Practical training method and system for electric power metering equipment, computer equipment and medium

Abstract

The invention relates to the technical field of equipment practical training, and discloses an electric power metering equipment practical training method and system, computer equipment and a medium, and the method comprises the steps: carrying out knowledge capability evaluation, learning style analysis and interest entity extraction on each practical training student, and obtaining a capability value, a knowledge weak point set, a learning style and an interest vector of the practical training student; constructing student portraits corresponding to practical training students; according to the trainee portrait, a learning sub-graph is extracted from the electric power metering equipment knowledge graph, a KgRank algorithm is adopted, knowledge nodes of the learning sub-graph are sorted, and a learning path is generated; learning content based on AR interaction is generated according to the learning path; and in response to completion of interactive learning, updating the knowledge graph and the trainee portrait according to the learning behavior of the training trainee, and adjusting the learning path. According to the method, through immersive interaction of the AR technology and structured reasoning of the knowledge graph, dynamic evaluation and path recommendation are combined, and the safety, efficiency and individuation level of electric power practical training are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment training, and particularly to a training method, system, computer device and medium for electric power metering equipment. Background Art

[0002] In the traditional training process of electric power metering equipment, it mainly relies on paper materials, on-site explanations and actual operation demonstrations. The defects and deficiencies of these conventional methods are mainly manifested in low information acquisition efficiency, high understanding threshold, poor training effect and prominent safety hazards. Among them, the low acquisition efficiency is manifested in the large number of paper materials and difficult retrieval, and it is difficult for newly recruited personnel to quickly obtain accurate information (such as the internal structure of the electric energy meter, wiring specifications, etc.). The high understanding threshold is manifested in the complex equipment principle and operation process, and it is difficult to intuitively display hidden components (such as the internal coil of the transformer) and dynamic operation mechanism by simple explanation. The poor training effect is manifested in the lack of personalized support and interactive design. It is easy for trainees to have insufficient skill mastery due to progress differences, and the traditional evaluation methods (written test + practical operation) cannot accurately quantify the ability short board. The prominent safety hazard is manifested in the risk of electric shock in the actual operation of the high-voltage environment, and the high loss cost caused by frequent use of equipment (such as the loss rate of electric energy meter disassembly and assembly training ≥ 15%).

[0003] Therefore, there is an urgent need for a new training method that can meet the safety, efficiency and personalized needs of electric power metering equipment training. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides a training method, system, computer device and medium for electric power metering equipment, so as to solve the problem that the traditional training of electric power metering equipment cannot meet the safety, efficiency and personalized needs, and achieve the technical effects of improving training efficiency and training safety.

[0005] In the first aspect, the present invention provides a training method for electric power metering equipment, and the method includes:

[0006] Conduct knowledge and ability evaluation, learning style analysis and interest entity extraction for each training trainee to obtain the ability value, knowledge weak point set, learning style and interest vector of the training trainee;

[0007] Construct a trainee portrait corresponding to the training trainee according to the ability value, the knowledge weak point set, the learning style and the interest vector;

[0008] Extract a learning subgraph from a pre-constructed knowledge graph of electric power metering equipment according to the trainee portrait, and use the KgRank algorithm to sort the knowledge nodes of the learning subgraph to generate a learning path for the training trainee;

[0009] Generate learning content based on AR interaction according to the learning path, so that the trainees can interactively learn the learning content;

[0010] In response to the completion of the interactive learning, update the knowledge graph and the trainee portrait according to the learning behavior of the trainees, and adjust the learning path until the training of power metering equipment ends.

[0011] Further, the steps of performing knowledge and ability evaluation, learning style analysis, and interest entity extraction on each trainee to obtain the ability value, knowledge weak point set, learning style, and interest vector of the trainee include:

[0012] Judge whether the trainee is being evaluated for the first time. If so, generate initial test questions randomly according to the registration information of the trainee and conduct an initial evaluation on the trainee. If not, generate personalized test questions according to the previous evaluation results and conduct a dynamic evaluation on the trainee;

[0013] Obtain the ability value of the trainee according to the correct rate of the dynamic evaluation, and use the knowledge points that failed the evaluation as knowledge weak points to form a knowledge weak point set;

[0014] Obtain the explicit learning style of the trainee according to a preset scale, and correct the explicit learning style according to the learning behavior of the trainee to obtain the learning style of the trainee;

[0015] Use a preset interest recognition model to extract interest entities from the learning behavior log of the trainee to obtain the interest vector of the trainee, and the interest recognition model is constructed based on a bidirectional long short-term memory neural network and a random conditional field.

[0016] Further, the steps of extracting a learning subgraph from a pre-constructed power metering equipment knowledge graph according to the trainee portrait, sorting the knowledge nodes of the learning subgraph using the KgRank algorithm, and generating the learning path of the trainee include:

[0017] Extract a learning subgraph from the power metering equipment knowledge graph according to the knowledge weak point set;

[0018] Determine the mastery degree of each knowledge node in the learning subgraph according to the ability value;

[0019] Calculate the resource type weight of each knowledge node according to the matching degree between the learning style and the resource type;

[0020] Calculate the interest matching degree of each knowledge node according to the similarity between the interest vector and the knowledge node;

[0021] According to the mastery level, the resource type weight, and the interest matching degree, use the KgRank algorithm to calculate the dynamic weight values of each knowledge node;

[0022] According to the dynamic weight values, sort each knowledge node of the learning sub-graph to obtain a learning path.

[0023] Further, after the step of using the KgRank algorithm to calculate the dynamic weight values of each knowledge node according to the mastery level, the resource type weight, and the interest matching degree, it further includes:

[0024] According to the PageRank value of each knowledge node, correct the dynamic weight value to obtain a corrected dynamic weight value;

[0025] The corrected dynamic weight value is represented by the following formula:

[0026]

[0027] In the formula, n i represents the i-th knowledge node, Score(n i ) represents the corrected dynamic weight value of the i-th knowledge node, PR(n i ) represents the PageRank value of the i-th knowledge node, Similarity(n i ) represents the interest matching degree between the i-th knowledge node and the interest vector, mi represents the ability value of the i-th knowledge node, ω k represents the initial weight of the k-th resource type corresponding to the i-th knowledge node, S uk represents the matching degree between the learning style and the k-th resource type, and α, β, γ, and δ all represent weight coefficients.

[0028] Further, the step of updating the knowledge graph and the trainee portrait according to the learning behavior of the training trainees and adjusting the learning path includes:

[0029] Re-conduct knowledge ability assessment, learning style analysis, and interest entity extraction for each training trainee to obtain the latest ability value, the latest set of knowledge weaknesses, the latest learning style, and the latest interest vector;

[0030] Taking the number of trainees as a unit, count each knowledge weakness in the latest set of knowledge weaknesses;

[0031] If the number of trainees corresponding to the knowledge weakness is greater than the quantity threshold, adjust the associated weight of the predecessor knowledge point of the knowledge weakness in the knowledge graph;

[0032] Adjust the trainee portrait according to the latest ability value, the latest set of knowledge weaknesses, the latest learning style, and the latest interest vector of each training trainee, and regenerate the corresponding learning path according to the adjusted trainee portrait and the adjusted knowledge graph.

[0033] Further, the steps for constructing the power metering equipment knowledge graph include:

[0034] Obtain the knowledge data of power metering equipment from the professional model library and the learning database, and perform data preprocessing on the knowledge data to obtain a cleaned structured data set;

[0035] Define entities as knowledge points, questions, trainees, and majors, and define relationships as hierarchical relationships and mapping relationships, and perform knowledge extraction on the structured data set to obtain a power metering equipment knowledge graph.

[0036] Further, the steps for performing knowledge extraction on the structured data set to obtain a power metering equipment knowledge graph include:

[0037] Extract entity relationship multi-tuples from the structured data set to obtain explicit knowledge;

[0038] Extract potential associations between entities from historical fault cases through a preset knowledge extraction model, and infer the logic between knowledge points through a rule engine to obtain implicit knowledge, where the knowledge extraction model is constructed based on a bidirectional long short-term memory neural network;

[0039] Perform knowledge fusion on the explicit knowledge and the implicit knowledge to obtain a power metering equipment knowledge graph.

[0040] In a second aspect, the present invention provides a power metering equipment training system, and the system includes:

[0041] A trainee portrait construction module, configured to perform knowledge ability evaluation, learning style analysis, and interest entity extraction on each training trainee to obtain the ability value, the set of knowledge weaknesses, the learning style, and the interest vector of the training trainee;

[0042] Construct a trainee portrait corresponding to the training trainee according to the ability value, the set of knowledge weaknesses, the learning style, and the interest vector;

[0043] A learning path generation module, configured to extract a learning subgraph from a pre-constructed power metering equipment knowledge graph according to the trainee portrait, and use the KgRank algorithm to sort the knowledge nodes of the learning subgraph to generate a learning path for the training trainee;

[0044] The AR interaction learning module is used to generate AR interaction-based learning content according to the learning path, so that the trainees can conduct interactive learning on the learning content;

[0045] The learning path update module is used to, in response to the completion of the interactive learning, update the knowledge graph and the trainee portrait according to the learning behaviors of the trainees, and adjust the learning path until the training of the power metering equipment ends.

[0046] In a third aspect, an embodiment of the present invention further provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the above method are implemented.

[0047] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above method are implemented.

[0048] The present invention provides a method, a system, a computer device, and a medium for training power metering equipment. Through the three-dimensional visualization of AR interaction, the present invention improves the understanding efficiency of trainees for complex structures, ensures the safety of training, reduces equipment loss, accurately depicts the trainee portrait through personalized dynamic evaluation, improves the systematicness of training content by constructing a hierarchical knowledge network, and improves the learning efficiency of trainees through personalized learning path recommendation. Through the immersive interaction of AR technology and the structured reasoning of the knowledge graph, combined with dynamic evaluation and path recommendation, the present invention effectively improves the safety, efficiency, and personalization level of power metering equipment training. Description of the Drawings

[0049] Figure 1 is a schematic flowchart of the method for training power metering equipment according to an embodiment of the present invention;

[0050] Figure 2 is a schematic structural diagram of the system for training power metering equipment according to an embodiment of the present invention;

[0051] Figure 3 is an internal structural diagram of the computer device in an embodiment of the present invention. Detailed Embodiments

[0052] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0053] Please refer to Figure 1 , a training method for power metering equipment proposed in the first embodiment of the present invention, which includes steps S10 to S50:

[0054] Step S10, conduct knowledge and ability assessment, learning style analysis, and interest entity extraction for each training student to obtain the ability value, knowledge weakness set, learning style, and interest vector of the training student;

[0055] Step S20, construct a student portrait corresponding to the training student according to the ability value, the knowledge weakness set, the learning style, and the interest vector;

[0056] Step S30, extract a learning subgraph from a pre-constructed power metering equipment knowledge graph according to the student portrait, use the KgRank algorithm to sort the knowledge nodes of the learning subgraph, and generate a learning path for the training student;

[0057] Step S40, generate AR interaction-based learning content according to the learning path, so that the training student can perform interactive learning on the learning content;

[0058] Step S50, in response to the completion of the interactive learning, update the knowledge graph and the student portrait according to the learning behavior of the training student, and adjust the learning path until the power metering equipment training ends.

[0059] The present invention provides a training method for power metering equipment based on the combination of AR interaction and knowledge graph. By analyzing the learning situation of training students and the knowledge graph of power metering equipment, personalized learning paths are planned for training students, and AR interaction technology is used for training to improve the learning efficiency of training students.

[0060] In this embodiment, first analyze the personal situation of the training student to construct a student portrait for each training student. The specific analysis steps include:

[0061] Judge whether the training student is undergoing the initial assessment. If so, randomly generate initial test questions according to the registration information of the training student and conduct an initial assessment on the training student. If not, generate personalized test questions according to the previous assessment results and conduct a dynamic assessment on the training student;

[0062] According to the correct rate of the dynamic assessment, obtain the ability value of the training students, and use the knowledge points that fail the assessment as knowledge weak points to form a knowledge weak point set;

[0063] According to a preset scale, obtain the explicit learning style of the training students, and correct the explicit learning style according to the learning behavior of the training students to obtain the learning style of the training students;

[0064] Use a preset interest recognition model to extract interest entities from the learning behavior log of the training students to obtain the interest vector of the training students. The interest recognition model is constructed based on a bidirectional long short-term memory neural network and a stochastic conditional random field.

[0065] In this embodiment, the situation analysis of the training students includes three aspects, namely, the analysis of knowledge mastery ability, the analysis of learning style, and the analysis of personal interests. Specifically, for the analysis of knowledge mastery ability, in this embodiment, the current knowledge mastery ability of the students is analyzed by the way of students answering questions. For students who have never been assessed, initial questions will be randomly generated according to the registration information of the students for the students to answer, so as to obtain the initial knowledge mastery ability of the students. The registration information includes name, age, gender, education background, work position, hobbies, career planning, professional title, skill level, etc.; the knowledge mastery ability includes the ability value and knowledge weak points. The ability value refers to the correct rate of the assessment, that is, the ratio of the number of correct answers to the total number of questions. Each knowledge point corresponds to several questions, and the correct rate of answering questions through this knowledge point is used to judge whether the student has mastered the knowledge point. If the assessment fails, that is, the correct rate is less than the threshold, then this knowledge point is used as a knowledge weak point. The initial assessment questions can be randomly selected from the corresponding question bank according to information related to the professional knowledge required for the position, such as the education background, work position, and skill level of the students.

[0066] After the initial assessment is completed, according to the assessment situation of the students, a formal dynamic assessment is carried out to accurately estimate the students' abilities and diagnose their knowledge. In this embodiment, a knowledge graph of power metering equipment is pre-constructed, and the specific construction steps will be described later. According to the results of the initial assessment and combined with the precursor and successor relationships of the knowledge points in the knowledge graph, the questions for the dynamic assessment are selected. First, query whether there is assessment data for the nearest precursor knowledge point and the farthest successor knowledge point of the knowledge point to be assessed; if there is measured data (the number of questions exceeds a certain threshold) for the nearest precursor knowledge point of the knowledge point and the mastery status is proficient (the answering correct rate exceeds the correct rate threshold), then skip the easy questions when selecting assessment questions and directly select medium-difficulty questions. On the contrary, if the mastery status is poor, then skip the difficult questions when selecting questions and select questions from the easy ones; if there is measured data for the farthest successor knowledge point of the knowledge point and the mastery status is proficient, then it is inferred that the mastery status of this knowledge point is proficient. If the mastery status of the farthest successor knowledge point of the knowledge point is poor, then it is considered that this knowledge point has not been mastered.

[0067] Preferably, the maximum information entropy question selection strategy can be adopted to select the questions for the dynamic assessment, and the questions that can best distinguish the students' ability levels are preferentially selected to avoid repeated testing of the mastered content. In addition, a cognitive diagnosis model can be constructed based on the item response theory or the Bayesian network to quantify the matching degree between the students' learning abilities and the difficulty of the questions, so as to achieve more targeted question selection.

[0068] Finally, according to the answering results of the dynamic assessment, the students' ability values and knowledge weak points are determined. The ability values in this embodiment include the ability values of the mastery status of each knowledge point and the ability values of the mastery status of all knowledge points. Only when the ability values of the mastery status of each knowledge point meet the standards, it is considered that the ability values of all knowledge points meet the standards.

[0069] For the analysis of learning styles, in this embodiment, a scale method is adopted. By the content filled in by the students in the scale, the explicit learning styles of the students are judged, such as video-based learning preferences, book-based learning preferences, auditory-based learning preferences, etc. Weights corresponding to different types of learning are assigned according to the preference degree. This explicit learning style will be used as the initial learning style. After the learning starts, the explicit learning style is corrected according to the students' learning behaviors. For example, if the student frequently operates the AR animation, then the weight of video-based learning is increased, etc., so as to obtain the students' learning styles.

[0070] For the analysis of the interest vector, in this embodiment, an interest recognition model is adopted to extract interest entities from the learning behavior logs of the training students, and the interest vectors of the training students are obtained. The interest recognition model is constructed based on a bidirectional long short-term memory neural network and a conditional random field, that is, a Bi-LSTM+CRF model. At the same time, a pre-trained model based on BERT is introduced into the Bi-LSTM+CRF model. The learning behavior logs of the students are analyzed through the Bi-LSTM+CRF model, and the interest entities of the students are output and mapped into the vector space to obtain the interest vectors. Finally, according to the ability value, the set of knowledge weak points, the learning style and the interest vector, a student portrait corresponding to the training student is constructed.

[0071] Then, based on the student portrait of the student, the corresponding learning path is extracted from the knowledge graph. Before explaining the generation steps of the learning path, first, the construction process of the knowledge graph of the power metering equipment is explained:

[0072] Obtain the knowledge data of the power metering equipment from the professional model library and the learning resource library, and perform data preprocessing on the knowledge data to obtain a cleaned structured data set;

[0073] Define the knowledge extraction of the structured data set with knowledge points, questions, students and majors as entity definitions, and hierarchical relationships and mapping relationships as relationship definitions to obtain the knowledge graph of power metering equipment.

[0074] In this embodiment, through a multi-source data fusion and dynamic reasoning mechanism, a structured network of power metering equipment knowledge is established. First, the relevant data of the power metering equipment is cleaned. The data includes power industry standard documents, equipment principle manuals, fault case libraries, etc. in the professional model library, such as knowledge data of professional skill evaluations and business handling such as power consumption inspection, meter reading, charging and collection, meter installation and connection, electric energy meter repair and calibration, line loss investigation, and terminal operation and maintenance under the jurisdiction of the power marketing department; teaching materials texts, AR three-dimensional model metadata, question banks (including the mapping relationship between questions and knowledge points), etc. in the learning resource library, such as textbooks, teaching audio and video, VR training courseware, teaching PPT, questions, test papers, etc. on the principle knowledge, operation steps, and fault repair of power metering equipment.

[0075] Then, the data in the professional model library and the learning resource library is preprocessed. Among them, for unstructured texts, NLP tools (such as Spacy) are used to perform word segmentation and entity recognition on textbooks and regulations (such as extracting entities such as "electric energy meter", "transformer", and "wiring principle"); for structured data, data alignment is performed to associate the questions in the question bank with the knowledge point IDs (such as question Q001 corresponding to knowledge points K001 and K002). Finally, a cleaned structured data set is output.

[0076] Then define the entities and relationships in the knowledge graph. Among them, the entity definitions include:

[0077] Knowledge points: equipment principles, operation steps, safety specifications, etc. (such as "K001 - Structure of Three-Phase Watt-Hour Meter", "K002 - Operation Specification for Energized Wiring").

[0078] Questions: assessment questions and their associated knowledge points, difficulty levels (such as "Q001 - True / False Question about Wiring, Difficulty L2").

[0079] Trainees: trainee ID, job type, historical assessment records.

[0080] Majors: power professional classifications such as electrical inspection, meter installation and connection.

[0081] The relationship definitions include:

[0082] Hierarchical relationship:

[0083] Predecessor-successor dependence between knowledge points (such as "K001→K002" means that K001 needs to be mastered first to learn K002).

[0084] The attribution relationship between majors and knowledge points (such as "Meter installation and connection major→K001, K002").

[0085] Mapping relationship:

[0086] Association between questions and knowledge points (such as "Q001→K001, K003").

[0087] The mastery status of trainees and knowledge points (such as "S001→K001: Ability value 80%").

[0088] After defining the entities and relationships, perform knowledge extraction on the structured data set, extract entity-relationship multi-tuples from the structured data set, so as to obtain the knowledge graph of power metering equipment. In fact, the directly extracted entity-relationship multi-tuples are explicit knowledge. There is also some implicit knowledge in the fault cases. Therefore, in a preferred embodiment, the present invention extracts the potential associations between entities from historical fault cases (such as "Wiring error→Watt-hour meter burnout") through a knowledge extraction model, and infers the logic between knowledge points through a rule engine (such as "If knowledge point A is the predecessor of B, and B is the predecessor of C, then A→C is an indirect dependence") to obtain implicit knowledge. Optionally, the knowledge extraction model is constructed based on a bidirectional long short-term memory neural network. Finally, fuse the explicit knowledge and implicit knowledge to obtain the power metering equipment knowledge graph.

[0089] After obtaining the trainee portrait and the power metering equipment knowledge graph, extract the learning path from the power metering equipment knowledge graph according to the trainee portrait. The specific steps include:

[0090] Extract a learning sub - graph from the power metering equipment knowledge graph according to the set of weak knowledge points;

[0091] Determine the mastery degree of each knowledge node in the learning sub - graph according to the ability value;

[0092] Calculate the resource type weight of each knowledge node according to the matching degree between the learning style and the resource type;

[0093] Calculate the interest matching degree of each knowledge node according to the similarity between the interest vector and the knowledge node;

[0094] Calculate the dynamic weight value of each knowledge node by using the KgRank algorithm according to the mastery degree, the resource type weight, and the interest matching degree;

[0095] Sort each knowledge node of the learning sub - graph according to the dynamic weight value to obtain a learning path.

[0096] In this embodiment, according to the set of weak knowledge points of the trainee \(K = \{K001, K002,\cdots, KN\}\), a connected sub - graph \(G\) with \(K\) as the core is extracted from the knowledge graph. The connected sub - graph \(G\) contains the precursor knowledge points and successor knowledge points of the weak knowledge points. Specifically, trace back along the knowledge graph to the root node to obtain the precursor knowledge points to ensure the completeness of the basic data, and extend forward along the knowledge graph to the leaf node to obtain the successor knowledge points to prevent knowledge gaps. For example, taking the weak knowledge point: K003 - Electricity theft detection as an example, the connected sub - graph contains precursor knowledge points: K001 - Structure of electric energy meter → K002 - Wiring principle → K003, and successor knowledge points: K003 → K004 - Fault troubleshooting. Further, for knowledge points involving high - risk operations (such as live wiring), safety specification nodes are forcibly inserted. In this embodiment, it is determined whether it is a high - risk operation by the preset operation risk value of each knowledge point.

[0097] Determine the mastery degree of each knowledge node in the sub - graph according to the ability value. For example, if the ability value of this knowledge point is 30%, it is converted into a non - linear weight through the Sigmoid function, so as to obtain the mastery degree of this knowledge node:

[0098]

[0099] In the formula, \(m_i\) represents the ability value of the \(i\) - th knowledge node.

[0100] Through the mastery degree, unmastered knowledge points can be preferentially recommended to avoid repeated training of mastered content.

[0101] According to the matching degree between the learning style and the resource type, calculate the resource type weights of each knowledge node. Specifically, assume that the resource types associated with a certain knowledge point include videos, texts, and test questions. Different initial weights are preset for different resource types. For example, video = 0.6, text = 0.3, test question = 0.1. Then, according to the rule table, determine the matching degree between the learning style and the resource type. For example, if the trainee's style is video-based, the matching degree of video resources is 1, and the matching degrees of other resource types are 0. Then, sum the products of the initial weights of all resource types and their corresponding matching degrees to obtain the overall matching degree between the learning resource type associated with the knowledge point and the trainee's learning style, and use it as the resource type weight. Through the resource type weight, resources suitable for the trainee's learning style can be recommended.

[0102] According to the similarity between the interest vector and the knowledge node, calculate the interest matching degree of each knowledge node in the learning subgraph. Preferably, the cosine similarity can be used to calculate the interest matching degree. Quantify the interest correlation through the cosine similarity between the trainee's interest vector and the knowledge point representation vector, and improve the trainee's participation by increasing the scores of preference-related knowledge points (such as "the principle of electricity theft"). Of course, other similarity calculation methods such as Euclidean distance can also be used, and no more restrictions are imposed here.

[0103] Then use the KgRank algorithm to fuse multi-dimensional features to calculate the dynamic weight values of each knowledge node:

[0104]

[0105] In the formula, n i represents the i-th knowledge node, Score′(n i ) represents the dynamic weight value of the i-th knowledge node, Similarity(n i ) represents the interest matching degree between the i-th knowledge node and the interest vector, mi represents the ability value of the i-th knowledge node, ω k represents the initial weight of the i-th knowledge node corresponding to the k-th resource type, S uk represents the matching degree between the learning style and the k-th resource type, and β, γ, δ all represent weight coefficients. The weight coefficients are preset values and can be set according to different job requirements.

[0106] After calculating the dynamic weight values of each knowledge node, sort them in descending order to obtain the learning path of the training trainees.

[0107] In a preferred embodiment, based on the above dynamic weight values, the present invention uses the PageRank values of each knowledge node to correct the dynamic weight values. The PageRank value is an index value obtained by calculating the importance of a web page through the PageRank algorithm, which reflects the logical importance of a knowledge point in the overall knowledge network. The specific calculation process can refer to the calculation process of the PageRank algorithm and will not be elaborated here. By correcting the dynamic weight values with the PageRank values, the logical coherence of the learning path is ensured, and important knowledge points (such as basic principles) can be preferentially recommended. The corrected dynamic weight value is expressed as:

[0108]

[0109] In the formula, n i represents the i-th knowledge node, Score(n i ) represents the corrected dynamic weight value of the i-th knowledge node, PR(n i ) represents the PageRank value of the i-th knowledge node, Similarity(n i ) represents the interest matching degree between the i-th knowledge node and the interest vector, mi represents the ability value of the i-th knowledge node, ω k represents the initial weight of the k-th resource type corresponding to the i-th knowledge node, S uk represents the matching degree between the learning style and the k-th resource type, and α, β, γ, and δ all represent weight coefficients.

[0110] It is assumed that the weight coefficients are preset as: α = 0.4, β = 0.3, γ = 0.2, δ = 0.1. In actual applications, they can be adjusted according to different scenarios. For example, for trainees in the electricity inspection post, the β value can be increased, and for high-risk operation knowledge points, the α value can be increased, etc.

[0111] The learning path generation method in this embodiment ensures the logical integrity, personalized adaptation, and adaptive dynamic optimization of the learning path through the collaborative analysis of multi-dimensional parameters, and ensures safety and compliance, realizing efficient, safe, and personalized path recommendation.

[0112] When the trainee is learning, the trainee conducts AR interactive learning through the AR glasses, scans the QR code on the device through the AR glasses to take the training course of the power metering device, and at the same time calls the trainee's current latest learning path to generate an AR interactive course, and supports multi-modal interactions such as gestures and voices to achieve AR learning by superimposing three-dimensional disassembly animations and virtual operation guides.

[0113] After the learning in the current stage is completed, the knowledge graph and the trainee portrait are updated according to the trainee's learning behavior, and the learning path is adjusted. The specific steps include:

[0114] Re - conduct knowledge and ability assessment, learning style analysis, and interest entity extraction for each practical training student to obtain the latest ability value, the latest set of knowledge weak points, the latest learning style, and the latest interest vector;

[0115] Taking the number of students as the unit, count each knowledge weak point in the latest set of knowledge weak points;

[0116] If the number of students corresponding to the knowledge weak point is greater than the quantity threshold, adjust the association weight of the predecessor knowledge point of the knowledge weak point in the knowledge graph;

[0117] Adjust the student portrait according to the latest ability value, the latest set of knowledge weak points, the latest learning style, and the latest interest vector of each practical training student, and regenerate the corresponding learning path according to the adjusted student portrait and the adjusted knowledge graph.

[0118] In this embodiment, each practical training student re - conducts knowledge and ability assessment, learning style analysis, and interest entity extraction, and updates the ability value, the set of knowledge weak points, the learning style, and the interest vector, so as to realize the update of the student portrait.

[0119] Then count the knowledge weak points of all students. If the number of students corresponding to a certain knowledge weak point exceeds the threshold, that is, a certain knowledge point belongs to the knowledge weak points of most students, then adjust the association weight of the predecessor knowledge point of this knowledge weak point in the knowledge graph. By increasing the association weight, the association reinforcement learning ability of this knowledge point is enhanced. Finally, according to the latest student portrait and the latest knowledge graph, regenerate the learning path, thus realizing the dynamic optimization and closed - loop update of the learning path. This closed - loop feedback mechanism realizes the self - optimization of the learning path and can continuously improve the training effect.

[0120] Iteratively optimize the learning path according to the above steps, and conduct AR - interactive training and learning for practical training students according to the learning path until the students master all knowledge points and complete the practical training course of power metering equipment.

[0121] A training method for power metering equipment provided in this embodiment. Through the three-dimensional visualization of AR interaction, the present invention improves the understanding efficiency of trainees for complex structures, ensures the safety of training, reduces equipment loss, accurately depicts the trainee portraits through personalized dynamic evaluation, improves the systematicness of training content by constructing a hierarchical knowledge network, and improves the learning efficiency of trainee through personalized learning path recommendation. Through the immersive interaction of AR technology and the structured reasoning of knowledge graph, combined with dynamic evaluation and path recommendation, the present invention effectively improves the safety, efficiency and personalization level of power metering equipment training.

[0122] Please refer to Figure 2 , based on the same inventive concept, a training system for power metering equipment proposed in the second embodiment of the present invention includes:

[0123] A trainee portrait construction module 10, configured to perform knowledge and ability evaluation, learning style analysis, and interest entity extraction on each trainee, to obtain the ability value, knowledge weakness set, learning style, and interest vector of the trainee;

[0124] Construct a trainee portrait corresponding to the trainee according to the ability value, the knowledge weakness set, the learning style, and the interest vector;

[0125] A learning path generation module 20, configured to extract a learning sub-graph from a pre-constructed power metering equipment knowledge graph according to the trainee portrait, use the KgRank algorithm to sort the knowledge nodes of the learning sub-graph, and generate a learning path for the trainee;

[0126] An AR interaction learning module 30, configured to generate AR interaction-based learning content according to the learning path, so that the trainee can perform interactive learning on the learning content;

[0127] A learning path update module 40, configured to respond to the completion of the interactive learning, update the knowledge graph and the trainee portrait according to the learning behavior of the trainee, and adjust the learning path until the power metering equipment training ends.

[0128] The technical features and technical effects of the power metering equipment training system proposed in the embodiment of the present invention are the same as those of the method proposed in the embodiment of the present invention, and will not be elaborated here. Each module in the above power metering equipment training system can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0129] In addition, an embodiment of the present invention further provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the above method are implemented.

[0130] Please refer to Figure 3 , the internal structure diagram of a computer device in an embodiment. The computer device may specifically be a terminal or a server. The computer device includes a processor, a memory, a network interface, a display, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the power metering device training method is implemented. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device may be a touch layer covered on the display screen, or a button, a trackball, or a touchpad provided on the outer shell of the computer device, or an external keyboard, a touchpad, or a mouse, etc.

[0131] Those of ordinary skill in the art can understand that Figure 3 the structure shown in

[0132] is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computing device may include more or fewer components than those shown in the figure, or combine some components, or have the same component arrangement.

[0133] In summary, a training method, system, computer device, and medium for power metering equipment proposed in the embodiments of the present invention. The method obtains the ability value, set of knowledge weak points, learning style, and interest vector of the training trainee by conducting knowledge and ability assessment, learning style analysis, and interest entity extraction for each training trainee; constructs a trainee portrait corresponding to the training trainee according to the ability value, the set of knowledge weak points, the learning style, and the interest vector; extracts a learning sub-graph from a pre-constructed power metering equipment knowledge graph according to the trainee portrait, and uses the KgRank algorithm to sort the knowledge nodes of the learning sub-graph to generate a learning path for the training trainee; generates AR interaction-based learning content according to the learning path to enable the training trainee to conduct interactive learning on the learning content; in response to completing the interactive learning, updates the knowledge graph and the trainee portrait according to the learning behavior of the training trainee, and adjusts the learning path until the power metering equipment training ends. The present invention improves the understanding efficiency of trainees for complex structures through the three-dimensional visualization of AR interaction, ensures the safety of training, reduces equipment loss, accurately depicts the trainee portrait through personalized dynamic assessment, improves the systematicness of training content by constructing a hierarchical knowledge network, and improves the learning efficiency of training trainees through personalized learning path recommendation. Through the immersive interaction of AR technology and the structured reasoning of the knowledge graph, combined with dynamic assessment and path recommendation, the present invention effectively improves the safety, efficiency, and personalization level of power metering equipment training.

[0134] Each embodiment in this specification is described in a progressive manner. For parts that are the same or similar in each embodiment, they can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0135] The above-described embodiments only represent several preferred embodiments of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be pointed out that for those of ordinary skill in the art in this technical field, without departing from the technical principle of the present invention, several improvements and substitutions can still be made, and these improvements and substitutions should also be regarded as the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the protection scope of the claims.

Claims

1. A training method for power metering equipment, characterized in that, Including: Conduct knowledge and ability assessment, learning style analysis, and interest entity extraction for each training trainee to obtain the ability value, knowledge weak point set, learning style, and interest vector of the training trainee; Construct a trainee portrait corresponding to the training trainee according to the ability value, the knowledge weak point set, the learning style, and the interest vector; Extract a learning subgraph from a pre-constructed power metering equipment knowledge graph according to the trainee portrait, and use the KgRank algorithm to sort the knowledge nodes of the learning subgraph to generate a learning path for the training trainee; Generate AR interaction-based learning content according to the learning path, so that the training trainee can interactively learn the learning content; In response to completing the interactive learning, update the knowledge graph and the trainee portrait according to the learning behavior of the training trainee, and adjust the learning path until the power metering equipment training ends.

2. The training method of the power metering device according to claim 1, characterized in that The step of conducting knowledge and ability assessment, learning style analysis, and interest entity extraction for each training trainee to obtain the ability value, knowledge weak point set, learning style, and interest vector of the training trainee includes: Judge whether the training trainee is undergoing an initial assessment. If so, randomly generate initial questions according to the registration information of the training trainee and conduct an initial assessment on the training trainee. If not, generate personalized questions according to the previous assessment results and conduct a dynamic assessment on the training trainee; Obtain the ability value of the training trainee according to the correct rate of the dynamic assessment, and use the knowledge points that fail the assessment as knowledge weak points to form a knowledge weak point set; Obtain the explicit learning style of the training trainee according to a preset scale, and correct the explicit learning style according to the learning behavior of the training trainee to obtain the learning style of the training trainee; Use a preset interest recognition model to extract interest entities from the learning behavior log of the training trainee to obtain the interest vector of the training trainee, and the interest recognition model is constructed based on a bidirectional long short-term memory neural network and a random conditional field.

3. The power metering equipment training method according to claim 1, characterized in that, The step of extracting a learning subgraph from a pre-constructed power metering equipment knowledge graph according to the trainee portrait, using the KgRank algorithm to sort the knowledge nodes of the learning subgraph, and generating a learning path for the training trainee includes: Extract a learning subgraph from the power metering equipment knowledge graph according to the knowledge weak point set; Determine the mastery degree of each knowledge node in the learning subgraph according to the ability value; Calculate the resource type weight of each knowledge node according to the matching degree between the learning style and the resource type; Calculate the interest matching degree of each knowledge node according to the similarity between the interest vector and the knowledge node; Calculate the dynamic weight value of each knowledge node using the KgRank algorithm according to the mastery degree, the resource type weight, and the interest matching degree; Sort the knowledge nodes of the learning subgraph according to the dynamic weight value to obtain a learning path.

4. The power metering equipment training method according to claim 3, characterized in that, After the step of calculating the dynamic weight value of each knowledge node using the KgRank algorithm according to the mastery degree, the resource type weight, and the interest matching degree, it further includes: According to the PageRank values of each knowledge node, correct the dynamic weight values to obtain the corrected dynamic weight values; The corrected dynamic weight values are represented by the following formula: Where n i represents the i-th knowledge node, Score(n i ) represents the revised dynamic weight value of the i-th knowledge node, PR(n i ) represents the PageRank value of the i-th knowledge node, Similarity(n i ) represents the interest matching degree between the i-th knowledge node and the interest vector, mi represents the ability value of the i-th knowledge node, ω k represents the initial weight of the k-th resource type corresponding to the i-th knowledge node, S uk represents the matching degree between the learning style and the k-th resource type, and α, β, γ, δ all represent weight coefficients.

5. The training method for power metering equipment according to claim 1, characterized in that The step of updating the knowledge graph and the trainee portrait according to the learning behaviors of the trainees in practice and adjusting the learning path includes: Conduct knowledge ability assessment, learning style analysis, and interest entity extraction for each trainee in practice again to obtain the latest ability value, the latest set of knowledge weak points, the latest learning style, and the latest interest vector; Taking the number of trainees as a unit, count each knowledge weak point in the latest set of knowledge weak points; If the number of trainees corresponding to the knowledge weak point is greater than the number threshold, adjust the association weight of the predecessor knowledge point of the knowledge weak point in the knowledge graph; According to the latest ability value, the latest set of knowledge weak points, the latest learning style, and the latest interest vector of each trainee in practice, adjust the trainee portrait, and regenerate the corresponding learning path according to the adjusted trainee portrait and the adjusted knowledge graph.

6. The power metering equipment training method according to claim 1, characterized in that, The construction steps of the power metering equipment knowledge graph include: Obtain the knowledge data of power metering equipment from the professional model library and the learning database, and perform data preprocessing on the knowledge data to obtain the cleaned structured data set; Define entities as knowledge points, questions, trainees, and majors, and define relationships as hierarchical relationships and mapping relationships, and perform knowledge extraction on the structured data set to obtain the power metering equipment knowledge graph.

7. The power metering equipment training method according to claim 6, characterized in that The step of performing knowledge extraction on the structured data set to obtain the power metering equipment knowledge graph includes: Extract entity relationship multi-tuples from the structured data set to obtain explicit knowledge; Through a preset knowledge extraction model, extract the potential associations between entities from historical fault cases, and infer the logic between knowledge points through a rule engine to obtain implicit knowledge. The knowledge extraction model is constructed based on a bidirectional long short-term memory neural network; Perform knowledge fusion on the explicit knowledge and the implicit knowledge to obtain the power metering equipment knowledge graph.

8. A training system for power metering equipment, characterized in that, Include: A trainee portrait construction module, which is used to conduct knowledge ability assessment, learning style analysis, and interest entity extraction for each trainee in practice to obtain the ability value, the set of knowledge weak points, the learning style, and the interest vector of the trainee in practice; Construct a trainee portrait corresponding to the trainee in practice according to the ability value, the set of knowledge weak points, the learning style, and the interest vector; A learning path generation module, which is used to extract a learning subgraph from the pre-constructed power metering equipment knowledge graph according to the trainee portrait, and use the KgRank algorithm to sort the knowledge nodes of the learning subgraph to generate the learning path of the trainee in practice; An AR interactive learning module, which is used to generate AR interactive learning content according to the learning path, so that the trainee in practice can perform interactive learning on the learning content; A learning path update module, configured to, in response to completion of the interactive learning, update the knowledge graph and the trainee portrait according to the learning behavior of the training trainee, and adjust the learning path until the training of the power metering equipment ends.

9. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 7.

Images