An intelligent training system and method for energy courses

Abstract

The application discloses an energy course intelligent training system and method, and relates to the technical field of skill training, which comprises the following steps: collecting standard norms, user behaviors and test data; based on a pre-defined ability post model, extracting knowledge points from standard data and correlating and mapping them to build a dynamic knowledge graph; fusing multi-source behavior data to build a user portrait depicting learning style and knowledge level; generating an individualized learning path based on the graph and the portrait, and adapting course resources through a multi-strategy intelligent recommendation engine; relying on three-dimensional modeling, physical simulation and VR technology to build a virtual-real integrated practical training environment, realizing high-fidelity operation training and data recording; and performing multi-dimensional fusion evaluation on learning, operation and test data to generate a comprehensive evaluation report with equal emphasis on process and result. The application changes the traditional energy training from an experience-driven, content-solidified unified teaching mode to a data-driven, dynamically adapted individualized and precise training mode.

Description

Technical Field

[0001] This invention relates to the field of skills training technology, specifically to an intelligent training system and method for energy-related courses. Background Technology

[0002] With the transformation of the global energy structure and the rapid development of integrated energy systems, the industry's demand for professionals with cross-disciplinary knowledge, proficient operational skills, and rapid problem-solving abilities is increasingly urgent. However, traditional energy-related training courses generally suffer from the following limitations: training content relies heavily on static textbooks and theoretical lectures, making it difficult to keep pace with dynamically evolving technical specifications and actual job requirements, leading to a disconnect between learning and application; practical training is severely lacking or merely a formality due to the high-risk, high-cost, or inaccessible nature of real energy equipment, and even when simulation software is used, it often lacks high-fidelity physical characteristics and immersive operational experiences, making it difficult for trainees to obtain near-realistic skills training. Furthermore, existing training systems typically lack in-depth understanding and personalized guidance of the trainee's learning process. Learning behavior data, operational data, and test data are isolated from each other, failing to be effectively integrated to build accurate user profiles, resulting in course recommendations relying on general rules and lacking specificity. At the same time, assessment methods mostly focus on final theoretical test scores, failing to provide a comprehensive, multi-dimensional quantitative evaluation of skill proficiency, problem-solving logic, and teamwork abilities throughout the learning process. These shortcomings collectively lead to problems such as low efficiency in traditional training and a gap between talent skills and job requirements. Therefore, there is an urgent need for a training method that can integrate multi-source data, accurately match job skills, provide a highly realistic training environment, and achieve personalized intelligent guidance, in order to drive the transformation of energy professional talent training towards digitalization and intelligence. Summary of the Invention

[0003] To provide a training method that can integrate multi-source data, accurately match job skills, provide a highly realistic training environment, and achieve personalized intelligent guidance, this invention proposes an intelligent training method for energy-related courses, including the following steps: S1: Collect standard and specification data in the energy field, as well as user learning behavior data, operational behavior data, and test evaluation data; S2: Based on a predefined competency model and job system, knowledge points and concepts extracted from standards and norms data in the energy field will be linked to establish a mapping relationship between knowledge points, skills, competencies and jobs, so as to dynamically build and update the knowledge graph in the energy field. S3: Based on learning behavior data, operational behavior data, and test evaluation data, analyze users' learning styles and knowledge mastery levels, and combine them with competency models and job systems to determine their competency requirements in order to build user profiles; S4: Based on knowledge graphs and user profiles, generate personalized course learning paths, and use a multi-dimensional recommendation strategy that integrates knowledge graphs and learning styles, while dynamically optimizing based on user feedback to recommend course resources to users; S5: Construct a virtual training environment for energy equipment and establish a data interaction channel between the virtual environment and the real equipment. Users can use VR devices to train their operations in the virtual environment and record their operation data. S6: Based on learning behavior data, test evaluation data, and operational data, conduct a multi-dimensional evaluation that combines process and outcome, and generate a comprehensive evaluation report.

[0004] This invention transforms traditional energy training from an experience-driven, content-fixed, unified teaching model into a data-driven, dynamically adaptable, personalized, and precise training model by constructing a closed-loop process encompassing data collection, knowledge graphs, user profiling, intelligent recommendation, virtual and real-world training, and multi-dimensional evaluation. This significantly improves the relevance, effectiveness, and job suitability of the training.

[0005] Furthermore, in step S2, the dynamic construction and updating of the energy domain knowledge graph is specifically achieved through the following methods: A hypergraph structure is constructed to represent the multidimensional relationships between knowledge points, skills, equipment and fault handling methods, and a hypergraph convolutional network containing a hyperedge aggregation layer and a hypergraph attention mechanism is used to model these relationships. A topology-driven incremental update mechanism is adopted to dynamically update the weights of relevant relationships in the knowledge graph in response to the user's learning content; The evolution trend of technology popularity is analyzed by time-scanning variables and probability distribution models, and the knowledge graph is optimized based on the evolution trend.

[0006] Furthermore, in step S3, the construction of the user profile is specifically achieved in the following way: Clustering algorithms are used to analyze learning behavior data in order to identify users' learning styles; Key operation instruction sequences are extracted from operation behavior data by regular expression matching, and key technical points and standard requirements are extracted from standard and specification data by applying the TF-IDF algorithm. Based on test evaluation data and extracted key operation sequences, a user-knowledge point mastery matrix is ​​constructed to quantify the user's mastery of each knowledge point. By combining learning styles, knowledge mastery levels, and competency requirements determined by competency models and job systems, a structured user profile is generated.

[0007] Furthermore, in step S4, the multi-dimensional recommendation strategy is a hybrid recommendation strategy, which is generated and optimized in the following way: The system integrates at least two of the following strategies to generate initial recommendation results: association recommendation based on knowledge graphs, adaptation recommendation based on learning styles, difficulty recommendation based on the knowledge mastery and operational proficiency reflected in user profiles, and social recommendation based on similar user choices. The Deep Reinforcement Learning (DQN) algorithm is used to dynamically adjust the weights of each recommendation strategy based on real-time user feedback on the initial recommendation results.

[0008] Furthermore, in step S5, the virtual training environment for energy equipment is constructed in the following way: Using 3D modeling software and a physics engine, high-fidelity 3D models of energy equipment are created, and the physical characteristics and working logic of the energy equipment are simulated. Establish a data interaction channel between the virtual environment and the real device to achieve state synchronization and fault simulation between the virtual environment and the real device; Using VR headsets and interactive devices, the system records user operation data during equipment operation training in a virtual training environment.

[0009] Furthermore, in step S6, the multi-dimensional evaluation is achieved in the following way: Process assessment: Based on learning behavior data, assess learning progress and knowledge mastery trends, and based on operational data, assess the accuracy, time consumption, and proficiency of skill operation sequence; Outcome-based assessment: Based on test evaluation data, assess the ability to apply knowledge, and combine the completion of simulation training or actual project cases to assess problem-solving ability; Comprehensive evaluation: Assign preset weights to each indicator of process evaluation and outcome evaluation, and generate a comprehensive evaluation report through weighted calculation.

[0010] Furthermore, the assessment of knowledge application capabilities based on test evaluation data is performed in the following manner: From a pre-set assessment question bank, various question types, including multiple choice, fill-in-the-blank, operation, and project questions, are selected to generate assessment tasks; The assessment tasks are conducted in the form of online assessments, timed assessments, or randomly generated tests, and scores are given based on the user's completion results to obtain test evaluation data.

[0011] Furthermore, in step S1, the data collected specifically includes: Operational behavior data, including operation sequence, operation time, and operation error type; Test evaluation data includes test results, training scores, and project evaluations; Learning behavior data includes user eye movement patterns, facial expressions, or heart rate changes collected through biosensors.

[0012] This invention also proposes an intelligent training system, comprising: The data acquisition module is configured to execute S1; The knowledge graph construction module is configured to execute S2. The user profile building module is configured to execute S3; The intelligent recommendation and course management module is configured to execute S4. The virtual-real fusion training module is configured to execute S5. The assessment module is configured to execute S6.

[0013] Furthermore, it also includes: The data interaction module is configured to enable data calls and transfers between modules through a standardized RESTful API interface, and to use message queue technology to handle high-concurrency requests.

[0014] Compared with the prior art, the present invention has at least the following beneficial effects: (1) The present invention proposes an intelligent training system and method for energy courses, which utilizes multi-source data fusion and knowledge graph dynamically constructed by hypergraph convolutional network to map discrete knowledge points, skills, abilities and job requirements in multiple layers, thereby solving the problem of disconnect between content and job requirements in traditional training; (2) By integrating learning behavior, operation logs and even physiological data to construct a precise user profile, the system can deeply understand the individual status of trainees and generate dynamically optimized personalized learning paths and course recommendations based on this, which significantly improves the relevance and efficiency of training. (3) The virtual-real integration training environment created by using 3D modeling, physics engine and VR technology can not only safely and cost-effectively simulate high-risk operations and complex faults, but also realize the linkage between virtual operation and real equipment status through data channels, which greatly enhances the immersion and practicality of skills training. (4) Through a multi-dimensional assessment system that runs through the entire learning process, it is possible to conduct quantitative analysis and comprehensive feedback on knowledge acquisition, skill operation and problem-solving ability, and form a closed loop with the recommendation system, so that the training process can continuously optimize itself and ultimately achieve adaptive matching between talent training and industry job requirements. Attached Figure Description

[0015] Figure 1 A flowchart illustrating the steps of an intelligent training system and method for energy-related courses; Figure 2 This is a diagram of an intelligent training system for energy-related courses and its modules. Detailed Implementation

[0016] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0017] With the profound transformation of the global energy structure and the accelerated construction of new power systems, the energy industry is facing unprecedented technological changes and talent challenges. Traditional training models, relying on centralized lectures, static textbooks, and limited practical exercises, are insufficient to meet the training needs for advanced skills such as integrated energy system operation, intelligent maintenance, and rapid fault response. Problems such as the disconnect between training content and technological development, the mismatch between skills training and job requirements, the lack of personalized guidance in the learning process, and the difficulty in scientifically quantifying training effectiveness have become common bottlenecks restricting the development of a talent pool in the energy industry. To address these challenges, such as... Figure 1 As shown, this invention proposes an intelligent training method for energy-related courses, which mainly includes the following steps: S1: Collect standard and specification data in the energy field, as well as user learning behavior data, operational behavior data, and test evaluation data; S2: Based on a predefined competency model and job system, knowledge points and concepts extracted from standards and norms data in the energy field will be linked to establish a mapping relationship between knowledge points, skills, competencies and jobs, so as to dynamically build and update the knowledge graph in the energy field. S3: Based on learning behavior data, operational behavior data, and test evaluation data, analyze users' learning styles and knowledge mastery levels, and combine them with competency models and job systems to determine their competency requirements in order to build user profiles; S4: Based on knowledge graphs and user profiles, generate personalized course learning paths, and use a multi-dimensional recommendation strategy that integrates knowledge graphs and learning styles, while dynamically optimizing based on user feedback to recommend course resources to users; S5: Construct a virtual training environment for energy equipment and establish a data interaction channel between the virtual environment and the real equipment. Users can use VR devices to train their operations in the virtual environment and record their operation data. S6: Based on learning behavior data, test evaluation data, and operational data, conduct a multi-dimensional evaluation that combines process and outcome, and generate a comprehensive evaluation report.

[0018] Furthermore, this invention also proposes an intelligent training system, such as... Figure 2 As shown, the system includes a data acquisition module, a knowledge graph construction module, a user profile construction module, an intelligent recommendation and course management module, a virtual-real fusion training module, and an assessment module to perform the above steps respectively. A data interaction module enables data calls and transmission between modules through a standardized RESTful API interface, and message queue technology is used to handle high-concurrency requests to achieve asynchronous processing and system load balancing.

[0019] As described in the steps above, the operation of this intelligent training system begins with the comprehensive collection and aggregation of multi-dimensional data. The data collection module continuously acquires information from multiple channels: on the one hand, it automatically crawls and integrates energy-related standards and specifications, such as the "Preventive Testing Regulations for Power Equipment," from the company's internal knowledge base, industry standard platforms, and publicly available technical documents to form a structured initial knowledge base. On the other hand, during all interactions between the trainee and the system, this module simultaneously captures detailed learning behavior data, operational behavior data, and test evaluation data. Learning behavior data records every step the trainee takes on the platform, including the viewing time and replay segments of course videos, the time spent on document pages, and the frequency of participation in forum discussions. When trainees enter the simulation training phase, all their actions, whether clicking buttons on the virtual control panel, turning simulated switches, or using virtual tools to disassemble equipment, are precisely recorded as operational behavior data, including the precise sequence of operation steps, the time taken for each step, and any error types that deviate from standard procedures. The scores, ratings, and evaluations generated by trainees after completing chapter quizzes, mock exams, or project assignments constitute the test evaluation data.

[0020] To gain a deeper understanding of learners’ cognitive load and stress response, the system also integrates biosensors to collect physiological data during learning or operation. For example, it uses eye trackers to track and analyze the visual focus path of learners when studying complex electrical diagrams, analyzes their emotional tension when dealing with simulated sudden failures through facial recognition, and monitors psychophysiological changes during operation through heart rate sensors.

[0021] Next, the knowledge graph construction module is responsible for transforming static knowledge into a dynamic intelligent network. This module first uses natural language processing technology to automatically extract key conceptual entities such as "transformer," "relay protection," and "insulation aging," along with their interrelationships, from the aggregated standard and specification texts, forming a preliminary set of knowledge points. Subsequently, based on pre-built capability models and job systems derived from industry expert experience and job analysis, it doesn't simply list knowledge points, but constructs a four-layer mapping network that closely links specific knowledge points, practical skills, comprehensive abilities, and target job positions. For example, the knowledge point "transformer oil chromatography data analysis" is mapped to the skill "dissolved gas analysis in oil," which is further linked to the core capability of "accurate equipment condition assessment," ultimately pointing to the job requirements of the "senior substation maintenance technician" position. To efficiently express the complex relationships that a single fault phenomenon may involve multiple equipment principles, several diagnostic skills, and a series of operational steps, the knowledge graph construction module uses a hypergraph structure for modeling: various entities serve as nodes, while a complete task or capability requirement acts as a hyperedge connecting multiple nodes.

[0022] To enable the knowledge graph to learn from learners' actual operations and remain up-to-date, the module also introduces a dynamic modeling tool: a hypergraph convolutional network. This network first uses a Res2Net model to perform deep analysis on the learners' operation videos, extracting the spatial location features and temporal sequence features of the actions. Optical flow binarization filtering is then used to refine the features, focusing on the meaningful actions themselves. These extracted dynamic features are used to continuously optimize the feature representation of corresponding nodes in the hypergraph (e.g., "circuit breaker debugging"). The hyperedge aggregation layer within the network integrates information from all nodes within a hyperedge, while the hypergraph attention mechanism intelligently determines the differences in the importance of different knowledge points for mastering a particular skill.

[0023] Meanwhile, the knowledge graph is updated using a topology-driven incremental update mechanism. When the system detects that a trainee is currently learning about "photovoltaic inverter fault diagnosis," it only recalculates and optimizes the local hyperedge networks directly related to this topic in the knowledge graph, rather than performing a global update, thus improving update efficiency. The knowledge graph construction module also uses a multi-source data fusion algorithm to correlate and compare successful or typical error cases generated by trainees in virtual training with a database of real-world fault handling cases from enterprises, continuously validating and enriching the theoretical graph with practical data. In addition, the system has a knowledge evolution monitoring function, which regularly scans the frequency of specific technical terms in industry technical documents, patents, and accident reports, analyzes their popularity trends using a probability distribution model, and automatically adjusts the weight and salience of relevant content in the knowledge graph to ensure that the training content is always in sync with the forefront of industry technology.

[0024] Based on the constructed dynamic knowledge graph and collected behavioral data, the user profile building module begins to create a digital portrait for each learner. This module first uses clustering analysis algorithms to mine learners' learning behavior patterns, automatically categorizing them into visual, auditory, operational, or social learning styles based on their preferences for video learning, text reading, hands-on activities, or collaborative discussions. Subsequently, the module deeply analyzes learners' operational behavior logs, using pre-defined regular expression templates to accurately match and extract key action sequences from operational instructions. For example, it identifies the standard safety procedure "testing for electricity → connecting grounding wire → setting up a fence" from a long operational log. Simultaneously, it applies the TF-IDF algorithm to analyze the technical standard texts learned by learners, extracting core terminology and mandatory regulatory clauses. Combining learners' test scores and practical records, the module constructs a dynamic "user-knowledge point mastery matrix," using a quantitative model to calculate their real-time mastery level of each knowledge point. Ultimately, the system integrates the identified learning style, the quantified knowledge mastery matrix, and the core competency requirements mapped from the competency model based on the user's career goals to generate a unique and structured user profile.

[0025] Building upon this foundation, the intelligent recommendation and course management module, upon receiving the knowledge graph and user profiles, launches a multi-strategy integrated intelligent recommendation engine. This engine operates multiple recommendation logics in parallel: First, association-based recommendation based on the knowledge graph, recommending prerequisite content and advanced courses to students' knowledge gaps according to the graph's association paths, systematically filling these gaps in their abilities; second, adaptation recommendation based on learning style, prioritizing animated courses for visual learners and arranging more VR training modules for hands-on learners; third, difficulty-based recommendation based on ability level, dynamically adjusting the difficulty gradient of recommended content according to the real-time mastery level reflected in the user profile; and fourth, social recommendation based on collective intelligence, referencing the choices and positive reviews of other students with similar profiles to discover potentially high-quality courses.

[0026] These strategies are not fixed; they are integrated through a fusion model, with their weights dynamically managed by a deep reinforcement learning (DQN) algorithm. The algorithm continuously observes learners' feedback behavior on recommended content: clicks, learning time, completion rate, and ratings, and uses this feedback as reward signals to continuously learn and adjust the weights of each strategy. This makes the entire recommendation system increasingly understand learners, achieving personalized adaptive recommendations and planning the optimal personalized learning path for each learner.

[0027] To transform learned knowledge into muscle memory and practical skills, a high-fidelity, repeatable, and zero-risk training environment is constructed through a virtual-real integrated training module. 3D modeling software is used to create life-size models of energy equipment such as transformers, circuit breakers, and wind turbines. A physics simulation engine injects realistic physical properties into these virtual models, providing corresponding physical feedback during operation. Functional simulation logic ensures that the virtual equipment can operate, respond, and even "fail" like real equipment. By employing industrial communication protocols such as OPC UA, this module bridges the virtual environment with the real equipment monitoring system, achieving state synchronization and data exchange. This means that instructors can safely simulate a system failure with operational risks in the real world (such as "total power loss") within the system, while trainees, wearing VR headsets, apply their learned knowledge to diagnose, make decisions, and handle the situation in the virtual power plant environment. The system records every detail of the trainee's operation during this process, including sequence, time, path, and errors, forming a complete skill operation file.

[0028] This invention not only focuses on the final result but also emphasizes the learning process. The assessment module analyzes learners' learning behavior data to evaluate their learning progress and knowledge assimilation trends in real time. By parsing operation logs, it accurately assesses the standardization, proficiency, and efficiency of their skills. At stages requiring result verification, the system can flexibly generate online tests, timed skills challenges, or randomly generated assessments from a vast question bank containing various question types, including multiple-choice, operational, and comprehensive project questions. Tasks completed or project solutions submitted by learners during simulated training also receive a comprehensive evaluation combining automatic system scoring and teacher feedback. Finally, all these process indicators (such as learning participation and operational accuracy) and outcome indicators (such as test scores and project scores) are assigned scientific and reasonable weights, summarized and integrated through a weighted calculation model, producing a comprehensive assessment report. Based on this report, a radar chart of learners' abilities in knowledge, skills, and qualities is drawn, and a direct comparison is made with the ability model of their target positions, thus providing a data foundation for continuous improvement.

[0029] In summary, the intelligent training system and method for energy-related courses proposed in this invention utilizes multi-source data fusion and a knowledge graph dynamically constructed using hypergraph convolutional networks to map discrete knowledge points, skills, and abilities to job requirements in multiple layers, thereby solving the problem of the disconnect between content and job requirements in traditional training.

[0030] By integrating learning behavior, operation logs, and even physiological data to construct precise user profiles, the system can deeply understand the learner's individual state and generate dynamically optimized personalized learning paths and course recommendations based on this, significantly improving the relevance and efficiency of training.

[0031] The virtual-real fusion training environment created by leveraging 3D modeling, physics engines, and VR technology can not only safely and cost-effectively simulate high-risk operations and complex faults, but also achieve linkage between virtual operations and real equipment status through data channels, greatly enhancing the immersion and practicality of skills training.

[0032] Through a multi-dimensional assessment system that runs through the entire learning process, quantitative analysis and comprehensive feedback can be provided on knowledge acquisition, skills operation, and problem-solving abilities. This forms a closed loop with the recommendation system, enabling the training process to continuously self-optimize and ultimately achieve adaptive matching between talent development and industry job requirements.

[0033] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0034] Furthermore, in this invention, descriptions involving terms such as "first," "second," and "a" are for descriptive purposes only and should not be construed as indicating or implying their 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 invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0035] In this invention, unless otherwise explicitly specified and limited, the terms "connection," "fixed," etc., should be interpreted broadly. For example, "fixed" can mean a fixed connection, a detachable connection, or an integral part; it can mean a mechanical connection or an electrical connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0036] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but only if they are feasible for those skilled in the art. If the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

Claims

1. A smart training method for energy-related courses, characterized in that, Including the following steps: S1: Collect standard and specification data in the energy field, as well as user learning behavior data, operational behavior data, and test evaluation data; S2: Based on a predefined competency model and job system, knowledge points and concepts extracted from standards and norms data in the energy field will be linked together to establish a mapping relationship between knowledge points, skills, competencies and jobs, so as to dynamically build and update the knowledge graph in the energy field. S3: Based on learning behavior data, operational behavior data, and test evaluation data, analyze users' learning styles and knowledge mastery levels, and combine them with competency models and job systems to determine their competency requirements in order to build user profiles; S4: Based on knowledge graphs and user profiles, generate personalized course learning paths, and use a multi-dimensional recommendation strategy that integrates knowledge graphs and learning styles, while dynamically optimizing based on user feedback to recommend course resources to users; S5: Construct a virtual training environment for energy equipment and establish a data interaction channel between the virtual environment and the real equipment. Users can use VR devices to train their operations in the virtual environment and record their operation data. S6: Based on learning behavior data, test evaluation data, and operational data, conduct a multi-dimensional evaluation that combines process and outcome, and generate a comprehensive evaluation report.

2. The intelligent training method for energy-related courses as described in claim 1, characterized in that, In step S2, the dynamic construction and updating of the energy domain knowledge graph is specifically achieved in the following ways: A hypergraph structure is constructed to represent the multidimensional relationships between knowledge points, skills, equipment and fault handling methods, and a hypergraph convolutional network containing a hyperedge aggregation layer and a hypergraph attention mechanism is used to model these relationships. A topology-driven incremental update mechanism is adopted to dynamically update the weights of relevant relationships in the knowledge graph in response to the user's learning content; The evolution trend of technology popularity is analyzed by time-scanning variables and probability distribution models, and the knowledge graph is optimized based on the evolution trend.

3. The intelligent training method for energy-related courses as described in claim 1, characterized in that, In step S3, the user profile is constructed in the following way: Clustering algorithms are used to analyze learning behavior data in order to identify users' learning styles; Key operation instruction sequences are extracted from operation behavior data by regular expression matching, and key technical points and standard requirements are extracted from standard and specification data by applying the TF-IDF algorithm. Based on test evaluation data and extracted key operation sequences, a user-knowledge point mastery matrix is ​​constructed to quantify the user's mastery of each knowledge point. By combining learning styles, knowledge mastery levels, and competency requirements determined by competency models and job systems, a structured user profile is generated.

4. The intelligent training method for energy-related courses as described in claim 1, characterized in that, In step S4, the multi-dimensional recommendation strategy is a hybrid recommendation strategy, which is generated and optimized in the following way: The system integrates at least two strategies from knowledge graph-based association recommendation, learning style-based adaptation recommendation, difficulty recommendation based on the knowledge mastery and operational proficiency reflected in user profiles, and social recommendation based on similar user choices to generate initial recommendation results. The Deep Reinforcement Learning (DQN) algorithm is used to dynamically adjust the weights of each recommendation strategy based on real-time user feedback on the initial recommendation results.

5. The intelligent training method for energy-related courses as described in claim 1, characterized in that, In step S5, the virtual training environment for energy equipment is constructed in the following way: Using 3D modeling software and a physics engine, high-fidelity 3D models of energy equipment are created, and the physical characteristics and working logic of the energy equipment are simulated. Establish a data interaction channel between the virtual environment and the real device to achieve state synchronization and fault simulation between the virtual environment and the real device; Using VR headsets and interactive devices, the system records user operation data during equipment operation training in a virtual training environment.

6. The intelligent training method for energy-related courses as described in claim 1, characterized in that, In step S6, the multi-dimensional evaluation is achieved in the following way: Process assessment: Based on learning behavior data, assess learning progress and knowledge mastery trends, and based on operational data, assess the accuracy, time consumption, and proficiency of skill operation sequence; Outcome-based assessment: The ability to apply knowledge is evaluated based on test data, and the problem-solving ability is evaluated in conjunction with the completion of simulation training or actual project cases. Comprehensive evaluation: Assign preset weights to each indicator of process evaluation and outcome evaluation, and generate a comprehensive evaluation report through weighted calculation.

7. The intelligent training method for energy-related courses as described in claim 6, characterized in that, The assessment of knowledge application capabilities based on test evaluation data is performed in the following ways: From a pre-set assessment question bank, various question types, including multiple choice, fill-in-the-blank, operation, and project questions, are selected to generate assessment tasks; The assessment tasks are conducted in the form of online assessments, timed assessments, or randomly generated tests, and scores are given based on the user's completion results to obtain test evaluation data.

8. The intelligent training method for energy-related courses as described in claim 1, characterized in that, The data collected in step S1 specifically includes: Operational behavior data, including operation sequence, operation time, and operation error type; Test evaluation data includes test results, training scores, and project evaluations; Learning behavior data includes user eye movement patterns, facial expressions, or heart rate changes collected through biosensors.

9. An intelligent training system for implementing the intelligent training method for energy-related courses as described in any one of claims 1 to 8, characterized in that, include: The data acquisition module is configured to execute S1; The knowledge graph construction module is configured to execute S2. The user profile building module is configured to execute S3; The intelligent recommendation and course management module is configured to execute S4. The virtual-real fusion training module is configured to execute S5. The assessment module is configured to execute S6.

10. The intelligent training system as described in claim 9, characterized in that, Also includes: The data interaction module is configured to enable data calls and transfers between modules through a standardized RESTful API interface, and to use message queue technology to handle high-concurrency requests.

Images