students' learning time, progress in mastering knowledge points, frequency of repetition of practical training operations, and types of errors, clearly depicting personalized learning status and providing a basis for differentiated instruction; the fault data integration unit summarizes real fault cases, fault diagnosis processes, and maintenance and handling data of railway power supply, signaling, communication, and traction power supply, enriching the real scenarios of practical training, strengthening students' practical fault handling capabilities, and providing comprehensive raw data for optimizing teaching models, implementing personalized teaching, and improving the practical training system, thereby enhancing the pertinence and effectiveness of teaching.

[0027] For example, in the intelligent teaching process of the four electrical systems of high-speed railways, the learning trajectory tracking unit can dynamically collect students' practical operation data, knowledge mastery progress, and skill application frequency throughout the entire process. When students omit steps or deviate from operations during contact network fault troubleshooting practice, the tracking unit can quickly locate their weak knowledge points and the transmission path of the errors. Compared with traditional manual spot checks, which can only find surface problems, it can more accurately reflect the characteristics of students' ability gaps. If students have cognitive misconceptions during signal system parameter debugging training, the tracking unit can predict potential obstacles to skill solidification through the deviation trend of learning behavior. This dynamic learning data provides accurate input for the learning simulation of the teaching twin, helping the intelligent teaching center to formulate personalized reinforcement training plans, thereby intervening in inefficient learning states in advance, reducing practical errors caused by ability gaps, and significantly improving the training quality and job suitability of personnel in the four electrical systems of high-speed railways.

[0028] In this embodiment of the application, the virtual simulation modeling module 200 includes: Figure 3 As shown, twin building blocks and standard fusion units.

[0029] Among them, the twin construction unit is used to restore the equipment layout, pipeline routing and electrical connection relationship of the high-speed railway's four electrical systems based on the BIM model, and integrate teaching process data, student learning trajectory and real industry fault data to construct a high-precision digital twin; the standard integration unit is used to extract equipment parameter standards, construction process requirements and fault handling criteria from railway technical specification texts and embed the constraints of the digital twin.

[0030] It is understood that the embodiments of this application, through the twin construction unit, accurately recreate the equipment layout, pipeline routing, and electrical connection relationships of the high-speed railway's four electrical systems based on the BIM model. They integrate teaching process data, student learning trajectories, and real-world industry fault data to construct a high-precision digital twin, achieving precise digital replication of the physical system and providing a core carrier for building immersive virtual teaching and training scenarios. The standard integration unit extracts equipment parameter standards, construction process requirements, and fault handling guidelines from railway technical specifications and embeds them into the constraints of the digital twin, giving the virtual simulation scenario professional industry standard attributes. This prevents training operations from deviating from standard requirements, providing standard support for standardized and regulated intelligent teaching implementation, and overall improving the accuracy, standardization, and teaching adaptability of virtual simulation modeling.

[0031] It should be noted that the construction of a high-precision digital twin of the high-speed railway's electrical, electronic, and communication systems based on a BIM model begins with extracting spatial layout parameters (such as power supply cabinet installation locations, signal placement points, and communication base station installation heights), pipeline routing and laying parameters (such as cable trench routing, contact wire support spacing, and optical cable laying routes), and topological and interface parameters of electrical connections (such as power supply circuit wiring methods, signal interlocking logic relationships, and communication equipment networking protocols) from the BIM model to form the initial basic model of the digital twin. Simultaneously, data from the teaching process (classroom interaction data, practical training behavior data, and theoretical assessment answer data), student learning trajectory data (learning time, knowledge point mastery progress, frequency of repeated practical training operations, and types of errors) are integrated. Real-world fault data (real fault cases, fault diagnosis processes, and maintenance data from railway power supply, signaling, communication, and traction power supply specialties) is categorized, analyzed for correlation, and standardized through mapping. This data is then transformed into teaching and training feature data that can be embedded in a digital twin. Subsequently, the initial basic model is registered, correlated, and integrated with the three types of feature data mentioned above. This corrects the problem of missing teaching scenarios caused by model abstraction. By integrating key information such as equipment physical attributes, electrical operation logic, teaching interaction rules, and fault evolution mechanisms, a high-precision digital twin is finally constructed that can accurately map the geometry, electrical characteristics, operating status, and teaching and training scenarios of the four electrical systems of high-speed railways. This provides a high-precision digital carrier for virtual simulation teaching, practical operation simulation, fault diagnosis drills, and teaching effectiveness evaluation. For example, taking the virtual training of fault diagnosis in the overhead contact system of high-speed railway as an example, the system uses natural language processing algorithms to accurately extract equipment parameter standards such as the allowable deviation of the contact wire height (±30mm) and the limit of the installation slope of the droppers, based on the clauses and details of documents such as the "Railway Electric Traction Power Supply Design Specification" and the "High-speed Railway Overhead Contact System Operation and Maintenance Regulations". It also extracts the core fault handling process of "power outage isolation - voltage testing and grounding - fault location - repair and reset", while capturing the constraints of priority handling criteria for different fault types on the training operation. The concurrently launched rule-structured conversion module transforms the extracted specification content into constraint codes recognizable by the digital twin based on parameter classification logic, clearly presenting core constraint elements such as equipment parameter threshold ranges, operation process verification nodes, and fault handling judgment criteria. Subsequently, the parameter constraint code was bound to the attribute fields of the overhead contact line geometric model, and the process verification logic was written into the interactive control module of the virtual training. Through rule matching and real-time verification, a linkage mechanism between standard constraints and virtual operations was constructed to monitor key behaviors of students in the training, such as the accuracy of guide height adjustment and the compliance of operating steps, including parameter over-limit warning trigger points and the activation conditions of the process skipping violation judgment logic. The final output includes not only records of students' violations in the training operation and accurate annotation of corresponding standard clauses, but also a compliance score report of the training process. This data provides quantitative support for optimizing the design of virtual training teaching plans for the four electrical systems, adjusting the focus of standardized teaching to reduce the student's violation rate in practical operations, and improving the industry adaptability of talent training.

[0032] In this embodiment of the application, the AI ​​intelligent analysis engine 300 includes: Figure 4 As shown, there are three units: capability profile building unit, learning path generation unit, and intelligent Q&A unit.

[0033] The unit comprises three components: a competency profile building unit and a learning path generation unit. The former uses decision tree algorithms and neural network models to analyze students' learning trajectories and practical training data, quantify weak points in knowledge and shortcomings in practical skills, and build a competency profile for each student. The latter uses a learning path generation unit to dynamically generate personalized learning paths that include knowledge point supplementation, targeted practical training, and fault simulation exercises, based on the student's competency profile and industry job requirements. The former uses an intelligent Q&A unit to respond to students' theoretical questions and practical difficulties based on a railway professional knowledge base and natural language processing algorithms, providing accurate Q&A explanations and reference material recommendations.

[0034] It is understood that the competency profile construction unit in this application relies on decision tree algorithms and neural network models to deeply analyze students' learning trajectories and practical training data, quantitatively locate weak points in knowledge and shortcomings in practical skills, and accurately construct multi-dimensional student competency profiles, providing a core basis for personalized teaching implementation; the learning path generation unit, based on the generated student competency profiles and the core requirements of the high-speed railway electrical system industry positions, dynamically plans a tiered personalized learning path that includes targeted knowledge point supplementation, targeted practical training reinforcement, and typical fault simulation exercises, avoiding blind learning and improving efficiency; the intelligent Q&A unit, relying on the railway professional knowledge base and natural language processing algorithms, responds in real time to students' theoretical knowledge questions and practical training confusion, outputs accurate Q&A analysis and supporting reference material recommendations, promptly removes learning obstacles, improves the intelligence, personalization and efficiency of teaching guidance, and helps cultivate professional and technical talents that meet industry needs.

[0035] It should be noted that the decision tree algorithm formula is as follows: ; ; ; ; in, For sample set Information entropy; The number of categories in the sample; For sample set The Middle The proportion of samples of each class; It is a logarithmic function with base 2; For index symbols; For sample set Based on features Information gain from splitting; For sample set Information entropy; Features The set of all possible values; For sample set Chinese characteristics Pick A subset of values; For subset The number of samples; For sample set The total number of samples; For subset Information entropy; For sample set The Gini coefficient; For sample set Based on features Split Gini gain; For subset The Gini coefficient; Let A be the value of feature A.

[0036] Neural network model formula: ; ; ; ; ; ; in, For the first The linear output of the layer; For the first Layer weight matrix; For the first Layer activation output; For the first Layer bias terms; This serves as an index for the network's layers. For the first Layer activation output; To modify the activation function of the linear unit; This is a function to find the maximum value. The active output of the output layer (layer T); Use the Sigmoid activation function; This is the linear output of the output layer (the T-th layer); is a natural constant; T is the layer index of the output layer; The value of the loss function; The number of samples; For the first The true label of each sample; For the first Predicted values ​​for each sample; For index symbols; For the first Layer weight matrix; The learning rate; loss function Weights The partial derivatives; This serves as an index for the network's layers. For the first Layer bias terms; loss function For bias The partial derivatives of .

[0037] Natural Language Processing Algorithm Formula: ; ; ; ; ; ; ; ; ; ; in, The vector corresponding to the text; For vocabulary The number of times it appears in the current text; The first word in the predefined dictionary; It is the second word in the predefined dictionary; The nth word in the predefined dictionary; The total number of words in the dictionary; For vocabulary The number of times it appears in the current text; For vocabulary The number of times it appears in the current text; For vocabulary In the text and corpus TF-IDF value in; For vocabulary In the text Word frequency in; For vocabulary In the corpus Inverse document frequency; For target vocabulary; The current text; For the entire corpus; For vocabulary In the text The number of times it appears in; For text The total number of occurrences of all words in the text; For text Any word in the text; It is the natural logarithm; For corpus Total number of documents in the database; For corpus Contains vocabulary The number of documents; For vectors and Cosine similarity; Let the vector represent the student's questions; For the knowledge base A vector of answers; It is the dot product of two vectors; For vectors L2 norm (modulus); For vectors L2 norm (modulus); For vectors The One component; For vectors The One component; Let be the dimension of the vector; For index symbols; For set and The Jaccard similarity coefficient; A collection of vocabulary for students' questions; A vocabulary set for the answers in a knowledge base; The size of the intersection of the two sets; The size of the union of the two sets; For given text features At that time, it belongs to the intention category. The posterior probability; For a given intent category At that time, text features appeared The likelihood probability; For Intent Category The prior probability; Text features Evidence factors; The vocabulary characteristics that students have questions about; Intent category; For the question of the first One word; For a given intent category When words appear The probability of; The number of words in the question; Given a linear output At that time, it belonged to the first The probability of class intent; Use the Softmax activation function; For the model to the first Linear output of class intent; The total number of intent categories; For the model to the first Linear output of class intent; This is the output of the self-attention mechanism; For query vector; The key vector; For querying the matrix multiplication of the key vector and the query vector; Key vector The dimension; is the scaling factor; j is the summation index.

[0038] The learning path generation unit is based on a multi-dimensional competency profile that accurately identifies students' weaknesses. It deeply aligns with the core skill requirements of positions in the high-speed railway's four electrical systems (power supply, signaling, communication, and traction power supply). Through an intelligent matching algorithm that matches skill gaps with job competency standards, it prioritizes identifying students' core weaknesses in areas such as professional theoretical knowledge, equipment operation skills, and fault diagnosis and handling. Subsequently, it dynamically plans a tiered and progressive learning content system based on the characteristics of different weaknesses. For modules where knowledge points are not firmly grasped, it provides customized theoretical lectures and specialized practice exercises to fill knowledge gaps. For areas where practical skills are insufficient, it designs scenarios that closely resemble real-world job positions. Jing's targeted training projects aim to strengthen hands-on skills. For areas with weak fault handling capabilities, it selects typical real-world fault cases from the industry to construct highly realistic fault simulation drills to improve emergency response levels. At the same time, it tracks students' learning progress and training effects in real time, and dynamically adjusts the difficulty level of learning content, the frequency of training projects, and the complexity of fault drills based on the results of phased ability assessments. Ultimately, it generates a personalized learning path that is adapted to students' individual ability levels and meets the actual needs of industry positions, greatly improving the relevance and efficiency of learning, helping students quickly make up for their skill gaps, and grow into professional and technical personnel who meet the requirements of high-speed railway electrical systems.

[0039] In this embodiment of the application, the virtual-real teaching interaction module 400 includes, as follows: Figure 5 As shown, the units include a personalized training scenario building unit, a VR immersive training unit, a motion recognition and error correction unit, and an AR remote guidance unit.

[0040] The personalized training scenario building unit generates suitable training scenarios for the disassembly, assembly, and maintenance of the four electrical systems based on students' ability profiles and learning paths; the VR immersive training unit provides a realistic virtual disassembly and assembly environment for equipment, allowing students to train in disassembling and assembling components using virtual hand models; the motion recognition unit captures students' training actions in real time using a skeletal key point recognition algorithm, compares them with standard operating procedures, and performs 3D visualization error correction; and the AR remote guidance unit provides real-time voice guidance, operation step prompts, and virtual annotations remotely via AR based on error correction results and learning progress, recording student training data including operation accuracy, completion time, and error frequency.

[0041] It is understood that the personalized training scenario building unit in this application generates a suitable four-electric system disassembly, assembly, and maintenance training scenario based on student competency profiles and customized learning paths, accurately matching students' competency gaps and learning needs; the VR immersive training unit creates a highly realistic virtual equipment disassembly and assembly environment, driving virtual hand models to complete practical training such as component disassembly and assembly, allowing students to immerse themselves in the job operation process and avoid the safety risks and wear and tear costs of operating physical equipment; the action recognition and error correction unit relies on the skeletal key point recognition algorithm to capture students' training action trajectories in real time, accurately compares them with standard operating procedures, and achieves three-dimensional visualization error correction, promptly correcting non-standard operating habits; the AR remote guidance unit, based on the action error correction results and real-time learning progress, provides voice guidance, operation step prompts, and virtual annotations through AR, and simultaneously records training data such as operation accuracy, completion time, and error frequency, providing a basis for subsequent competency profile optimization and learning path adjustment, improving the immersion, standardization, and personalization of practical training, and efficiently cultivating students' job practical skills.

[0042] It should be noted that the personalized training scenario construction unit first performs a multi-dimensional analysis of the input student competency profile, accurately extracting the students' knowledge gaps, weak practical skills, and fault handling capabilities in each professional module of the high-speed railway's four electrical systems (power supply, signaling, communication, and traction power supply). Simultaneously, it deeply breaks down the clearly defined phased learning objectives, key skill enhancement areas, and assessment standards within the personalized learning path. Subsequently, it correlates and matches competency gaps with learning objectives, defining corresponding training scenario types and complexity levels for students at different competency levels. For example, for students with weak foundations in overhead contact line disassembly and assembly, a basic disassembly and assembly scenario including a single support column and dropper components is matched, focusing on basic skills training such as component identification and disassembly / assembly sequence; for students whose traction power supply system maintenance capabilities need improvement, a similar scenario is created. For students, the training program integrates daily equipment inspections, parameter adjustments, and common tripping fault location and handling scenarios to enhance multi-stage collaborative operation capabilities. For students who have already mastered basic skills, it matches advanced training scenarios involving multi-equipment linkage operations and complex fault superposition to hone emergency response and system debugging capabilities. Then, it retrieves equipment models, pipeline topologies, and fault case resource libraries from the digital twin of the high-speed railway's four electrical systems, and configures the equipment composition, operation process, fault triggering mechanism, and assessment indicator thresholds of the scenarios based on the matching results. Finally, through parameterized assembly and logical verification of the scenarios, it generates personalized four electrical system disassembly, assembly, and maintenance training scenarios that are fully adapted to students' current ability level and stage of learning needs, providing a precisely matched training platform for subsequent VR immersive training, motion recognition and error correction, and other stages.

[0043] Formula for skeletal key point recognition algorithm: ; ; ; ; ; ; ; ; ; in, For the first A key point at pixel coordinates Heatmap values ​​at the location; It is an exponential function; These are the pixel coordinates on the heatmap; For the first The actual coordinates of each key point; The standard deviation of the Gaussian kernel; Index for key points; The loss value predicted by the heatmap; The total number of key points; This represents the width and height of the heatmap; For the first A key point at pixel coordinates Predicted heatmap values ​​at the location; For the first Each skeleton is connected at pixel coordinates The PAF vector at that location; For the first Unit direction vectors connected to each skeleton; For the first The coordinates of the midpoints of the connected bones; Index for skeletal connections; This represents the loss value predicted by PAF. The total number of skeletal connections; For the first Each skeleton is connected at pixel coordinates The predicted PAF vector at the location; It is the L2 norm (vector magnitude). This represents the total loss value of the model; These are the weighting coefficients for the loss; For the first Predicted coordinates of key points; To obtain the coordinates corresponding to the maximum value; For the first Predicted 3D coordinates of key points; For parameters The regression network; It is a set of coordinates of two-dimensional key points; For the depthwise convolutional features of the image; This represents the loss value for 3D keypoint regression; For the first The true 3D coordinates of each key point; The L1 norm (sum of absolute values ​​of coordinate differences); For the first The first frame Deviation at a key point; For the first The first frame Predicted 3D coordinates of key points; For the first The first frame The true 3D coordinates of each key point; This is the time step of a video frame.

[0044] For example, in the training on troubleshooting broken catenary cables in the overhead contact system of a high-speed railway, the AR remote guidance unit plays a crucial role. When a student's improper operation (such as failing to perform grounding before testing for voltage) is captured and marked by the action recognition and correction unit during virtual training, the unit immediately activates the AR remote guidance function based on the real-time correction results and the student's current learning progress in fault handling. Through AR glasses, key operational points for voltage testing and grounding are virtually overlaid in the student's field of vision in real time, and voice guidance prompts are simultaneously pushed, such as "You must first use a voltage tester to confirm the absence of power before installing the grounding wire to complete the safety isolation." The unit also displays the standard handling steps of "power outage - voltage testing - grounding - fault location - repair" in a dynamic flowchart. Simultaneously, the unit records the student's operational accuracy, fault troubleshooting completion time, and frequency of errors in key steps throughout the training. This data can be transmitted back to the AI ​​intelligent analysis engine in real time, providing accurate basis for subsequent iterative optimization of student competency profiles and dynamic adjustment of learning paths. This effectively ensures the standardization of training operations and helps students quickly master the core skills of handling overhead contact system faults.

[0045] In this embodiment of the application, the cross-disciplinary fault simulation module 500 includes, as follows: Figure 6 As shown, there are fault correlation analysis unit, linkage simulation unit, and simulation result output unit.

[0046] The fault correlation analysis unit is used to identify the coupling relationship between different professional faults based on the trainees' training data and through the association rule mining algorithm; the linkage simulation unit is used to simulate the transmission path and impact range of cross-professional faults, and to conduct dynamic simulation and emergency response simulation; the simulation result output unit is used to output the fault evolution process, key nodes of the response, and the rationality assessment of the solution.

[0047] It is understood that the fault correlation analysis unit in this application embodiment relies on the association rule mining algorithm to deeply analyze the student training data, accurately identify the coupling relationship between faults in different specialties such as power supply, signaling, communication, and traction power supply of the four electrical systems of high-speed railway, and break the limitations of single-specialty fault simulation; the linkage simulation unit, based on the fault correlation rules mined, dynamically simulates the transmission path and overall impact range of cross-specialty faults, supports the deduction and practical simulation of emergency response plans in multiple scenarios, and hones students' system thinking and collaborative response capabilities; the deduction result output unit comprehensively outputs the entire process of fault evolution, key nodes of response, and a rationality evaluation report of the plan, providing quantitative basis for optimizing cross-specialty training teaching plans and adjusting teaching focus, improving the systematicness and practicality of fault simulation training, and helping to cultivate compound technical talents that meet industry needs.

[0048] It should be noted that the formula for the association rule mining algorithm is: ; ; ; in, For association rules Support level; for and The probability of them happening simultaneously; For transaction set It contains and The number of transactions; For transaction set Total number of transactions; For the set of preceding faults; For the set of subsequent faults; For association rules Confidence level; For a given hour The conditional probability of occurrence; For transaction set Includes The number of transactions; For association rules The degree of improvement; for Support level; for The probability of occurrence.

[0049] The linkage simulation unit, based on the coupling relationships of faults in the four electrical systems (power supply, signaling, communication, and traction power supply) of high-speed railways mined by the fault correlation analysis unit, first constructs a logical model and parameterized correlation matrix for the transmission of faults across multiple disciplines. It clarifies the triggering conditions, transmission media, and impact thresholds for different types of initial faults. For example, a short-circuit fault in the contact network of the traction power supply system will be transmitted to the signaling system through abnormal voltage in the power supply circuit, causing unstable power supply to the signal and leading to display errors. Simultaneously, it affects the dispatch command transmission link of the communication system. Subsequently, according to the set initial fault scenario, the unit dynamically simulates the entire process of the fault from its occurrence and spread to the formation of a chain reaction, accurately presenting the fault. This unit explores the spatiotemporal evolution of transmission paths and impact ranges between different professional modules, as well as the changes in the operating status of equipment in each professional field. Based on this, the unit supports students in conducting cross-disciplinary emergency response simulations. Students are allowed to take measures such as power outage isolation, fault location, parameter adjustment, and collaborative repair at different stages of a fault. The unit provides real-time feedback on the intervention effect of the response plan on the fault development. If the response measures are inappropriate, the unit simulates the consequences of further fault spread. If the response is timely and effective, the unit demonstrates the entire process of gradually eliminating the fault and restoring the system to normal operation. This approach cultivates students' systemic thinking and cross-disciplinary collaborative response capabilities, allowing them to intuitively understand the complexity of cross-disciplinary faults and the importance of coordinated response.

[0050] In this embodiment of the application, the teaching effectiveness evaluation module 600 includes, for example: Figure 7 As shown, there are process-based assessment units, practical skills assessment units, and integrated evaluation units.

[0051] The process assessment unit is used to quantify the mastery of knowledge points, learning attitude, and self-learning ability based on students' learning trajectory, classroom interaction data, and phased practical training performance; the practical skills assessment unit is used to evaluate equipment operation proficiency, fault diagnosis accuracy, and cross-professional collaborative handling ability based on fault simulation results and practical training operation data; the evaluation integration unit is used to integrate the results of process assessment and practical skills assessment to generate quantitative teaching effectiveness data including skill attainment rate, ability improvement, and job suitability.

[0052] It is understood that the process assessment unit in this application's embodiments uses multi-dimensional data, such as students' entire learning trajectory, frequency and quality of classroom interaction, and performance in phased practical training, to quantitatively evaluate students' mastery of professional knowledge points of the high-speed railway's four electrical systems, their learning engagement, and their ability to plan their own learning. The practical skills assessment unit relies on cross-professional fault simulation results and full-volume practical training data to accurately assess students' proficiency in operating the four electrical equipment, the accuracy of fault diagnosis and location, and their ability to handle multi-professional collaborative situations, focusing on the verification of core job skills. The evaluation integration unit weightedly integrates and cross-verifies the process growth data with the practical skills assessment results to generate a quantitative teaching effectiveness evaluation report containing core indicators such as professional skills attainment rate, ability improvement, and industry job suitability. This provides data support for teachers to optimize teaching plans, adjust practical training priorities, and improve the talent training system, thereby helping to improve the quality and efficiency of professional talent training in the high-speed railway's four electrical systems.

[0053] It should be noted that the process-based assessment unit is supported by the full-process data of intelligent teaching for the high-speed railway's four electrical systems. It comprehensively integrates students' learning trajectory data (covering theoretical learning time, frequency of repeated learning of knowledge points, distribution of incorrect answers and rectification status, and progress of personalized learning paths), classroom interaction data (including classroom question response speed, quality of professional question presentations, participation in group discussions, and frequency and depth of interaction with teachers / students), and phased practical training performance data (such as the completion rate of basic disassembly and assembly, fault diagnosis, etc., process standardization, fault location accuracy, and completeness of practical training reports). Subsequently, through a quantitative scoring model and multi-dimensional data correlation analysis, the scattered data is transformed into measurable ability indicators, such as knowledge point mastery rate and core concept understanding accuracy. The system quantifies students' mastery of professional knowledge points such as power supply, signaling, communication, and traction power supply through metrics like accuracy and error repetition rate. It quantifies students' learning engagement through classroom interaction, practical training attendance, and on-time task completion rates. Furthermore, it quantifies students' self-learning ability through metrics like the rationality of their self-study plans, proactively extending their study time, and the percentage of self-directed problem-solving. Simultaneously, it breaks away from the traditional "one exam determines your future" assessment limitations, tracking students' growth trajectory from basic entry-level to advanced skills, dynamically capturing strengths and weaknesses in the learning process. This not only provides a comprehensive basis for accurately evaluating students' stage-by-stage learning outcomes but also provides data support for teachers to adjust teaching strategies and optimize tutoring directions, ensuring that the assessment results objectively reflect students' true learning status and skill development process.

[0054] The practical skills assessment unit uses the full-process results of fault simulation for the four electrical systems of high-speed railway (power supply, signaling, communication, and traction power supply) and all training operation data as the core evaluation criteria. First, it comprehensively integrates two types of key data: fault simulation results, which cover the rationality of handling solutions for cross-disciplinary faults (such as failure of traction power supply contact network grounding linkage signal interlocking), the effectiveness of fault evolution intervention, the timeliness of key handling nodes, and the potential for solution optimization; and training operation data, which includes the compliance of equipment disassembly and assembly procedures, the accuracy of tool use, the standardization of operational actions, the frequency of repeated errors, the completion time of a single task, and the efficiency of task connection and communication and collaboration records between various professional modules in cross-disciplinary training. Then, a quantitative scoring model is used to accurately assess students' core practical skills: based on the accuracy of equipment operation steps, the standardization of actions, and the completion rate. The assessment comprehensively evaluates the operational proficiency of various equipment in the four electrical systems (such as overhead contact line supports, signal machines, and power supply cabinets) using indicators such as completion time, completion rate, and repetition error rate. It quantifies fault diagnosis accuracy using indicators such as fault location time, the scientific validity of diagnostic criteria, the accuracy of fault type identification, and the conformity of handling plans with industry standards. It evaluates cross-disciplinary collaborative handling capabilities using indicators such as the rationality of task allocation in cross-disciplinary fault handling, the speed of multi-disciplinary module collaborative response, the effectiveness of coordinated handling plans, and team communication efficiency. Ultimately, it forms a quantitative assessment result focusing on core job skills, objectively reflecting students' true level of transforming theoretical knowledge into practical ability, and accurately matching the actual skill requirements of high-speed railway four electrical systems industry positions. This provides core data support for optimizing teaching effectiveness, adjusting practical training focus, and improving the quality of talent cultivation.

[0055] The evaluation integration unit first processes the core data from both process-based assessments and practical skills assessments. It first analyzes multi-dimensional indicators from the process-based assessments, such as "knowledge point mastery rate, quantitative score of learning attitude, and level of self-learning ability," to clarify students' performance in theoretical learning, learning habits, and self-planning within the four electrical systems of high-speed railways (power supply, signaling, communication, and traction power supply). Then, it extracts core skill indicators from the practical skills assessments, such as "equipment operation proficiency score, fault diagnosis accuracy rate, and cross-professional collaborative handling ability rating," to pinpoint students' true levels of practical application and collaborative handling. Subsequently, based on the skill priorities of positions within the four electrical systems of high-speed railways (e.g., practical skills like fault diagnosis and cross-professional collaboration have higher weights than theoretical knowledge mastery), a scientific weighted integration model is established. This model differentiates the weights of the two types of assessment indicators and performs cross-validation, avoiding the bias of a single assessment dimension. (For example, if a student has a solid grasp of knowledge points in the process-based assessment but low fault handling efficiency in practice, the model will objectively reflect their "strong theoretical knowledge but weak practical skills" through weight adjustments.) Based on the characteristics of "weakness", three types of core quantitative teaching effectiveness data are precisely generated: First, the skill attainment rate, which compares students' various assessment indicators with industry job skill standard thresholds to statistically analyze the percentage of students who have mastered core skills such as knowledge points, equipment operation, and fault handling; Second, the improvement in ability, which uses students' initial enrollment assessment data or stage assessment benchmarks as a reference to calculate the growth difference and percentage improvement in dimensions such as theoretical knowledge, practical skills, and self-learning; Third, the job suitability, which precisely matches the assessment results with the core competency requirements of relevant positions in the high-speed railway electrical system (such as overhead contact line maintenance and signal system debugging), quantifying the degree of fit between students' current abilities and the requirements of target positions. The final output quantitative data not only comprehensively integrates "process growth" and "outcome effectiveness", but also realizes the transformation of teaching effectiveness from "qualitative description" to "quantitative evaluation", providing teachers with objective and accurate decision-making basis for optimizing teaching plans, adjusting the allocation of practical training resources, and for colleges and universities to build a talent training system that meets industry needs.

[0056] The formula for the quantitative scoring model is: ; ; ; ; in, As an indicator The standardized value; As an indicator The original value; This is the minimum value of the indicator in the entire sample; This is the maximum value of the indicator in the entire sample; negative indicator The standardized value; For the first The score of each primary indicator; For the first The first primary indicator The weights of each secondary indicator; For the first The first primary indicator Standardized values ​​of each secondary indicator; Index for primary indicators (1 = mastery of professional knowledge, 2 = attitude towards learning, 3 = self-learning ability). This serves as an index for secondary indicators; The overall process assessment score; The weights of the three primary indicators; The scores are for the three primary indicators.

[0057] The formula for the weighted fusion model is: ; ; ; ; in, The score is based on the overall ability after integration. The weighting for process-based assessment; The overall score is based on the process assessment. The weighting for practical skills assessment; The overall score for practical skills assessment; meets the requirements ; For skill qualification rate; The number of indicators required to reach the industry standard threshold; The total number of core indicators for achieving the target; The percentage increase in ability; Score for the current assessment dimension; For initial admission assessment or stage benchmark scores; For job suitability; For the first The weight of each job's core competency indicator; For students Scores for each capability indicator; For the target position Standard thresholds for each capability indicator; The number of core competency indicators for the position; For indexing indicators.

[0058] This application proposes an AI-based intelligent teaching system for the four electrical systems of high-speed railways. It collects multi-source teaching and fault data through a teaching data acquisition module, and constructs a digital twin of the high-speed railway's four electrical systems using a virtual simulation modeling module, providing high-precision, scenario-based digital support for teaching. A virtual-real teaching interaction module builds personalized digital training scenarios, and a cross-disciplinary fault simulation module enables cross-disciplinary fault linkage simulation, effectively improving the adaptability of training content to complex on-site conditions. An AI intelligent analysis engine constructs student competency profiles and generates personalized learning paths. Combined with a binary quantitative assessment system in the teaching effectiveness evaluation module, it accurately identifies and optimizes learning weaknesses. This system solves the problems of traditional teaching scenarios being singular and lacking cross-disciplinary fault drills, while improving teaching relevance and skill conversion efficiency. It can significantly shorten the training cycle for maintenance personnel, enhance their ability to handle complex cross-disciplinary faults, and provide a solid talent guarantee for the safe and stable operation of the high-speed railway's four electrical systems. Therefore, it solves the problems of rigid teaching strategies and low accuracy in existing technologies.

[0059] The following will illustrate an AI-based intelligent teaching system for the four electrical systems of a high-speed railway through a specific embodiment, such as... Figure 8 As shown, it includes: The high-speed railway electrical systems training center of a railway vocational college has long faced the pain points of traditional teaching methods: physical training equipment (such as overhead contact line supports and signal interlocking systems) is expensive and wears out quickly; cross-disciplinary fault scenarios are difficult to reproduce; students face high operational risks during training; and there is a lack of quantitative basis for evaluating teaching effectiveness. To solve these problems, the college introduced an AI-based intelligent teaching system for high-speed railway electrical systems, achieving deep integration of theoretical teaching and practical training, and precise cultivation of cross-disciplinary skills and job suitability. The following are the specific implementation details of the system.

[0060] Deployment and Data Collection Process of Teaching Data Acquisition Module The teaching data acquisition module serves as the system's data foundation. Through a three-pronged architecture of "hardware sensing + software tracking + data integration," it achieves accurate collection and standardized storage of teaching data across all dimensions. The teaching process data acquisition unit deploys multi-dimensional sensing devices: 16 4K high-definition cameras (30fps sampling rate, 120° field of view) are installed in the training workshop to capture students' practical operations and classroom interactions in real time; each training terminal integrates a microphone (48kHz audio sampling rate, noise suppression threshold ≤30dB) to record voice data such as teacher-student Q&A and group discussions; the theoretical assessment system collects student answer data through an online quiz platform (question types include single-choice, multiple-choice, and practical operation process judgment questions, with answer time accuracy down to the millisecond level), simultaneously recording details such as the distribution of incorrect answers and answer duration. The learning trajectory tracking unit records students' learning behavior in real time through data integration with the college's academic affairs system and practical training management platform: theoretical learning time (accurate to the minute), knowledge point click frequency (such as the number of times core knowledge points such as catenary structure and signal machine principle are accessed), practical training operation repetition frequency (such as the number of times a certain equipment is disassembled and assembled), and error operation type (categorized as parameter setting error, process skipping error, tool usage error, etc.). All data is uploaded in real time through WiFi 6 wireless communication network (transmission rate ≥1.2Gbps, latency ≤20ms). The fault data integration unit has established data cooperation with a certain railway bureau group and high-speed rail construction companies, and has compiled more than 1,200 real fault cases of railway electrical systems. These cases cover four major categories of faults: contact network grounding faults in power supply, interlocking system failures in signaling, dispatching link interruptions in communication, and substation tripping in traction power supply. Each case includes the fault occurrence scenario, diagnosis process, maintenance and handling steps, and corresponding standard basis. After being structured (stored according to "fault type-triggering condition-handling process-standard number"), the data is stored in the system's distributed database (supporting tens of millions of concurrent data accesses with data storage latency ≤5ms), providing real data support for subsequent modeling and simulation.

[0061] Virtual simulation modeling module construction details The virtual simulation modeling module, through the collaboration of the twin construction unit and the standard fusion unit, creates a digital twin of the four electrical systems that is highly consistent with the real scene. The twin construction unit is based on the actual engineering CAD drawings of the four electrical systems of a high-speed railway line. It uses laser 3D scanning technology (scanning accuracy ±0.03mm) to scan the physical equipment such as the catenary, signal machines, and power supply cabinets in the college's training base, acquiring full-element 3D point cloud data. Through point cloud registration (registration error ≤0.05mm), mesh reconstruction (triangular mesh density ≥1000 points / cm²), and parameter optimization, a digital twin is constructed that includes equipment layout (such as the spacing between catenary support pillars and the placement of signal machines consistent with the real line), pipeline routing (cable laying paths and fiber optic connection topology completely restored), and electrical connection relationships (precise mapping of power supply circuits and signal control links). The geometric dimensional deviation between the virtual model and the physical equipment is ≤0.1mm. Meanwhile, this unit integrates teaching process data, student learning trajectory data, and real industry fault data into the model, giving the twin a teaching attribute: such as dynamically updating the model's knowledge point annotations according to the student's learning progress (hiding detailed parameters for unlearned knowledge points and displaying extended content for mastered knowledge points), and setting the model's fault triggering parameters based on real fault cases (such as the resistance threshold for contact network grounding faults and the voltage fluctuation range for signal faults). The standard integration unit employs natural language processing algorithms to perform structured analysis on 15 core industry standard texts, including the "Railway Electric Traction Power Supply Construction Specification" and the "High-Speed ​​Railway Signal Engineering Construction Quality Acceptance Standard." It precisely extracts equipment parameter standards (such as allowable deviation of catenary conductor height ±30mm, signal cable laying bending radius ≥15 times cable diameter), construction process requirements (such as power supply equipment installation verticality deviation ≤0.5‰, interlocking system debugging process "power-off inspection - parameter configuration - linkage test"), and fault handling guidelines (such as the "power-off isolation - fault location - parameter retest - power-on reset" process after traction power supply tripping). These standard contents are then transformed into structured rules recognizable by the digital twin (classified and coded according to "parameter constraints - process constraints - handling constraints"). Through model parameter binding (writing equipment parameter standards into geometric model attribute fields) and process node embedding (writing construction process requirements into operation simulation logic), the digital twin is deeply embedded, enabling the virtual model to possess standard verification capabilities.

[0062] AI intelligent analysis engine algorithm implementation The AI-powered intelligent analysis engine leverages machine learning algorithms to provide a closed-loop service encompassing student ability analysis, personalized learning path generation, and intelligent Q&A. The ability profile construction unit employs an algorithm model combining decision trees and multilayer perceptron (MLP) neural networks: the decision tree algorithm uses the Gini coefficient (with a threshold of ≤0.3) to filter key features for ability assessment (such as training error types and the number of times knowledge points are repeatedly studied), determining the core assessment indicators across three dimensions: "knowledge point mastery," "practical skills," and "self-directed learning." The MLP neural network comprises an input layer (20 neurons, corresponding to 20 assessment features), two hidden layers (128 neurons per layer, using ReLU activation function), and an output layer (6 neurons, corresponding to 6 ability levels). After training with 80,000 sets of student learning data (including samples from students of different grades and backgrounds), the model's prediction accuracy reached 94%. This unit analyzes students' learning trajectories and practical training data to quantify their mastery of knowledge points (e.g., 85% mastery of overhead contact system knowledge points and 72% mastery of signal system knowledge points), weaknesses in practical skills (e.g., weak fault diagnosis capabilities and unfamiliarity with equipment disassembly and assembly procedures), and self-learning ability scores (quantified on a scale of 0-100), constructing a multi-dimensional student competency profile. The learning path generation unit, based on these competency profiles and industry job requirements (referencing skill requirements for positions such as high-speed rail maintenance and signal debugging), establishes a weighted fusion model (practical skills weight 0.6, theoretical knowledge weight 0.3, and self-learning ability weight 0.1), dynamically generating personalized learning paths: for students with weak knowledge points, corresponding theoretical lectures and online exercises are provided (precisely matched according to the type of incorrect answers); for students with insufficient practical skills, targeted practical training projects are configured (e.g., basic disassembly and assembly of overhead contact systems and signal parameter debugging); for students with advanced skills, cross-disciplinary fault simulation exercises are arranged (e.g., comprehensive handling of overhead contact system grounding linkage signal system failures). The intelligent Q&A unit constructs a railway electrical system knowledge base containing over 50,000 professional terms (covering theoretical knowledge, practical guidelines, and regulatory clauses). It uses the TF-IDF algorithm to extract keywords from student questions (keyword extraction accuracy ≥ 92%) and combines a scaled dot product attention mechanism (attention coefficient dynamically adjusted via a softmax function) to calculate the semantic similarity between the question and the knowledge base answer (similarity threshold ≥ 0.85) to achieve accurate matching. When students raise practical questions such as "how to handle excessive overhead contact line conductivity", the system outputs a Q&A analysis within 1 second, including handling steps, regulatory basis, and operational precautions, and recommends relevant training video materials to help students quickly understand.

[0063] Application and Implementation of Virtual and Real Teaching Interaction Module The virtual-real teaching interaction module creates immersive and personalized practical training scenarios through multi-unit collaboration. The personalized practical training scenario building unit dynamically generates suitable training scenarios based on students' ability profiles and learning paths: for students with weak foundations, low-complexity scenarios such as "disassembly and assembly of a single component of the overhead contact line" and "basic parameter settings for signal controllers" are generated (number of devices ≤ 5, no fault triggering); for intermediate students, medium-complexity scenarios such as "power supply cabinet debugging + signal controller linkage testing" are generated (includes 1-2 single-discipline faults); for advanced students, complex cross-discipline scenarios such as "overhead contact line grounding fault + signal interlocking failure + communication link interruption" are generated (number of devices linked ≥ 10, faults triggered by multiple faults). After scenario generation, the scenarios are loaded into the VR immersive training unit through a virtual simulation engine (supporting real-time rendering, frame rate ≥ 60fps, rendering accuracy ≤ 0.1mm). The VR immersive training unit uses VR headsets (resolution 3840×2160, field of view 110°, refresh rate 90Hz) to provide students with an immersive environment for assembling and disassembling equipment. Through data gloves (finger motion tracking accuracy ≤0.5mm, tactile feedback pressure range 0-5N) driving virtual hand models, students can complete practical training such as assembling contact wire supports, wiring signal cables, and debugging power supply parameters. The weight and operating resistance of the equipment in the virtual environment are consistent with the real equipment (such as the torque feedback of tightening bolts simulating the real feel). The motion recognition and error correction unit deploys a KinectV4 motion capture device (≥25 skeletal keypoints recognized, tracking frame rate 30fps), and captures students' training movements in real time through a skeletal keypoint recognition algorithm: first, heatmaps of key points such as hands and arms are generated using a Gaussian heatmap regression formula (Gaussian standard deviation σ=2), then discrete keypoints are connected into complete motion postures using a partial affinity domain (PAF) formula, and finally, the Euclidean distance error between the student's movement and the standard movement is calculated (error threshold ≤5mm); when students perform non-standard operations such as exceeding the verticality standard of the contact wire support installation or insufficient bending radius of cable laying, the system immediately displays a three-dimensional visual error correction prompt through the VR headset (the standard movement trajectory is marked with a red dotted line, and the error reason explanation pops up simultaneously). The AR remote guidance unit uses AR smart glasses (display resolution 1920×1080, field of view 85°) to provide real-time guidance based on action correction results and learning progress: voice guidance (voice recognition accuracy ≥95%, supporting natural language interaction such as "What is the next operation?" "Is this parameter set wrong?"), operation step prompts (dynamic flowcharts overlaid in the AR field of view), and virtual annotations (highlighting key operation points with highlighted icons). At the same time, it records the student's operation accuracy (calculated as: number of correct operations / total number of operations × 100%), completion time, error frequency, and other training data, and transmits them back to the system backend in real time.

[0064] Cross-disciplinary fault simulation module operation mechanism The cross-disciplinary fault simulation module, based on student training data, enables accurate simulation and deduction of cross-disciplinary faults in the four electrical systems. The fault correlation analysis unit employs an improved Apriori association rule mining algorithm (minimum support 0.05, minimum confidence 0.8) to analyze fault types, operational behaviors, equipment statuses, and other data in the student training data. It identifies the coupling relationships between faults in different disciplines, such as the correlation between "contact network grounding fault" and "sudden drop in signal system power supply voltage," and the triggering logic between "substation tripping" and "communication dispatch link interruption." A cross-disciplinary fault correlation matrix is ​​constructed (matrix dimension 4×4, covering the pairwise correlation strength of faults in the four major disciplines). Based on this correlation matrix, the linkage simulation unit constructs a fault propagation logic model to dynamically simulate the propagation path and impact range of cross-disciplinary faults. Taking "interlocking failure of contact wire grounding fault linkage signal" as an example, the system first simulates the occurrence of contact wire grounding fault (triggering condition: distance between contact wire and grounding body ≤ 0.5m). Through circuit simulation algorithm (simulation step size ≤ 1ms), it calculates the voltage change of the power supply circuit (voltage drops from 27.5kV to ≤ 10kV), which in turn triggers unstable power supply to the signal system, causing abnormal signal display (such as red light changing from solid to flashing), and finally causing interlocking logic disorder (train routes cannot be arranged normally). Throughout the process, students can observe the changes in the operating status of various professional equipment through virtual terminals (such as voltage meter value fluctuations and signal indicator light status switching). After the fault simulation is completed, the simulation output unit generates a multi-dimensional simulation report: a timeline document of the entire fault evolution process (marking the fault status and impact range of each stage according to the timeline), a list of key handling nodes (clearly defining core nodes such as "complete power outage isolation of the fault section within 3 minutes" and "switch to backup signal power supply within 5 minutes"), and a rationality assessment of the student's handling plan (quantitative scoring dimensions include handling efficiency, process compliance, and risk control level). The report is exported in PDF format, supporting teachers to review and students to debrief.

[0065] Implementation process of teaching effectiveness evaluation module The teaching effectiveness evaluation module achieves quantitative evaluation of teaching effectiveness through a three-tiered framework that integrates process-based assessment, practical skills assessment, and evaluation. The process-based assessment unit establishes a multi-dimensional quantitative indicator system based on the full-process data collected by the teaching data acquisition module: the mastery of knowledge points is quantified by the mastery rate of knowledge points (number of correct answers / total number of questions × 100%) and the repetition rate of wrong questions (number of times the same wrong question appears repeatedly / total number of wrong questions × 100%); learning attitude is assessed by classroom interaction activity (number of interactions / total classroom time × 100%), practical training attendance rate (actual number of practical training sessions / number of practical training sessions that should be attended × 100%), and on-time completion rate of tasks (number of tasks completed on time / total number of tasks × 100%); self-learning ability is quantified by self-learning time (percentage of time exceeding the prescribed learning time), frequency of access to extended knowledge points, and self-solving rate of difficult problems (number of self-solved problems / total number of problems × 100%). Each indicator is weighted (mastery of knowledge points 0.4, learning attitude 0.3, self-learning ability 0.3) to calculate the total score of the process-based assessment (out of 100). The practical skills assessment unit evaluates three core skills based on cross-disciplinary fault simulation results and practical training data: equipment operation proficiency (standards are ≥90% accuracy in operation steps, ≤120% of standard time in completion time, and ≤2 errors), fault diagnosis accuracy (number of times fault type correctly diagnosed / total number of fault simulations × 100%), and cross-disciplinary collaborative handling ability (scored based on the rationality of task division, collaborative response speed, and effectiveness of joint handling plan). Each skill is assigned a weight according to job requirements (equipment operation proficiency 0.3, fault diagnosis accuracy 0.4, cross-disciplinary collaborative handling ability 0.3), resulting in a total practical skills assessment score (out of 100). The evaluation integration unit employs a weighted integration model, combining the total score of process-based assessment (weight 0.3) with the total score of practical skills assessment (weight 0.7) to generate three types of core quantitative teaching effectiveness data: skill attainment rate (the percentage of students who have achieved the required scores in six core skills, including knowledge point mastery, equipment operation, and fault handling); ability improvement rate (calculating the percentage improvement in each dimension based on the initial assessment upon enrollment, such as an improvement of 41.7% in fault diagnosis accuracy from 60% to 85%); and job suitability (matching the assessment results with the ability requirements of target positions such as overhead contact line maintenance and signal system debugging, quantifying the degree of fit, with a fit rate ≥80% considered highly suitable). The final quantitative evaluation report provides objective and accurate decision-making basis for teachers to optimize teaching plans (such as increasing fault simulation training hours for classes with weak fault diagnosis capabilities), adjust the allocation of practical training resources (such as adding interlocking system training terminals for signal majors), and for the college to improve its talent training system, effectively enhancing the relevance and practicality of talent training in the four electrical systems major.

[0066] In summary, this application's embodiments utilize a teaching data acquisition module to collect comprehensive teaching, learning, and real-world fault data, providing precise data input for system operation; a virtual simulation modeling module constructs a high-precision digital twin and deeply integrates industry standards, both recreating the real four-electric system scenario and significantly reducing the cost of physical equipment and wear and tear; an AI intelligent analysis engine relies on machine learning algorithms to construct precise competency profiles, generate personalized learning paths, and provide real-time intelligent Q&A, efficiently addressing students' knowledge gaps and skill deficiencies; and a virtual-real teaching interaction module creates an immersive training scenario, using real-time action recognition error correction and AR remote... The course provides guidance to mitigate practical risks and enhance the standardization of training; the cross-disciplinary fault simulation module accurately reproduces complex fault transmission and coordinated handling scenarios, honing students' systemic thinking and cross-disciplinary collaborative abilities; the teaching effectiveness evaluation module establishes a quantitative evaluation system that combines process-oriented and practical skills, providing objective decision-making basis for teaching optimization. Ultimately, it effectively addresses the pain points of traditional teaching, such as high equipment costs, difficulty in reproducing fault scenarios, high practical risks, and lack of quantitative evaluation, achieving deep integration of theory and practice, improving the relevance, practicality, and efficiency of training professionals in the four electrical systems, and helping to cultivate compound technical talents that meet industry needs.

[0067] Next, referring to the accompanying drawings, an AI-based intelligent teaching method for the four electrical systems of a high-speed railway is described according to an embodiment of this application.

[0068] like Figure 9 As shown, this AI-based intelligent teaching method for the four electrical systems of high-speed railway includes the following steps: In step S101, teaching process data, student learning trajectory and real industry fault data are acquired.

[0069] It is understood that, by acquiring teaching process data, student learning trajectories, and real-world industry fault data, the intelligent teaching system for the high-speed railway electrical system can dynamically synchronize the entire teaching implementation process, student learning progress, and real-world industry fault scenarios. It can capture in real time deviations in teaching implementation, students' core competency gaps, and the coupling evolution of industry faults. This provides high-quality data for subsequent BIM model integration to construct digital twins, machine learning algorithm modeling of student competency profiles, and intelligent generation of personalized learning paths. It enables precise optimization of teaching implementation and targeted guidance for student learning, reducing the disconnect between teaching and industry needs and the mismatch between learning paths and student weaknesses. Ultimately, it improves the intelligence level of high-speed railway electrical system teaching and the effectiveness of professional talent training.

[0070] In step S102, based on teaching process data, student learning trajectory and real industry fault data, and by integrating BIM model and railway technical specification text data, a digital twin of the high-speed railway's four electrical systems is constructed. A student competency profile is built through machine learning algorithms, and personalized learning paths are generated and intelligent Q&A is provided based on the student competency profile.

[0071] Among them, the BIM model is the digital carrier of building information model. It is a parametric three-dimensional model based on three-dimensional geometry, integrating all information such as geometry, physics, function and technology throughout the entire life cycle of a project, and supporting collaborative operations and information management at all stages such as design, construction and operation and maintenance.

[0072] It is understood that the embodiments of this application integrate teaching process data, student learning trajectories, real industry fault data, and railway technical specification text data, and associate them with the full information of the BIM model. Relying on the characteristics of three-dimensional geometric morphology and the integration of full information throughout the entire life cycle of the project, it accurately restores the physical form, functional characteristics, and technological standards of the high-speed railway's four electrical systems. This provides core model and information support for the accurate construction of the digital twin of the high-speed railway's four electrical systems, the modeling of student ability profiles using machine learning algorithms, the intelligent generation of personalized learning paths, and the professional answers from intelligent Q&A units. It improves the accuracy of model and data fusion, comprehensively integrates the related information of teaching implementation, student learning, and industry practice, accurately matches the practical needs of high-speed railway four electrical systems positions, and provides three-dimensional model support for the construction of digital twin teaching scenarios, accurate assessment of student abilities, personalized teaching, and professional intelligent Q&A. It optimizes teaching implementation strategies and resource allocation, and enhances the scenario-based, professional, and intelligent efficiency of talent training in the high-speed railway four electrical systems.

[0073] For example, in the digital twin teaching scenario of the four electrical systems of high-speed railways, BIM models can accurately restore the three-dimensional geometric shape, physical attributes, process assembly standards, and spatial layout relationships of core facilities and surrounding supporting projects of the four electrical systems, such as catenary, signal machines, power supply cabinets, and traction substations. This allows for a full-dimensional, concrete presentation of details such as equipment structure, pipeline routing, and interface matching. Furthermore, it can accurately associate real-world single-discipline faults and cross-discipline coupled fault data (such as catenary grounding linkage signal interlock failure, power supply system faults causing communication interruptions, etc.) with the corresponding facility locations in the model, simulating equipment status changes, fault propagation paths, and the entire evolution process when a fault occurs, intuitively demonstrating the chain reaction caused by the fault. Simultaneously, the practical operation trajectories of students in equipment disassembly and assembly, fault diagnosis, and cross-discipline collaborative handling can be marked in real time on the BIM model. The 3D model clearly presents students' weaknesses in areas such as equipment structure understanding, operational procedures, fault location logic, and collaborative work, facilitating teachers' accurate assessment of students' skill gaps. Furthermore, students can utilize the model for immersive, risk-free equipment operation simulations, fault diagnosis, and troubleshooting training. They can repeatedly practice equipment disassembly and assembly steps, tool usage methods, fault diagnosis procedures, and key points of cross-disciplinary collaboration in a 3D environment, improving their practical skills and fault handling abilities. The intelligent Q&A unit, using the BIM model as a 3D visualization support, provides visual and step-by-step answers to students' professional questions regarding equipment structure disassembly, fault cause analysis, troubleshooting procedure formulation, and process standard execution, making abstract professional knowledge and practical points intuitive and easy to understand, helping students quickly grasp and master them.

[0074] In step S103, based on student ability profiles and personalized learning paths, a personalized digital training scenario is built to conduct VR immersive equipment disassembly and assembly training. Real-time action recognition is used for 3D visualization error correction and AR remote dynamic teaching guidance. Student training data is output, and cross-professional fault linkage simulation and dynamic deduction are carried out through association rule mining algorithms, and fault deduction results are output.

[0075] Among them, association rule mining algorithm is a classic algorithm in the field of data mining. It refers to the technical method of mining the implicit association and coupling rules between different things from massive structured data, and quantifying the effectiveness of the rules through indicators such as support, confidence, and lift.

[0076] It is understood that this application's embodiments, through processing and analyzing student training data, fault simulation results, and real fault data from the high-speed railway's electrical systems, uncover the implicit correlations and coupling evolution patterns between faults of different specialties. This transforms scattered fault data into a quantifiable and predictable fault association rule system, providing core algorithmic support and data basis for the accurate implementation of cross-professional fault linkage simulation and dynamic simulation. This enhances the realism and industry relevance of fault simulation, accurately capturing the transmission paths, linkage logic, and evolution trends of cross-professional faults. Students directly face real-world fault scenarios to effectively hone their ability to identify, analyze, and handle cross-professional coupled faults. Simultaneously, it provides data support for optimizing and adjusting personalized digital training scenarios and precisely setting training priorities, assisting teachers in providing targeted fault handling instruction, reducing students' blind spots in understanding cross-professional faults, and comprehensively improving the intelligence level, professional adaptability, and teaching effectiveness of digital training in the high-speed railway's electrical systems.

[0077] For example, in personalized digital training scenarios for the four electrical systems of high-speed railways, association rule mining algorithms can be used to deeply analyze massive amounts of student training fault data, real-world industry fault case data, and operational correlation data of the four electrical systems. By quantifying the effectiveness of rules through indicators such as support, confidence, and lift, the algorithms can accurately uncover implicit correlations and coupling evolution patterns of a series of cross-disciplinary faults, such as the tendency for "traction power supply contact network grounding" to couple to "signal system interlocking failure" and "track circuit short circuit," the tendency for "power supply cabinet relay protection device malfunction" to link to "communication base station power outage" and "signal display abnormality," and the tendency for "turnout switch machine jamming" to trigger "section signal blockage fault." Furthermore, the algorithms can also uncover the tendency for "contact network disassembly and wiring errors" during student training to trigger subsequent... Continuing the correlation between operational errors and fault handling deviations, such as "failure to troubleshoot grounding faults" and "deviation in signal controller parameter adjustment" leading to "failure to verify interlocking relationships," these discovered correlation rules can be used to build highly realistic cross-professional fault linkage simulation scenarios in VR immersive equipment disassembly and fault handling training. This allows students to immerse themselves in the entire process of a single fault triggering a chain reaction of problems across multiple professions. Targeted fault handling training tasks can also be designed, enabling students to simultaneously master the troubleshooting and handling methods for both primary and related faults. Furthermore, teachers can use these rules to guide students in mastering the key points of handling related professional faults, addressing their weaknesses in handling faults within a specific profession, thus preventing students from focusing solely on single-professional faults while neglecting chain reactions during training.

[0078] In step S104, based on the fault simulation results and trainee training data, a dual evaluation system combining process assessment and practical skills assessment is established, and teaching effectiveness data is quantitatively output.

[0079] Among them, the binary evaluation system is a systematic evaluation system built around a specific evaluation goal, from two core dimensions (such as process and result, theory and practice), setting corresponding evaluation indicators, standards and weights for each dimension, carrying out targeted quantitative or qualitative evaluation and making comprehensive judgments.

[0080] It is understood that this application embodiment systematically sorts out and quantifies the fault simulation results and student training data. Relying on a dual evaluation system that combines process assessment and practical skills assessment, it integrates the evaluation content of the two core dimensions of students' learning and growth process in the high-speed railway electrical system and practical training results. It sets exclusive evaluation indicators, quantitative standards, and scientific weights for each dimension to meet the needs of industry positions. It conducts targeted and refined assessment and evaluation and makes comprehensive judgments, providing systematic and standardized evaluation support for the accurate quantitative output of teaching effectiveness data. This improves the comprehensiveness, objectivity, and professional adaptability of teaching effectiveness evaluation, effectively avoids the one-sidedness of evaluation caused by a single assessment dimension, accurately captures students' strengths and weaknesses in theoretical learning, self-learning, equipment operation, fault handling, etc., and provides scientific evaluation basis for teachers to optimize teaching plans, adjust training focus, and conduct targeted teaching guidance. At the same time, it makes the teaching effectiveness evaluation more in line with the skill requirements of the high-speed railway electrical system industry positions, helps colleges and universities accurately control the quality of talent training, and improves the adaptability of professional talent training in the high-speed railway electrical field to the needs of industry positions.

[0081] For example, in the evaluation scenario of practical training in the four electrical systems of high-speed railways, a dual evaluation system combining process-based assessment and practical skills assessment can be used to conduct a comprehensive evaluation based on the results of fault simulation and all training data. The process-based assessment dimension quantifies and analyzes the entire process data, including the execution trajectory of equipment disassembly and assembly, the evolution of fault diagnosis strategies, the interactive process of cross-disciplinary collaboration, and the rectification of operational errors. It also refines assessment indicators such as knowledge mastery rate, compliance of operating procedures, and timeliness of problem rectification. The practical skills assessment dimension focuses on the final handling effect of fault simulation, quantifying and evaluating core outcome indicators such as equipment operation proficiency, fault diagnosis accuracy, effectiveness of cross-disciplinary fault linkage handling, and quality of training task completion. This provides a scientific basis for both dimensions. The evaluation weights are set to align with the job requirements of the high-speed rail electrical engineering industry, such as 40% for process-based assessment and 60% for practical skills assessment. By comprehensively evaluating the results of the two dimensions, it is possible to accurately identify different types of student skill gaps, such as "standardized training procedures but low fault handling efficiency" and "acceptable final fault simulation results but omissions in the operation process." It is also possible to comprehensively and objectively quantify and output teaching effectiveness data. Based on the evaluation results, teachers can focus on providing standardized practical steps guidance to students with non-standardized process operations and design targeted fault simulation acceleration training for students with low handling efficiency, making teaching guidance more targeted. At the same time, it is possible to accurately control the teaching effectiveness from both process and result dimensions, providing an objective basis for optimizing teaching plans and adjusting training priorities.

[0082] According to the embodiments of this application, an AI-based intelligent teaching method for the four electrical systems of high-speed railways is proposed. This method collects multi-source teaching and fault data through a teaching data acquisition module, and constructs a digital twin of the high-speed railway's four electrical systems using a virtual simulation modeling module, providing high-precision, scenario-based digital support for teaching. A personalized digital training scenario is built using a virtual-real teaching interaction module, and a cross-professional fault simulation module enables cross-professional fault linkage simulation, effectively improving the adaptability of training content to complex on-site conditions. An AI intelligent analysis engine constructs student competency profiles and generates personalized learning paths. Combined with a binary quantitative assessment system in the teaching effectiveness evaluation module, learning weaknesses are accurately identified and optimized. This method solves the problems of traditional teaching scenarios being singular and lacking cross-professional fault drills, while improving the relevance of teaching and the efficiency of skill transformation. It can significantly shorten the training cycle for maintenance personnel, enhance their ability to handle complex cross-professional faults, and provide a solid talent guarantee for the safe and stable operation of the high-speed railway's four electrical systems. Therefore, it solves the problems of rigid teaching strategies and low accuracy in existing technologies.

[0083] The following will illustrate an AI-based intelligent teaching method for the four electrical systems of high-speed railways through a specific embodiment, such as... Figure 10 As shown, it includes: This study selects a vocational high-speed railway electrical system training center as the application scenario, focusing on four core professional areas: power supply, signaling, communication, and traction power supply. Using overhead contact lines (cantilever type), signal lights (high-mast color lights), traction substations (27.5kV GIS cabinets), and communication base stations (fiber optic repeaters) as core training equipment, an AI-based intelligent teaching system is constructed. This system enables intelligent teaching throughout the entire process, from data acquisition and twin modeling to training guidance and assessment. The data acquisition process employs a three-dimensional acquisition scheme combining hardware sensing, software integration, and manual supplementation to ensure comprehensive coverage and accurate usability of multi-source data. Teaching process data is collected synchronously through the training platform and classroom interaction system: a smart blackboard (Honghe HD-I7980D) is deployed to record in real time the duration of knowledge point explanations in theoretical lectures, the accuracy rate of classroom question and answer (30 questions are sampled per class, and an accuracy rate of ≥80% is considered passing), the frequency of key points marked in courseware, and other data; the training platform (Railway Electrical Training Simulation System V3.0) has a built-in data acquisition module to capture in real time students' trigger commands for equipment simulation operations, parameter adjustment records, and process jump trajectories, with a sampling frequency of 10Hz to reproduce operation details. The student learning trajectory relies on the deep integration of the campus LMS (Learning Management System) and the training terminal: through the association of student accounts, the system collects the online course learning time (≥40 hours for a single course), the quality of homework completion (graded according to the accuracy of knowledge points, with 85 points or above being excellent), and the frequency and duration of access to self-expanded learning resources (such as railway industry standards and fault cases); the training terminal (industrial-grade tablet, IP65 protection level) has a built-in positioning module (GPS + Beidou dual-mode, positioning accuracy ±3m) to record the student's movement trajectory in the training site, the time spent operating equipment, and to identify the proportion of time spent on self-practice and passive guidance. The real-world fault data was obtained through cooperation with the operation and maintenance departments of railway bureau groups. More than 200 typical fault cases of the high-speed rail electrical system over the past 5 years were selected, covering single-discipline faults (such as wear of contact wires and signal light failure) and cross-discipline coupled faults (such as failure of traction power supply grounding linkage signal interlocking and failure of power supply monitoring due to interruption of communication optical cable). Each fault data includes full-element information such as fault cause, handling process, equipment parameter change curve, and industry standard basis, and was imported into the system after being de-identified.

[0084] Data transmission and storage adopt an "edge aggregation + cloud collaboration" architecture: Five industrial Ethernet gateways (Huawei AR502H) are deployed in the training site, connecting to equipment sensors and training terminals via the Profinet protocol. Data is initially filtered for redundant information by an edge computing node (Intel Core i7-12700K) (removing duplicate data collection and retaining ≥95% of valid data), and then transmitted to the local server via a 5G industrial module (Huawei MH5000-31), with transmission latency controlled within 30ms. At the storage level, a hybrid database architecture is used: InfluxDB time-series database stores high-frequency training operation data (retained for 90 days, supporting fast queries by device and time dimension); MySQL database stores structured data such as teaching process data, student files, and industry fault cases; MongoDB stores unstructured data (such as student practical videos and fault handling documents). Real-time data exchange between multiple databases is achieved through a data synchronization tool (DataStage), building a complete data traceability system.

[0085] The digital twin was constructed using a "BIM modeling + multi-source data fusion + physical simulation" approach to ensure scene fidelity and teaching suitability. In the BIM modeling phase, Revit 2024 and Navisworks 2024 software were used to build a 1:1 scale 3D model of the four electrical systems, including over 300 core components (overhead catenary arms, signal lights, GIS cabinet circuit breakers, fiber optic distribution frames, etc.). The model accurately reproduced the equipment's geometric dimensions (overhead catenary arm length 2.8m, GIS cabinet dimensions 800mm×1500mm×2200mm), physical properties (material, weight, pressure rating), process assembly standards (bolt tightening torque, pipeline routing), and spatial layout relationships (overhead catenary and track spacing, equipment foundation pre-embedded locations). Model optimization employed a balanced strategy of topology simplification and detail preservation, removing non-critical features (such as bolts with a diameter <3mm) and reducing the model's polygon count from 8 million to 2 million, while ensuring a geometric accuracy of ±0.5mm for critical mating parts (such as the overhead catenary pantograph contact points and signal terminal blocks). By integrating railway technical specification text data (such as the "Railway Signal Equipment Installation Standard" and the "Railway Electric Traction Power Supply Design Specification"), the specification clauses are associated with model components. Clicking on a component will display a pop-up window showing the corresponding installation requirements, operational prohibitions, and fault handling specifications. Physical simulation is achieved collaboratively through ANSYS and Unity3D: ANSYS establishes a catenary pantograph force model and a GIS cabinet arc simulation model to simulate the mechanical changes and electrical characteristics of the equipment under fault conditions (such as the current distribution when the catenary is grounded and the voltage changes when the signal is short-circuited); Unity3D loads the BIM model and integrates simulation data through the FMU interface to achieve real-time visualization of the fault condition (such as displaying a red current halo at the catenary grounding point and displaying a temperature gradient effect on the faulty component of the GIS cabinet), with an update frequency of 25Hz to ensure smoothness.

[0086] The student competency profile was constructed using a fusion model of random forest and collaborative filtering algorithms. After preprocessing, the input data was used to build a 42-dimensional feature vector (12 dimensions of theoretical knowledge, 18 dimensions of practical skills, and 12 dimensions of learning attitude). Theoretical knowledge features included knowledge point mastery rate (e.g., accuracy rate of traction power supply principle knowledge) and familiarity with industry standards (accuracy rate in answering questions about standard clauses). Practical skills features covered the accuracy rate of equipment operation procedures, fault location time, and compliance rate of tool use. Learning attitude features included the percentage of self-study time, frequency of asking questions, and timeliness of error rectification. The model was trained using the Python TensorFlow framework, based on historical teaching data from 1000 students (70% training set, 20% validation set, and 10% test set). After 80 epochs, the model achieved an accuracy rate of 96.8%, accurately outputting student competency levels (excellent, good, satisfactory, need improvement) and weaknesses (e.g., weak handling of overhead contact line faults, insufficient cross-disciplinary collaboration ability). Personalized learning path generation employs an adaptive recommendation algorithm to customize differentiated solutions for students with varying ability profiles: for students with weak theoretical knowledge but strong practical skills, online theoretical courses plus specialized exercises on regulatory clauses are recommended; for students with weak practical skills, VR immersive training plus one-on-one practical guidance is prioritized; for students lacking cross-disciplinary collaboration, cross-disciplinary fault handling team tasks are assigned. Intelligent Q&A relies on a BERT pre-trained model to build a question-and-answer knowledge base covering equipment operation, fault handling, industry standards, and other content. Students can ask questions via voice (with a recognition accuracy of over 95%) or text, and the system can provide visual answers using a BIM model. For example, if the question is "steps for handling contact network grounding faults," the system simultaneously demonstrates the entire process of grounding location, power outage operation, and fault troubleshooting in a virtual twin scenario, highlighting key precautions.

[0087] Personalized digital training scenarios are built based on Unity3D 2023.1 and are compatible with the PicoNeo4VR headset (2160×2160 resolution / eye, 90Hz refresh rate) to achieve immersive training in equipment disassembly, assembly, and troubleshooting. The VR immersive equipment disassembly and assembly training constructs interactive scenarios for core equipment: In the catenary arm disassembly and assembly scenario, students use VR controllers to simulate the operation of tools such as wrenches and screwdrivers. The system captures hand movements through a skeletal key point recognition algorithm (94% accuracy) and judges the standardization of operations in real time (such as whether the bolt tightening torque is up to standard and whether the wiring sequence is correct). It provides 3D visual correction for incorrect actions (such as reverse bolt disassembly or reversed terminal connections)—displaying red warning marks at the incorrect locations, synchronously playing standard operation animations (slow-motion demonstration of the correct steps), and alerting students through controller vibration feedback (vibration intensity correlated with the severity of the error). AR remote dynamic teaching guidance uses the Microsoft HoloLens 2 headset, which supports teachers to remotely access students' training scenarios. Teachers can intervene in the training process through gesture marking (such as pointing an arrow to the faulty part) and voice guidance (real-time voice latency <10ms). Standardized operation prompts can be pushed for common problems (such as pop-up tool usage instructions), and one-on-one action correction can be performed for individual problems.

[0088] Trainee training data is output to the system in real time, including equipment operation step records (time spent on each step, correct / incorrect markings), fault handling procedures (location time, diagnostic basis, and handling measures), tool usage records (tool type, usage duration, and compliance counts), and cross-disciplinary collaborative interaction logs (communication content, task allocation, and coordination efficiency). Cross-disciplinary fault linkage simulation and dynamic extrapolation employ the Apriori association rule mining algorithm. Based on trainee training data and industry fault data, association rules are mined, setting a support threshold of 15% and a confidence threshold of 70%, accurately identifying cross-disciplinary fault association patterns such as "excessive wear of contact wire → pantograph arcing → signal malfunction" and "GIS cabinet circuit breaker failure → traction power supply interruption → communication base station backup power activation." The simulation process is visualized in a digital twin scenario: taking the "failure of the contact network grounding linkage signal interlocking" fault as an example, the system simulates the entire process from the occurrence of contact network grounding, to the abnormal current transmission to the signal system, and then to the failure of the interlocking device. It displays the changes in parameters of each device in real time (sudden drop in contact network voltage, flashing of signal lights, and changes in interlocking status codes). Students can try fault location and cross-disciplinary collaborative handling in the scenario (power supply professionals cut off power, signal professionals investigate interlocking faults, and communication professionals ensure information transmission). The system records the simulation process data and results.

[0089] The dual evaluation system is constructed based on fault simulation results and trainee training data. It adopts a weighted allocation that combines process-based assessment (40%) and practical skills assessment (60%), aligning with the skill requirements of the high-speed rail electrical engineering industry. The process-based assessment dimension is refined into four core indicators: knowledge point mastery rate (weight 10%, calculated based on theoretical test and standardized answer scores), compliance rate of practical operation steps (weight 12%, statistically analyzing the percentage of correct steps in equipment disassembly and assembly and fault handling), timeliness of error rectification (weight 8%, assessing the time spent and effectiveness of rectification after an error occurs), and cross-professional collaborative participation (weight 10%, scored based on the frequency of communication and the rationality of division of labor in group tasks). The practical skills assessment dimension focuses on three core indicators: equipment operation proficiency (weight 20%, comprehensively evaluated based on operation time, tool usage accuracy, and step accuracy), fault diagnosis and handling ability (weight 25%, assessing the time spent on fault location, the scientific nature of the diagnostic basis, and the conformity of the handling plan with industry standards), and effectiveness of cross-professional fault simulation (weight 15%, scored based on the quality of simulation completion and the accuracy of related fault identification). Each indicator has a quantitative standard. For example, a proficiency level of ≥85 points is excellent, 70-84 points is good, 60-69 points is qualified, and <60 points is in need of improvement.

[0090] The quantitative output of teaching effectiveness data includes three core data categories: skill attainment rate (statistically calculating the percentage of students who meet each indicator, such as the attainment rate of handling overhead contact line faults and the attainment rate of cross-disciplinary collaboration), ability improvement rate (calculating the percentage improvement in theoretical knowledge and practical skills based on the initial assessment upon enrollment; for example, if a student's practical skills score improves from 62 to 81, the improvement rate is 30.6%), and job suitability (matching the assessment results with the ability requirements of target positions such as overhead contact line maintenance and signal system debugging, quantifying the degree of fit; a fit rate of ≥80% is considered highly suitable). The system automatically generates a teaching effectiveness report for each student, including a competency profile, weakness analysis, improvement suggestions, and detailed practical training data. It also outputs overall class teaching effectiveness statistics (average attainment rate, distribution of weak points), providing teachers with a basis for optimizing teaching plans—for example, if it is found that the class's cross-disciplinary collaboration ability is generally weak, the number of cross-disciplinary fault handling practical training hours can be increased; for weaknesses in overhead contact line practical skills, the intensity and frequency of VR practical training scenarios can be adjusted.

[0091] The system performance was simultaneously optimized during the teaching process: by continuously collecting student training data and teaching feedback, the threshold parameters of the association rule mining algorithm were iteratively updated (e.g., adjusting the confidence threshold of cross-professional fault association rules to 75% to improve rule accuracy); BIM model details were optimized, and 3D models and simulation data of newly added training equipment (such as new intelligent signal machines) were supplemented; the question-and-answer knowledge base was updated to include the latest industry fault cases and technical specifications, ensuring that teaching content is in sync with industry development. Through the application of this system, problems such as insufficient scenario-based learning, high practical risks, and biased evaluation in traditional high-speed rail electrical engineering teaching were effectively solved. Students' equipment operation proficiency increased by an average of 28%, and the accuracy rate of cross-professional fault handling increased by 35%. The adaptability of teaching effectiveness to industry job requirements was significantly improved, providing support for cultivating compound talents with solid practical skills and cross-professional collaborative qualities in the field of high-speed rail electrical engineering.

[0092] In summary, this application's embodiments effectively address the pain points of traditional high-speed rail electrical engineering teaching, such as insufficient scenario-based teaching, high operational risks, and biased evaluation, by collecting and storing comprehensive multi-source teaching data, learning trajectories, and industry fault data across multiple databases. This is achieved through BIM-integrated digital twin modeling, AI algorithm-based capability profiling and personalized recommendations, VR / AR immersive training guidance, and correlation rule-based fault deduction. Coupled with a binary evaluation system combining process-oriented and practical skills, these solutions precisely pinpoint students' skill gaps and weaknesses in classroom teaching. Personalized training paths and visualized Q&A guidance help students improve their equipment operation proficiency and cross-disciplinary fault handling capabilities, significantly reducing practical error rates. Dynamically optimized algorithm parameters and model details ensure that teaching content keeps pace with industry development, and quantitatively outputted teaching effectiveness data provides a scientific basis for scheme optimization. This comprehensively enhances the intelligence level, training effectiveness, and adaptability of high-speed rail electrical engineering teaching to industry job requirements, providing solid support for the delivery of interdisciplinary talents in the field.

[0093] Figure 11 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include: The memory 1101, the processor 1102, and the computer program stored on the memory 1101 and executable on the processor 1102.

[0094] When the processor 1102 executes the program, it implements an AI-based intelligent teaching method for the four electrical systems of a high-speed railway, as provided in the above embodiments.

[0095] Furthermore, electronic devices also include: Communication interface 1103 is used for communication between memory 1101 and processor 1102.

[0096] The memory 1101 is used to store computer programs that can run on the processor 1102.

[0097] The memory 1101 may include high-speed RAM (Random Access Memory) and may also include non-volatile memory, such as at least one disk storage.

[0098] If the memory 1101, processor 1102, and communication interface 1103 are implemented independently, then the communication interface 1103, memory 1101, and processor 1102 can be interconnected via a bus to complete communication between them. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 11 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0099] Optionally, in a specific implementation, if the memory 1101, processor 1102, and communication interface 1103 are integrated on a single chip, then the memory 1101, processor 1102, and communication interface 1103 can communicate with each other through an internal interface.

[0100] The processor 1102 may be a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of this application.

[0101] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described AI-based intelligent teaching method for the four electrical systems of a high-speed railway.

[0102] In the description of this specification, the references to "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0103] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0104] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0105] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any of the following techniques known in the art, or a combination thereof: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0106] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0107] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. An AI-based intelligent teaching system for high-speed railway four-electricity systems, characterized in that, include: The system includes modules for teaching data acquisition, virtual simulation modeling, AI intelligent analysis engine, virtual-real teaching interaction, cross-disciplinary fault simulation, and teaching effectiveness evaluation; among which, The teaching data acquisition module is used to collect teaching process data, student learning trajectories, and real-world fault data from the industry. The virtual simulation modeling module is used to construct a digital twin of the high-speed railway's four electrical systems based on the teaching process data, student learning trajectories, and real industry fault data, and by integrating BIM models and railway technical specification text data. The AI ​​intelligent analysis engine is used to analyze teaching data based on the digital twin, construct student ability profiles through machine learning algorithms, generate personalized learning paths, and provide intelligent Q&A. The virtual-real teaching interaction module is used to build personalized digital training scenarios based on the student's ability profile and the personalized learning path, conduct VR immersive equipment disassembly and assembly training, perform three-dimensional visualization error correction and AR remote dynamic teaching guidance through real-time action recognition, and output student training data. The cross-disciplinary fault simulation module is used to perform cross-disciplinary fault linkage simulation and dynamic inference based on the student training data and through association rule mining algorithm, and output fault inference results. The teaching effectiveness evaluation module is used to establish a binary evaluation system that combines process assessment and practical skills assessment based on the fault simulation results and the students' training data, and to quantitatively output teaching effectiveness data. 2.The AI-based intelligent teaching system for high-speed railway four-electricity system of claim 1, wherein The teaching data acquisition module includes a teaching process data collection unit, a learning trajectory tracking unit, and a fault data integration unit. The teaching process data collection unit is used to collect classroom interaction data, practical training operation behavior data, and theoretical assessment answer data in real time. The learning trajectory tracking unit is used to record students' learning time, knowledge point mastery progress, frequency of repeated practical training operations, and types of incorrect operations. The fault data integration unit is used to summarize real fault cases, fault diagnosis processes, and maintenance and handling data of railway power supply, signaling, communication, and traction power supply. 3.The AI-based intelligent teaching system for high-speed railway four-electricity system of claim 1, wherein The virtual simulation modeling module includes a twin construction unit and a specification fusion unit. The twin construction unit is used to recreate the equipment layout, pipeline routing, and electrical connection relationships of the high-speed railway's four electrical systems based on the BIM model, and integrates teaching process data, student learning trajectories, and real industry fault data to construct a high-precision digital twin. The specification fusion unit is used to extract equipment parameter standards, construction process requirements, and fault handling criteria from railway technical specifications and embed them into the constraints of the digital twin. 4.The AI-based intelligent teaching system for high-speed railway four-electricity system of claim 1, wherein, The AI ​​intelligent analysis engine includes a competency profile building unit, a learning path generation unit, and an intelligent Q&A unit. The competency profile building unit uses decision tree algorithms and neural network models to analyze students' learning trajectories and practical training data, quantifying weaknesses in knowledge points and shortcomings in practical skills to construct a student competency profile. The learning path generation unit dynamically generates personalized learning paths based on the student competency profile and industry job requirements, including knowledge point supplementation, targeted practical training, and fault simulation exercises. The intelligent Q&A unit, based on a railway professional knowledge base and natural language processing algorithms, responds to students' theoretical questions and practical difficulties, providing precise Q&A explanations and recommended reference materials. 5.The AI-based intelligent teaching system for high-speed railway four-electricity system of claim 1, wherein The virtual-real teaching interaction module includes a personalized training scenario building unit, a VR immersive training unit, a motion recognition and error correction unit, and an AR remote guidance unit. The personalized training scenario building unit generates suitable training scenarios for the disassembly, assembly, and maintenance of the four electrical systems based on student ability profiles and learning paths. The VR immersive training unit provides a realistic virtual disassembly and assembly environment for equipment, using virtual hand models for component disassembly and assembly training. The motion recognition unit captures student training movements in real time using a skeletal keypoint recognition algorithm, comparing them with standard operating procedures for 3D visualization and error correction. The AR remote guidance unit provides real-time voice guidance, operation step prompts, and virtual annotations based on error correction results and learning progress, recording student training data including operation accuracy, completion time, and error frequency. 6.The AI-based intelligent teaching system for high-speed railway four-electricity system of claim 1, wherein The cross-disciplinary fault simulation module includes a fault correlation analysis unit, a linkage simulation unit, and a simulation result output unit. The fault correlation analysis unit is used to identify the coupling relationship between faults in different disciplines based on trainee training data and through association rule mining algorithms. The linkage simulation unit is used to simulate the transmission path and impact range of cross-disciplinary faults, and to conduct dynamic simulation and emergency response simulation. The simulation result output unit is used to output the fault evolution process, key handling nodes, and fault simulation results for evaluating the rationality of the solution. 7.The AI-based intelligent teaching system for high-speed railway four-electricity system of claim 1, wherein The teaching effectiveness evaluation module includes a process assessment unit, a practical skills assessment unit, and an evaluation integration unit. The process assessment unit quantifies students' mastery of knowledge points, learning attitude, and self-learning ability based on their learning trajectory, classroom interaction data, and stage-by-stage practical training performance. The practical skills assessment unit evaluates equipment operation proficiency, fault diagnosis accuracy, and cross-disciplinary collaborative handling capabilities based on fault simulation results and practical training operation data. The evaluation integration unit integrates the results of the process assessment and practical skills assessment to generate quantitative teaching effectiveness data including skill attainment rate, ability improvement, and job suitability.

8. A method applied to the AI-based intelligent teaching system for high-speed railway four-electricity systems of any one of claims 1-7, characterized in that, The method includes: Acquire data on the teaching process, student learning trajectories, and real-world industry fault data; Based on the teaching process data, student learning trajectory and real industry fault data, and by integrating BIM model and railway technical specification text data, a digital twin of the high-speed railway's four electrical systems is constructed. A student competency profile is built through machine learning algorithms, and personalized learning paths are generated and intelligent Q&A is provided based on the student competency profile. Based on the student's ability profile and the personalized learning path, a personalized digital training scenario is built to conduct VR immersive equipment disassembly and assembly training. Real-time action recognition is used for 3D visualization error correction and AR remote dynamic teaching guidance. Student training data is output. Cross-professional fault linkage simulation and dynamic inference are conducted through association rule mining algorithm, and fault inference results are output. Based on the fault simulation results and the student training data, a dual evaluation system combining process assessment and practical skills assessment is established to quantify and output teaching effectiveness data.

9. An electronic device, comprising: It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the program to implement the AI-based intelligent teaching method for the four electrical systems of a high-speed railway as claimed in claim 8.

10. A computer readable storage medium having stored thereon a computer program or instructions, characterized in that, When a computer program or instruction is executed, it implements the AI-based intelligent teaching method for the four electrical systems of a high-speed railway as claimed in claim 8.

Images