A high-voltage test intelligent guidance and capability portrait generation method and related device

Abstract

This invention discloses a method and related apparatus for intelligent guidance and capability profile generation in high-voltage testing, relating to the field of power testing technology. The method includes: collecting high-voltage testing knowledge data; breaking down the categorized knowledge data into structured knowledge units; dividing the high-voltage testing operation into several operation stages to construct a dynamic operation context; calculating the relevance between the operation context and each knowledge unit to filter target knowledge units to generate a hierarchical prompting strategy and provide real-time guidance feedback to trainees; generating a capability profile containing several capability dimensions based on trainee operation behavior records and data analysis path records; comparing the scores of each capability dimension with corresponding thresholds to identify weak capabilities; identifying the causes of weaknesses based on the error type distribution and prompt usage associated with the weak capabilities; and recommending subsequent training tasks. This invention solves the problems of existing technologies, such as single capability evaluation dimensions, lack of personalized recommendations, and difficulty in solidifying expert experience.

Description

Technical Field

[0001] This invention relates to the field of power testing technology, and in particular to a method and device for intelligent guidance and capability profile generation in high-voltage testing. Background Technology

[0002] High-voltage testing is a crucial technical means for evaluating the insulation condition, diagnosing defects, and ensuring the safe operation of power equipment. Operators not only need to be proficient in instrument operation and standardized wiring procedures, but also must possess comprehensive abilities such as on-site risk identification, analysis of abnormal test data, reasoning about defect causes, and decision-making regarding subsequent handling. However, these advanced judgment abilities heavily rely on the long-term experience accumulated by senior engineers. For example, when the dielectric loss factor increases, it is necessary to conduct multi-source cross-verification by comprehensively considering wiring shielding, environmental humidity, procedural limits, historical trends, and parameters of similar equipment, rather than simply determining an insulation defect. Currently, the skill improvement of high-voltage testing personnel mainly relies on on-site learning through mentorship programs, intensive theoretical lectures, fixed-process virtual simulation systems, or general online training platforms.

[0003] While these methods have improved basic operational skills to some extent, they still have significant shortcomings in areas such as intelligent guidance, process evaluation, and personalized training. However, existing technologies still suffer from the following main problems: 1. Intelligent guidance lacks contextual awareness and deep data analysis and reasoning capabilities. Specifically, existing fixed-process virtual simulation systems focus on reproducing preset operational steps and cannot dynamically identify the specific operational stage and corresponding risk level of the trainee. Simultaneously, the introduced general AI question-and-answer system is detached from the specific stage constraints and triggering conditions of high-voltage testing, resulting in generalized and ambiguous prompts. These deficiencies prevent existing technologies from guiding trainees to perform deep reasoning according to the correct analytical path of "eliminating experimental factors—comparing multi-source data—judging the cause of the defect" when abnormal experimental data occurs (such as increased dielectric loss or decreased insulation resistance). This makes it difficult to achieve real-time professional guidance deeply integrated with the operational scenario. 2. Coarse-grained recording of operational processes leads to a single dimension of competency evaluation. Specifically, existing training systems typically collect trainee operational behavior data only for final assessment, resulting in coarse-grained recording that cannot be refined to the operational stage, specific operational object, error type, risk level, or the trainee's reliance on prompts. Due to a lack of granular behavioral data, competency assessments often only output a general overall score, failing to differentiate trainees' specific weaknesses in different professional dimensions such as instrument selection, wiring compliance, safety risk identification, test data analysis, and defect cause judgment. The assessment results lack specificity and guidance. Third, standardized training content leads to a lack of personalized recommendations and makes it difficult to solidify expert experience. Specifically, existing training systems use uniform standard operating procedures and training tasks for all trainees, making it difficult to automatically adjust subsequent training content based on trainees' different weaknesses and their causes, resulting in insufficient training focus and low efficiency. At the same time, the implicit analytical logic of "data anomaly → on-site troubleshooting → multi-source data comparison → defect characterization," accumulated by senior high-voltage test engineers over decades of work, lacks structured expression and storage methods, making it difficult to scale up and systematically reuse, further restricting the efficiency and quality of highly skilled personnel training. Summary of the Invention

[0004] This invention provides a method and related apparatus for intelligent guidance and capability profile generation in high-pressure testing, which addresses the problems of existing technologies lacking operational context awareness and data analysis and reasoning capabilities, coarse-grained operation records leading to a single dimension of capability evaluation, and standardized training content resulting in a lack of personalized recommendations and difficulty in solidifying expert experience.

[0005] In view of this, the first aspect of the present invention provides a method for intelligent guidance and capability profile generation in high-pressure testing, the method comprising:

[0006] Collect and classify high-pressure test knowledge data, and then break down the classified knowledge data into structured knowledge units;

[0007] The high-pressure test operation is divided into several operation stages. The operation stage, operation behavior and test data of the trainees during the current operation are collected in real time. A dynamic operation context is constructed in combination with the trainees' ability level.

[0008] The relevance between the task context and each knowledge unit is calculated to filter target knowledge units. A tiered prompting strategy is determined based on the target knowledge unit, the ability level, the current training mode, and the current risk level. Real-time guidance and feedback are then provided to the trainee according to the tiered prompting strategy.

[0009] Collect student's operational behavior records and data analysis path records throughout the entire assignment process, and generate a competency profile containing several competency dimensions based on the operational behavior records and the data analysis path records;

[0010] The scores of each capability dimension are compared with the corresponding thresholds to identify weak capabilities. The reasons for the weakness are identified based on the error type distribution and prompt usage associated with the weak capability. Subsequent training tasks are matched and recommended from the training task library based on the weak capability and the reasons for the weakness.

[0011] Optionally, the step of collecting and classifying high-pressure test knowledge data, and then breaking down the classified knowledge data into structured knowledge units, includes:

[0012] Collect high-pressure testing knowledge data, which includes at least one of the following: regulations and standards, work instructions, instrument data, manufacturer data, historical data, defect cases, expert experience, and risk rules;

[0013] Add source tags, scope of application tags, credibility tags, and update time tags to each of the knowledge materials to obtain the classified and labeled knowledge materials;

[0014] The categorized and labeled knowledge data is broken down into knowledge units. Each knowledge unit includes at least the applicable equipment, applicable test items, applicable operation stages, triggering conditions, knowledge content, risk level, judgment basis, recommended actions, and knowledge source.

[0015] Optionally, the high-voltage testing operation is divided into several operation stages, and the operation stage, operational behavior, and test data of the trainee during the current operation are collected in real time. A dynamic operation context is constructed based on the trainee's ability level, including:

[0016] After the high-pressure test operation is pre-divided into several operation stages, the operation stage, operation behavior and test data of the trainee in the current operation process are collected in real time, and the trainee's ability level is determined based on the trainee's historical operation records.

[0017] A dynamic job context is constructed based on the current equipment type, current test project, job stage, operation behavior, test data, environmental parameters, historical data retrieval status, and capability level.

[0018] Optionally, calculating the relevance between the job context and each of the knowledge units to filter target knowledge units includes:

[0019] Retrieve the current equipment type, current test item, current job stage, current operation behavior, and current test data from the current job context;

[0020] For each of the knowledge units, calculate the following: a first relevance sub-item matching the current device type; a second relevance sub-item matching the current test project; a third relevance sub-item matching the current operation stage; a fourth relevance sub-item matching whether the current operation behavior meets the triggering condition; a fifth relevance sub-item matching whether the current test data meets the triggering condition; and a sixth relevance sub-item matching the credibility level of the knowledge unit.

[0021] The first to sixth relevance sub-items are weighted and summed to obtain the relevance between the job context and each knowledge unit.

[0022] The knowledge units are sorted according to their relevance, and target knowledge units are selected from the knowledge units based on the sorting results.

[0023] Optionally, the step of determining a tiered prompting strategy based on the target knowledge unit, the ability level, the current training mode, and the current risk level, and providing real-time guidance and feedback to the trainee according to the tiered prompting strategy, includes:

[0024] Obtain the knowledge content of the target knowledge unit, and obtain the trainee's ability level, current training mode, and current risk level;

[0025] Based on the ability level, the current training mode, and the current risk level, and in conjunction with the knowledge content, a hierarchical prompting strategy is generated. The hierarchical prompting strategy includes a prompting method that matches the current scenario and a prompting message to be output generated based on the knowledge content.

[0026] Based on the hierarchical prompting strategy, the feedback method is determined according to the current risk level and the current interactive terminal type, and the prompting information to be output is output according to the feedback method.

[0027] Optionally, the process of collecting trainees' operational behavior records and data analysis path records throughout the entire assignment, and generating a competency profile containing several competency dimensions based on the operational behavior records and the data analysis path records, includes:

[0028] Collect records of trainees' operational behaviors throughout the entire assignment process. These records must include at least the operational results, error types, and risk levels.

[0029] Collect data on the completion status of necessary analysis steps by trainees during the data analysis phase, and calculate the data analysis path completeness based on the number of necessary analysis steps completed.

[0030] Based on the recorded operation behavior and the completeness of the data analysis path, scores for each dimension are calculated according to several preset capability dimensions to generate a capability profile of the trainee.

[0031] Optionally, the step of comparing the scores of each of the capability dimensions with corresponding thresholds to identify weak capability items, identifying the causes of weakness based on the error type distribution and prompt usage associated with the weak capability items, and matching and recommending subsequent training tasks from the training task library based on the weak capability items and the causes of weakness includes:

[0032] The scores of each capability dimension are compared with the corresponding pass thresholds, and capability dimensions with scores below the pass thresholds are identified as weak capability items.

[0033] Obtain the error type distribution and prompt usage frequency associated with the weak capability item; when the frequency of occurrence of the same error type exceeds a frequency threshold within a preset time period, the error type is identified as a weak cause; and / or, when the prompt usage frequency exceeds a prompt frequency threshold, prompt dependency is identified as a weak cause.

[0034] Based on the weak skills, the reasons for the weakness, and the trainee's historical training records, the matching degree between each training task in the training task library and the trainee is calculated. Subsequent training tasks are recommended based on the matching degree. Each training task in the training task library includes applicable skill dimensions, applicable equipment, applicable test items, difficulty level, training objectives, typical error points, and recommended prerequisites.

[0035] Optionally, the step of matching and recommending subsequent training tasks from the training task library based on the weak capability items and the reasons for the weaknesses further includes:

[0036] Record the trainees' ability profiles after multiple training sessions, and calculate the improvement in each ability dimension based on the ability profiles after each training session;

[0037] When the improvement in any of the aforementioned capability dimensions is less than a preset threshold, the difficulty level of the training task is reduced or basic knowledge prompts are added.

[0038] A second aspect of the present invention provides a high-pressure test intelligent guidance and capability profile generation system, the system comprising:

[0039] The data acquisition unit is used to collect and classify high-pressure test knowledge data, and to break down the classified knowledge data into structured knowledge units.

[0040] The construction unit is used to divide the high-voltage test operation into several operation stages, collect the operation stage, operation behavior and test data of the trainee in the current operation process in real time, and construct a dynamic operation context in combination with the trainee's ability level.

[0041] The calculation unit is used to calculate the relevance between the task context and each knowledge unit to filter target knowledge units, determine a hierarchical prompting strategy based on the target knowledge unit, the ability level, the current training mode and the current risk level, and provide real-time guidance and feedback to the trainee according to the hierarchical prompting strategy.

[0042] The generation unit is used to collect the student's operation behavior records and data analysis path records throughout the entire assignment process, and generate a capability profile containing several capability dimensions based on the operation behavior records and data analysis path records.

[0043] The matching unit is used to compare the scores of each capability dimension with the corresponding thresholds to identify weak capability items, identify the causes of weakness based on the error type distribution and prompt usage associated with the weak capability items, and match and recommend subsequent training tasks from the training task library based on the weak capability items and the causes of weakness.

[0044] A third aspect of the present invention provides a high-pressure test intelligent guidance and capability profile generation device, the device comprising a processor and a memory:

[0045] The memory is used to store program code and transmit the program code to the processor;

[0046] The processor is used to execute the steps of the high-pressure test intelligent guidance and capability profile generation method as described in the first aspect above, according to the instructions in the program code.

[0047] As can be seen from the above technical solutions, the present invention has the following advantages:

[0048] This invention provides a method for intelligent guidance and capability profile generation in high-pressure testing. First, it collects and categorizes high-pressure testing knowledge data, breaking it down into structured knowledge units. This transforms scattered procedures, manufacturer information, historical cases, and expert experience into computable and matchable knowledge units, solving the problem of the difficulty in solidifying expert experience and distinguishing it from the generalized answers of ordinary AI question-answering systems. Then, by dividing the high-pressure testing operation into several stages, it collects the trainee's current stage, operational behavior, and test data in real time. A dynamic operational context is constructed based on the trainee's capability level, and the correlation between the operational context and each knowledge unit is calculated to filter target knowledge units. A tiered prompting strategy is determined based on the target knowledge unit, capability level, current training mode, and current risk level. Real-time guidance and feedback are provided to the trainee according to this strategy, enabling dynamic perception of the trainee's current specific operational stage and corresponding risk level. When test data is abnormal, the system guides the trainee to perform in-depth reasoning according to the correct analytical path. It also achieves tiered prompts based on trainee level, training mode, and risk level, thus solving the problem of lack of context awareness and in-depth data analysis and reasoning capabilities in intelligent guidance, distinguishing it from existing systems with fixed-step prompts.

[0049] Next, by collecting operational behavior records and data analysis path records of trainees throughout the entire assignment process, a competency profile containing several capability dimensions is generated based on these records. This allows for accurate identification of trainees' specific competency weaknesses in each professional dimension based on fine-grained, end-to-end behavioral data, thus solving the problem of limited competency evaluation dimensions caused by coarse-grained operational records, and differing from existing evaluation methods that only output a general total score. Finally, by comparing the scores of each competency dimension with corresponding thresholds to identify weak competencies, the causes of weaknesses are identified based on the distribution of error types associated with weak competencies and the usage of prompts. Based on the weak competencies and their causes, subsequent training tasks are matched and recommended from the training task library. This allows for automatic adjustment of subsequent training content based on different trainees' weak competencies and their causes, thus solving the problem of lack of personalized recommendations due to standardized training content, and differing from existing methods that push uniform training content to all trainees.

[0050] In summary, this invention solves the problems of existing technologies, such as lack of job context awareness and data analysis reasoning capabilities, coarse-grained operation records leading to a single dimension of capability evaluation, and standardized training content resulting in a lack of personalized recommendations and difficulty in solidifying expert experience. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A schematic diagram illustrating the process of obtaining a high-pressure test intelligent guidance and capability profile generation method provided in an embodiment of the present invention;

[0053] Figure 2 The diagram shows the structure of a high-pressure test intelligent guidance and capability profile generation system provided in an embodiment of the present invention. Detailed Implementation

[0054] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0055] Please see Figure 1 The present invention provides a method for intelligent guidance and capability profile generation in high-pressure testing, comprising:

[0056] Step 101: Collect and classify high-pressure test knowledge data, and break down the classified knowledge data into structured knowledge units;

[0057] Step 102: Divide the high-voltage test operation into several operation stages, collect the operation stage, operation behavior and test data of the trainee in the current operation process in real time, and construct a dynamic operation context in combination with the trainee's ability level;

[0058] Step 103: Calculate the relevance between the task context and each knowledge unit to filter target knowledge units, determine a hierarchical prompting strategy based on the target knowledge unit, the ability level, the current training mode and the current risk level, and provide real-time guidance and feedback to the trainee according to the hierarchical prompting strategy.

[0059] Step 104: Collect the student's operation behavior records and data analysis path records throughout the entire assignment process, and generate a capability profile containing several capability dimensions based on the operation behavior records and the data analysis path records;

[0060] Step 105: Compare the scores of each capability dimension with the corresponding thresholds to identify weak capability items, identify the reasons for the weakness based on the error type distribution and prompt usage associated with the weak capability items, and match and recommend subsequent training tasks from the training task library based on the weak capability items and the reasons for the weakness.

[0061] In one embodiment, step 101 involves collecting and classifying high-pressure test knowledge data, and then breaking down the classified knowledge data into structured knowledge units, including:

[0062] Step 1011: Collect high-pressure testing knowledge data, which includes at least one of the following: regulations and standards, work instructions, instrument data, manufacturer data, historical data, defect cases, expert experience, and risk rules;

[0063] Specifically, the first step is to collect high-voltage testing knowledge data, which includes at least one of the following: regulations and standards, work instructions, instrument data, manufacturer data, historical data, defect cases, expert experience, and risk rules. For example, the "High-Voltage Testing Regulations" can be collected as regulations and standards; the instruction manual for the dielectric loss tester as instrument data; the main transformer's factory test report as manufacturer data; data from previous preventative tests as historical data; cases of insulation dampness as defect cases; the experience of experienced workers regarding "increased dielectric loss requires checking wiring shielding" as expert experience; and the safety rule regarding "leaky grounding" as risk rules. After collection, this knowledge data constitutes the original knowledge data set D. raw :

[0064] ;

[0065] Among them, d1 to d n This represents the first to nth original knowledge data. The original knowledge data is then categorized, including categories such as procedures and standards, work instructions, instrument data, manufacturer data, historical data, defect cases, expert experience, and risk rules.

[0066] This step categorizes the original knowledge data, enabling subsequent knowledge unit decomposition to proceed in an orderly manner according to data type, thus avoiding difficulties in retrieval and matching caused by the mixing of knowledge from different sources and types.

[0067] Step 1012: Add source tags, scope of application tags, credibility tags and update time tags to each of the knowledge materials to obtain the classified and labeled knowledge materials;

[0068] Specifically, firstly, source tags, scope of application tags, credibility tags, and update time tags are added to each piece of knowledge material to obtain categorized and labeled knowledge materials. For example, a certain article in the "High Voltage Test Procedure" is tagged with "DL / T 596 Procedure," "Main Transformer Dielectric Loss Test," "Mandatory Procedure," and "2020"; a text on the experience of a senior technician is tagged with "Expert from a Provincial Electric Power Research Institute," "Main Transformer Dielectric Loss Test Data Analysis," "Expert Experience," and "2022"; a defect case in a historical test report is tagged with "220kV Substation Handover Test Report," "Bushing Dielectric Loss Test," "Case Reference," and "2021." The categorized and labeled knowledge materials are represented as follows:

[0069] ;

[0070] Among them, type i Indicates the data type, source i Indicates the source of information, scope i Indicates the scope of application, time i Indicates the time or version of the data, level i The reliability level of the data is indicated. The categorized and labeled knowledge data is then stored in the high-pressure testing knowledge base.

[0071] This step adds multi-dimensional tags to each knowledge resource, giving the knowledge unit clear traceability and applicable boundaries.

[0072] Step 1013: Decompose the classified and labeled knowledge data into knowledge units. Each knowledge unit includes at least the applicable equipment, applicable test items, applicable operation stages, triggering conditions, knowledge content, risk level, judgment basis, suggested actions, and knowledge source.

[0073] Specifically, the categorized and labeled knowledge data is first broken down into knowledge units. Each knowledge unit includes at least the applicable equipment, applicable test items, applicable operation stage, triggering conditions, knowledge content, risk level, judgment basis, recommended actions, and knowledge source. For example, for the scenario of analyzing dielectric loss test data of main transformers, a knowledge unit is extracted from the categorized and labeled regulations and expert experience: the applicable equipment is set as the main transformer, the applicable test item is set as the dielectric loss test, the applicable operation stage is set as the data analysis stage, the triggering condition is set as an increase in dielectric loss value and a decrease in insulation resistance, the knowledge content is set as the possible presence of insulation dampness, insulation degradation, or influence from test wiring, the risk level is set as abnormal, the judgment basis is set as regulations limits, historical trends, and comparison with similar equipment, the recommended actions are set as rechecking wiring and shielding, checking environmental humidity, comparing historical data and retesting if necessary, and the knowledge source is set as the DL / T 596 regulations and the experience of experts from a provincial power research institute. For example, in the scenario of the DC test completion, another knowledge unit is extracted: the applicable equipment is set as high-voltage electrical equipment, the applicable test item is set as DC high-voltage test, the applicable operation stage is set as test completion and discharge stage, the triggering condition is set as incomplete discharge, the knowledge content is set as possible residual charge on the tested equipment, the risk level is set as severe, the judgment basis is set as safety procedures, the recommended action is set as using a discharge rod to fully discharge and confirm grounding, and the knowledge source is set as on-site safety procedures. Each knowledge unit is represented as follows:

[0074]

[0075] Among them, O i Indicates the applicable equipment, P i Stage indicates the applicable test item. i Indicates the applicable operation phase, Condi indicates the triggering condition, and Content... i Represents knowledge content, Risk i Indicates the risk level; Basisi indicates the basis for the judgment; Action i Source indicates a suggested action. i Indicates the source of knowledge, Level i The knowledge credibility level is then indicated. The resulting knowledge units are then stored in the high-pressure test knowledge base.

[0076] This step breaks down knowledge data into standardized knowledge units containing the aforementioned fields, enabling the precise retrieval of corresponding knowledge content, risk levels, and suggested actions based on the current work scenario. This achieves real-time intelligent guidance that matches the work stage and triggering conditions, avoiding providing generalized or irrelevant prompts.

[0077] In one embodiment, step 102 involves dividing the high-voltage test operation into several operation stages, collecting the trainee's current operation stage, operational behavior, and test data in real time, and constructing a dynamic operation context based on the trainee's ability level, including:

[0078] Step 1021: After pre-dividing the high-voltage test operation into several operation stages, collect the operation stage, operation behavior and test data of the trainee in the current operation process in real time, and determine the trainee's ability level based on the trainee's historical operation records;

[0079] Specifically, the high-voltage testing operation is first pre-divided into several stages, including task selection, test preparation, instrument selection, wiring operation, safety confirmation, parameter setting, test operation, data reading, data analysis, defect identification, test completion and discharge, and debriefing and evaluation. Then, the trainee's current work stage, operational behavior, and test data are collected in real time, and the trainee's competency level is determined based on their historical operation records. For example, if a trainee is currently in the data analysis phase of a dielectric loss test, real-time data is collected showing that the trainee is currently viewing test data, their current action is "directly submitting defect judgment," and the current test data is "dielectric loss value 0.85% (standard limit 0.8%), insulation resistance decreased." Based on the trainee's historical operation records (such as historical training times, accuracy rate of operation at each stage, error type distribution, etc.), the trainee's ability level is determined to be intermediate. As another example, if a trainee is currently in the wiring operation phase, real-time data is collected showing that the current work stage is wiring, the current action is "connecting the high-voltage output terminal of the dielectric loss tester to the tested equipment," and the current test data is empty. Based on the trainee's historical operation records, the trainee's ability level is determined to be new employee. This is based on the current work stage identification function. Automatically identify the current work stage;

[0080] Stage t Indicates the current stage of the operation; This represents the sequence of operations from the start to time t. This represents the current interactive interface or task page; h() represents the task stage identification function. This formula indicates that the stage is not manually specified, but rather automatically identified based on the student's operation sequence and the current interface.

[0081] This step involves pre-dividing the task into stages and collecting the student's task stage, operational behavior, and test data in real time. This enables the dynamic perception of the student's current task scenario and operational status, providing real-time data input for subsequent task context construction and accurate knowledge matching.

[0082] Step 1022: Construct a dynamic job context based on the current equipment type, current test item, job stage, operation behavior, test data, environmental parameters, historical data retrieval status, and capability level.

[0083] Specifically, a dynamic work context is constructed based on the current equipment type, current test item, work stage, operational behavior, test data, environmental parameters, historical data retrieval status, and capability level. For example, if the current equipment type is a main transformer, the current test item is a dielectric loss test, the current work stage is the data analysis stage, the current operational behavior is "directly submit defect judgment," the current test data is "dielectric loss value 0.85%, insulation resistance decrease," the environmental parameters are "ambient temperature 28℃, humidity 65%," the historical data retrieval status is "historical trends not viewed," and the trainee's capability level is intermediate, all of this information constitutes the dynamic work context at that moment.

[0084] ;

[0085] Where O represents the equipment type, P represents the test item, and Stage t A represents the operation phase at time t. t X represents the student's action at time t. t E represents the current experimental data. t H represents environmental parameters. t Indicates the status of historical data retrieval, U t This indicates the student's ability level.

[0086] This step constructs a dynamic job context containing multi-dimensional information, enabling the subsequent knowledge matching module to perform relevance calculations based on complete current scenario information, thereby improving the accuracy and relevance of knowledge matching.

[0087] In this embodiment, job context C t It updates dynamically based on student actions. For example:

[0088] After the trainee selects the main transformer dielectric loss test, the equipment type O and the test item P are determined;

[0089] After the trainee enters the wiring interface, the work stage begins. t Updated to the wiring operation stage;

[0090] When the trainee fails to confirm grounding, the operation behavior is A. t The record indicates that security confirmation is missing.

[0091] After the trainee reads the dielectric loss data, the experimental data X t Updated to current test data;

[0092] When multiple requests for prompts are detected from a student, ability level U t The update now indicates a high dependency level.

[0093] The dynamic update mechanism described above can reflect the latest assignment status of trainees in real time, ensuring that the context on which subsequent knowledge matching and prompt generation are based is always the current valid state.

[0094] In one embodiment, step 103 involves calculating the relevance between the task context and each knowledge unit to filter target knowledge units, determining a tiered prompting strategy based on the target knowledge units, the ability level, the current training mode, and the current risk level, and providing real-time guidance and feedback to the learner according to the tiered prompting strategy, including:

[0095] Step 1031: Obtain the current equipment type, current test item, current test stage, current operation behavior, and current test data from the current job context;

[0096] Specifically, it retrieves the current equipment type, current test item, current operation stage, current operation behavior, and current test data from the current job context. For example, the current equipment type is a main transformer, the current test item is a dielectric loss test, the current operation stage is the data analysis stage, the current operation behavior is "directly submit defect judgment," and the current test data is "dielectric loss value increases and insulation resistance decreases."

[0097] This step involves starting from the current job context. Extracting the above five types of information allows subsequent relevance calculations to be performed based on a unified current scenario input, avoiding matching deviations caused by missing or inconsistent information between different sub-steps.

[0098] Step 1032: For each of the knowledge units, calculate the first relevance sub-item that matches the current device type, the second relevance sub-item that matches the current test project, the third relevance sub-item that matches the current operation stage, the fourth relevance sub-item that matches whether the current operation behavior meets the triggering condition, the fifth relevance sub-item that matches whether the current test data meets the triggering condition, and the sixth relevance sub-item that matches the credibility level of the knowledge unit.

[0099] Specifically, for each knowledge unit, a first relevance sub-item matching the current device type, a second relevance sub-item matching the current test project, a third relevance sub-item matching the current operation stage, a fourth relevance sub-item matching whether the current operation behavior meets the triggering condition, a fifth relevance sub-item matching whether the current test data meets the triggering condition, and a sixth relevance sub-item matching the credibility level of the knowledge unit are calculated respectively. For example, the first relevance sub-item is determined based on whether the current equipment type "main transformer" matches the applicable equipment in the knowledge unit; the second relevance sub-item is determined based on whether the current test item "dielectric loss test" matches the applicable test item in the knowledge unit; the third relevance sub-item is determined based on whether the current operation stage "data analysis stage" matches the applicable operation stage in the knowledge unit; the fourth relevance sub-item is determined based on whether the current operation behavior "directly submit defect judgment" meets the triggering conditions in the knowledge unit (such as "dielectric loss value increases and multi-source comparison is not performed"); the fifth relevance sub-item is determined based on whether the current test data "dielectric loss value increases and insulation resistance decreases" meets the triggering conditions in the knowledge unit; and the sixth relevance sub-item is determined based on the credibility level of the knowledge unit (such as "mandatory procedures" being higher than "expert experience").

[0100] This step calculates relevance sub-items across six dimensions, enabling a comprehensive assessment of the matching degree between knowledge units and the current scenario from six perspectives: device, project, stage, operation, data, and credibility, thus avoiding matching biases caused by relying on only a single dimension.

[0101] Step 1033: Perform a weighted summation on the first to sixth relevance sub-items to obtain the relevance between the job context and each knowledge unit;

[0102] Specifically, the first to sixth relevance sub-items are weighted and summed to obtain the relevance between the task context and each knowledge unit. The relevance calculation is expressed as follows: ;

[0103] Among them, Rel(C t ,K i ) represents the current job context C t With knowledge unit K i Relevance; O indicates the current device type; O i Indicates the type of equipment to which the knowledge unit applies; P represents the current test item; P i Indicates the applicable test items for the knowledge unit; Stage t Indicates the current work stage; Stage i Indicates the applicable stage of the knowledge unit; A t Indicates the current user action; Cond iIndicates the triggering condition for a knowledge unit; X t Indicates current test data; Level i The represents the credibility level of the knowledge unit; Sim(⋅) represents the similarity or matching degree calculation function; (w1,w2,w3,w4,w5,w6) represents the weight of each matching factor. For example, the overall relevance is highest when the equipment matching, project matching, stage matching, operation behavior trigger condition matching, test data trigger condition matching, and the credibility level is "mandatory according to procedure".

[0104] This step involves weighted summation of the six relevance sub-items, which enables the generation of a unified relevance metric by integrating the matching results across multiple dimensions. This provides a comparable basis for subsequent sorting and filtering, and avoids difficulties in judging the overall matching degree due to inconsistencies in the matching results across different dimensions.

[0105] Step 1034: Sort each knowledge unit according to the relevance, and select target knowledge units from the knowledge units according to the sorting results;

[0106] Specifically, the knowledge units are ranked according to their relevance, and target knowledge units are selected from these units based on the ranking results. For example, after calculating the relevance of multiple knowledge units, they are ranked from highest to lowest relevance, and the top-ranked knowledge units are identified as target knowledge units. For instance, the knowledge unit with the highest relevance, "analysis of increased dielectric loss in main transformers," is selected as the target knowledge unit.

[0107] This step sorts and filters target knowledge units based on their comprehensive relevance, enabling the rapid location of the knowledge content that best matches the current work scenario from the high-pressure testing knowledge base. This avoids inaccurate matching or untimely retrieval due to the large amount of knowledge, thereby improving the professionalism and response speed of intelligent guidance.

[0108] Step 1035: Obtain the knowledge content of the target knowledge unit, and obtain the trainee's ability level, current training mode, and current risk level;

[0109] Specifically, the knowledge content of the target knowledge unit is obtained, along with the trainee's ability level, current training mode, and current risk level. For example, the knowledge content of the target knowledge unit is "There may be insulation dampness, insulation deterioration, or test wiring issues. It is necessary to review the wiring and shielding, check the environmental humidity, and make a comprehensive judgment after comparing historical data and data from similar equipment." The trainee's ability level is "intermediate," the current training mode is "learning mode," and the current risk level is "abnormal."

[0110] This step, by acquiring the above four types of information, enables the tiered prompting strategy to make comprehensive decisions based on the characteristics of the knowledge itself (i.e., the specific guidance information contained in the knowledge content), the trainee's personalized attributes (ability level, training mode), and the current risk status of the task, thus providing a data foundation for the subsequent generation of differentiated prompting strategies.

[0111] Step 1036: Based on the ability level, the current training mode, and the current risk level, and in conjunction with the knowledge content, generate a hierarchical prompting strategy. The hierarchical prompting strategy includes a prompting method that matches the current scenario and a prompting message to be output generated based on the knowledge content.

[0112] Specifically, even when relevant knowledge is matched, trainees shouldn't always be prompted in the same way. New employees need direct prompts, experienced employees are better suited to follow-up questions, answers shouldn't be given in real time during assessments, and the reasons for errors need to be explained in detail during debriefing. Therefore, it's necessary to transform the knowledge content into output prompts of different forms and depths based on the trainee's ability level, training mode, and risk level, and to select different prompting methods.

[0113] For example, for new employees in learning mode, a "direct prompt" strategy is generated to clearly inform trainees of the next steps, reducing the learning curve; for intermediate personnel in learning mode with a low risk level, a "heuristic questioning" strategy is generated to guide trainees to think independently without providing direct answers; when the current risk level is high, a "risk warning + step lock" strategy is generated to prevent trainees from continuing dangerous operations; when in assessment mode with low risk, a "silent recording" strategy is generated, recording only the process without providing prompts, thus not affecting the authenticity of the assessment; when in the data analysis phase, a "data analysis prompt" strategy is generated to guide comparison with procedures, manufacturers, historical data, and similar equipment data; when in debriefing mode, a "debriefing explanation" strategy is generated to explain the reasons for errors and improvement suggestions after training.

[0114] For example, the prompting strategy representation generated based on the current context, relevant knowledge units, learner's ability level, and training mode is as follows:

[0115] ;

[0116] in:

[0117]

[0118]

[0119]

[0120]

[0121] The hierarchical suggestion strategy satisfies the following rules:

[0122] ;

[0123] Where: Mode t Indicates the current prompt mode; L u Indicates the student's ability level; M t Indicates training mode; Risk t This indicates the current risk level; the severity threshold indicates the preset high-risk judgment criteria.

[0124] The rule states that answers are not always provided; instead, the method of prompting is chosen based on the learner's level and training objectives. When the risk is low, learners can be allowed to think for themselves; when the risk is high, a safety warning can be triggered even in assessment mode.

[0125] Step 1037: Based on the hierarchical prompting strategy, determine the feedback method according to the current risk level and the current interactive terminal type, and output the prompting information to be output according to the feedback method.

[0126] Specifically, the "prompt method" and "prompt information to be output" in the layered prompting strategy determine "what form of guidance" and "what content to say," while the "feedback method" addresses the question of "what channels and means to convey the prompt information to be output to the trainees." High-voltage testing operations are highly on-site, and text prompts alone may not be sufficient to draw enough attention. Therefore, the system needs to determine an appropriate feedback method based on the current risk level and the type of interactive terminal, delivering the prompt information to be output generated in step 1036 to the trainees in the most effective way.

[0127] Feedback methods may include at least one of the following: text prompts, voice prompts, 3D highlighting, step locking, data panel prompts, follow-up question pop-ups, and debriefing reports, and shall be determined according to the following rules:

[0128] When the current risk level is high, the feedback method of "voice warning + 3D highlighting + step locking" is adopted: the warning output prompt information generated in step 1036 is broadcast by voice, the relevant equipment components (such as discharge rod, grounding point, incorrect wiring terminal) are highlighted in the virtual scene or 3D interface, and dangerous operation steps are locked. Trainees must not continue to operate until they have completed the safety confirmation.

[0129] When the current risk level is low and the interaction terminal type is computer, the feedback method of "text prompt + follow-up question pop-up" is adopted: the prompt information to be output is displayed in the form of a pop-up window in the interface, along with interactive buttons for confirmation or continued operation;

[0130] When the current work phase is the data analysis phase, the feedback method of "data panel prompts" is adopted: in the data display panel, the prompt information to be output is displayed in the form of annotations or highlights, such as "Exceeded the limit of the procedure, it is recommended to compare with the historical trend" next to the dielectric loss data;

[0131] When the current training mode is the debriefing mode, the feedback method is "debriefing report". After the training is completed, a detailed evaluation report containing the error reasons and improvement suggestions is generated. The error reason analysis part directly uses the debriefing explanation class to be output prompt information generated in step 1036.

[0132] When the current interactive terminal type is a head-mounted display device or a mixed reality terminal, the feedback method of "3D highlighting + voice prompts" is given priority. The operation object in the prompt information to be output is visually marked in 3D space, and the prompt content is broadcast synchronously by voice.

[0133] The feedback method selection is represented as follows:

[0134] ;

[0135] Wherein: F t Indicates the final feedback method; Q() represents the feedback method selection function; T t This indicates the content of the tutor's prompts; Risk t Indicates the current risk level; Stage t Indicates the current task stage; Device t This indicates the type of the current interactive terminal, such as a head-mounted display device, a computer, or a tablet.

[0136] For example, in high-risk operations, "voice alerts + 3D highlighting + step locking" can be used; in the data analysis phase, "data panel prompts + heuristic questioning" can be used.

[0137] This step integrates prompts with the virtual work environment, providing actionable work guidance through various methods such as text, voice, 3D highlighting, step locking, data panel prompts, follow-up question pop-ups, and debriefing reports. Feedback methods are determined based on risk level and terminal type, ensuring that prompts are delivered to trainees in the most suitable format for the current scenario and terminal, improving the perceptibility and operability of the prompts. Especially in high-risk scenarios, step locking directly prevents dangerous operations, ensuring work safety.

[0138] In one embodiment, step 104 involves collecting the student's operational behavior records and data analysis path records throughout the entire assignment process, and generating a competency profile containing several competency dimensions based on the operational behavior records and the data analysis path records, including:

[0139] Step 1041: Collect the student's operation behavior records throughout the entire operation process. The operation behavior records shall include at least the operation result, error type, and risk level.

[0140] Specifically, the first step is to record the trainee's operational behavior throughout the entire operation. For example, when the trainee connects the high-voltage output terminal of the dielectric loss tester to the device under test during the wiring operation phase, the operation time is recorded as "14:30:05", the operation phase is "wiring operation phase", the operation type is "connecting cables", the operation object is "high-voltage output terminal of dielectric loss tester", the operation result is "correct", the error type is empty, the risk level is "low risk", the usage prompt is "no", and the error correction time is 0 seconds. When the trainee fails to confirm grounding during the safety confirmation phase, the operation time is recorded as "14:32:10", the operation phase is "safety confirmation phase", and the operation type is "confirming grounding". "Identify Grounding", the operation object is "Grounding Wire", the operation result is "Omission", the error type is "Lack of Grounding", the risk level is "High Risk", the prompt for whether to use is "Yes", and the error correction time is 15 seconds. When a student directly submits a defect judgment during the data analysis phase without checking historical trends, the operation time is recorded as "14:38:20", the relevant task phase is "Data Analysis Phase", the operation type is "Judgment of Defect", the operation object is "Defect Options", the operation result is "Incomplete", the error type is "Missing Analysis Path", the risk level is "Attention", the prompt for whether to use is "No", and the error correction time is 0 seconds. The student operation record set is represented as follows:

[0141] ;

[0142] Where b i Let represent the i-th operation record. Each operation record is represented as:

[0143] ;

[0144] Where t i Indicates the time when the operation occurred, Stage i Indicates the job phase to which the operation belongs, Act i Indicates the operation type, Obj i Indicates the object being operated on, Result i Err represents the result of the operation. i Indicates the error type, Risk i Hint indicates the risk level corresponding to this operation. i Indicates whether to use prompts, Dur i This indicates the time taken for the operation or error correction. The collected operation records are then stored in the student operation record collection.

[0145] This step involves collecting and refining the trainee's entire operational behavior record down to the operational result, error type, and risk level. This enables subsequent competency profiles to be evaluated based on fine-grained behavioral data, avoiding the output of only general evaluations due to coarse-grained recording.

[0146] Step 1042: Collect data on the completion status of necessary analysis steps by trainees during the data analysis phase, and calculate the data analysis path completeness based on the number of completed necessary analysis steps;

[0147] Specifically, the first step is to collect data on the completion status of necessary analysis steps by trainees during the data analysis phase. The necessary analysis steps for high-pressure test data analysis are pre-defined by expert rules, including determining data validity, comparing against standard limits, comparing against historical trends, comparing against manufacturers and similar equipment, identifying potential defects, and proposing remedial measures. For example, if a trainee only performs the steps "determining data validity" and "comparing against standard limits" during the data analysis phase, but fails to perform "comparing against historical trends," "comparing against manufacturers and similar equipment," "identifying potential defects," and "proposing remedial measures," the number N of necessary analysis steps completed is recorded. done =2, Total number of necessary analysis steps N required =6.

[0148] Data analysis records are represented as follows:

[0149] ;

[0150] Among them, Check r Indicates whether to view the procedure limits. f Indicates whether you want to view the manufacturer's factory data. h Indicates whether to view historical data, Check. c Indicates whether to view data for similar devices. e Indicates whether to investigate environmental or wiring factors. Judge x This indicates the result of the data anomaly level judgment. y Indicates the result of the defect cause judgment, Action s The proposed solutions are then presented. The data analysis path completeness is then calculated based on the number of necessary analysis steps completed.

[0151] ;

[0152] For example, C d =2 / 6≈0.33, meaning C d The lower value indicates that the student's data analysis path is incomplete; the other student completed all six necessary analysis steps, so Cd = 6 / 6 = 1.0, meaning C... d A higher score indicates that the student's data analysis path is complete.

[0153] This step, by recording and calculating the completion status of necessary analysis steps in the high-pressure test data analysis process, enables the evaluation of whether trainees have mastered the correct data analysis path, rather than solely evaluating the correctness of the final judgment result.

[0154] Step 1043: Based on the operation behavior record and the data analysis path completeness, calculate the score of each dimension according to several preset ability dimensions to generate the student's ability profile.

[0155] Specifically, based on the recorded operational behavior and the completeness of the data analysis path, scores are calculated for each of the preset capability dimensions. These capability dimensions include: instrument selection capability (A1), standardized process capability (A2), wiring operation capability (A3), safety risk identification capability (A4), test parameter setting capability (A5), data reading capability (A6), test data analysis capability (A7), defect cause judgment capability (A8), post-mortem improvement capability (A9), and independent completion capability (A1). 10 The ability profile is represented as follows:

[0156] ;

[0157] Where m represents the number of capability dimensions. For example, taking security risk identification capability A4 as an example, its calculation method is as follows:

[0158] ;

[0159] Where A4 represents the score for security risk identification capability, N risk R represents the number of risk omissions. max H indicates the highest risk level. risk This indicates the number of times risk warnings are used. The more risks are missed, the higher the highest risk level, and the more reliant on warnings, the lower the score for security risk identification ability. This indicates the weighting of the deduction; taking the test data analysis capability A7 as an example, its calculation method is as follows:

[0160] ;

[0161] Where A7 represents the score for experimental data analysis ability, N judge C represents the number of incorrect data judgments. d H represents the completeness of the data analysis path. d N represents the number of times the prompt is used during the data analysis phase. skip This indicates the number of times necessary data comparison steps are skipped. The more data judgment errors, the more incomplete the analysis path, the higher the dependence on prompts, and the more necessary comparison steps are skipped, the lower the data analysis ability score. μ, ν, ρ, and η are the deduction weight coefficients. Taking the defect cause judgment ability A8 as an example, its calculation method is as follows:

[0162] ;

[0163] Where A8 represents the score for the ability to determine the cause of defects, and N cause D represents the number of errors in determining the cause of a defect. cause H represents the degree of deviation between the trainee's judgment result and the standard cause. cause This indicates the number of times the defect assessment prompts were used, where λ, θ, and ω are the deduction weighting coefficients. For example, a trainee missed risks 1 time in three training sessions, had a highest risk level of "Severe," and used risk prompts 2 times, achieving a safety risk identification ability A4 score of 72. Another trainee missed risks 3 times, had a highest risk level of "Moderate," and used risk prompts 0 times, achieving a safety risk identification ability A4 score of 85. The trainee's ability profile is then generated based on the scores across these dimensions. A comprehensive ability score can also be generated further.

[0164] ;

[0165] Wherein: S core Indicates the overall ability score, A j w represents the score for the j-th ability dimension. j represents the weight of the j-th capability dimension, and m represents the number of capability dimensions. Different capability dimensions can be assigned different weights according to the importance of high-pressure testing.

[0166] This step breaks down high-pressure testing capabilities into multiple professional competency dimensions and generates profiles based on the completeness of behavioral data and data analysis paths throughout the process. This enables the output of scores for each dimension and key weaknesses, rather than just providing a single exam score, thus providing a fine-grained competency data foundation for subsequent weakness identification and personalized recommendations.

[0167] In one embodiment, step 105, comparing the scores of each capability dimension with corresponding thresholds to identify weak capabilities, identifying the causes of weakness based on the error type distribution and prompt usage associated with the weak capabilities, and matching and recommending subsequent training tasks from the training task library based on the weak capabilities and the causes of weakness, includes:

[0168] Step 1051: Compare the scores of each capability dimension with the corresponding pass thresholds, and identify capability dimensions with scores below the pass thresholds as weak capability items;

[0169] Specifically, the scores of each capability dimension are first compared with their corresponding pass / fail thresholds, and capability dimensions with scores below the pass / fail thresholds are identified as weak capability items. Weak capability item identification is represented as follows:

[0170] ;

[0171] Where Weak represents the set of weak capabilities, A j T represents the score of the j-th ability dimension. j This represents the passing threshold for that capability dimension. For example, if the instrument selection capability A1 scores 75 points, and its corresponding passing threshold T1 is 80 points, then instrument selection capability is identified as a weak capability. If the test data analysis capability A7 scores 68 points, and its corresponding passing threshold T7 is 75 points, then test data analysis capability is identified as a weak capability. If the safety risk identification capability A4 scores 82 points, and its corresponding passing threshold T4 is 80 points, then safety risk identification capability is not identified as a weak capability. Then, all capability dimensions with scores below the passing threshold are marked as weak capabilities and recorded in the weak capability set.

[0172] This step compares the scores of each ability dimension with the thresholds, enabling us to accurately pinpoint the specific ability dimensions in which the student is lacking, thus avoiding the inability to identify specific weaknesses by only outputting the overall score.

[0173] Step 1052: Obtain the error type distribution and prompt usage frequency associated with the weak capability item; when the frequency of occurrence of the same error type exceeds the frequency threshold within a preset time period, identify the error type as a weak cause; and / or, when the prompt usage frequency exceeds the prompt frequency threshold, identify the prompt dependency as a weak cause.

[0174] Specifically, firstly, the error type distribution and prompt usage frequency associated with the weak capability item are obtained. When the frequency of occurrence of the same error type exceeds a frequency threshold within a preset time period, the error type is identified as a weakness cause; and / or, when the prompt usage frequency exceeds a prompt frequency threshold, prompt dependency is identified as a weakness cause. Weakness cause identification is represented as follows:

[0175] ;

[0176] Among them, R easonj Let Z() represent the cause of the j-th weakness, and let A represent the weakness cause analysis function. j E represents the ability score. j H indicates the error type associated with this capability dimension. j This indicates the usage of this dimension, C. jThis indicates the completion status of tasks in this dimension. For example, if a trainee's weak skill is experimental data analysis, obtaining the error type distribution associated with this skill reveals that the trainee's "failure to check historical trends" error occurred 4 times in the past 5 training sessions, exceeding the preset frequency threshold of 3 times. Therefore, the weakness is identified as "failure to pay attention to historical trends." Simultaneously, obtaining the number of prompts used shows that the trainee used prompts 8 times during the data analysis phase, exceeding the preset prompt frequency threshold of 5 times. Therefore, "prompt dependence" is also identified as another weakness for this skill. As another example, if a trainee's weak skill is safety risk identification, obtaining the error type distribution associated with this skill reveals that the "missed grounding" error occurred 3 times, exceeding the preset frequency threshold of 2 times. Therefore, "missed grounding" is identified as a weakness. Then, the identified weaknesses are associated and recorded with the corresponding weak skill items.

[0177] This step involves identifying weak skills and then further analyzing the distribution of associated error types and the use of prompts. This allows us to not only determine which skill dimension the learner is weak in, but also to pinpoint the specific reasons for the weakness, providing a more concrete basis for the accurate matching of subsequent training tasks.

[0178] Step 1053: Based on the weak ability items, the reasons for the weaknesses, and the trainee's historical training records, calculate the matching degree between each training task in the training task library and the trainee, and recommend subsequent training tasks based on the matching degree. Each training task in the training task library includes applicable ability dimensions, applicable equipment, applicable test items, difficulty level, training objectives, typical error points, and recommended prerequisites.

[0179] Specifically, firstly, based on the weak skills, the reasons for the weaknesses, and the trainee's historical training records, the matching degree between each training task in the training task library and the trainee is calculated. Subsequent training tasks are recommended based on the matching degree. Each training task in the training task library includes applicable skill dimensions, applicable equipment, applicable experimental items, difficulty level, training objectives, typical error points, and recommended prerequisites. The training task is represented as follows:

[0180] ;

[0181] Among them, Target i This indicates the capability dimension targeted by the task, O i Indicates the applicable equipment, P i Indicates the applicable test item, Difficulty i Indicates the difficulty of the task, Goal i Indicates the training objective, Error i This indicates the typical errors that this task focuses on covering.i This indicates the prerequisites required for recommending this task. For example, the training task library includes the task "Data Analysis Case of Increased Dielectric Loss," with the applicable capability dimension being "Experimental Data Analysis Ability," the applicable equipment being "Main Transformer," the applicable test item being "Dielectric Loss Test," the difficulty level being "Intermediate," and the training objective being "Mastering the complete analysis process for increasing dielectric loss by comparing data from regulations, historical data, manufacturers, and similar equipment." Typical errors are "Failure to check historical trends" and "Failure to check environmental humidity." The recommended prerequisite is "Completion of basic data analysis training." The recommendation matching degree is calculated as follows:

[0182] ;

[0183] Among them, Rec u Let T represent the set of training tasks recommended to student u, g() represent the recommendation matching function, Weak represent the set of weak abilities, Reason represent the set of reasons for weaknesses, U represent the student's historical ability profile and training records, and T represent the set of training tasks recommended to student u. i This section represents candidate training tasks. For example, if a trainee's weak skill is "experimental data analysis ability," and the reasons for this weakness are "lack of attention to historical trends" and "reliance on prompts," and their historical training records show that they have completed "basic data analysis training," then after calculating the matching degree between each training task and the trainee, the "Dielectric Loss Increase Data Analysis Case" task has the highest matching degree, and therefore, this task is recommended as a subsequent training task. If the trainee's weak skill is "wiring operation ability," and the reason for this weakness is "frequent wiring errors," then the "Dielectric Loss Test Wiring Special" task is recommended. If the trainee's weak skill is "defect cause judgment ability," and the reason for this weakness is "insulation moisture review case," then the "Less Prompt Mode Training" task is recommended to help the trainee reduce reliance on prompts. Then, one or more training tasks with the highest matching degree are output as the recommended results.

[0184] This step automatically matches training tasks from the training task library based on the weak skills, the reasons for the weakness, and the trainee's historical training records. This allows for the recommendation of training content that addresses the specific weaknesses of different trainees, enabling targeted improvement, unlike the approach of pushing uniform training content to all trainees.

[0185] This invention provides a method for intelligent guidance and capability profile generation in high-pressure testing. First, it collects and categorizes high-pressure testing knowledge data, breaking it down into structured knowledge units. This transforms scattered procedures, manufacturer information, historical cases, and expert experience into computable and matchable knowledge units, solving the problem of the difficulty in solidifying expert experience and distinguishing it from the generalized answers of ordinary AI question-answering systems. Then, by dividing the high-pressure testing operation into several stages, it collects the trainee's current stage, operational behavior, and test data in real time. A dynamic operational context is constructed based on the trainee's capability level, and the correlation between the operational context and each knowledge unit is calculated to filter target knowledge units. A tiered prompting strategy is determined based on the target knowledge unit, capability level, current training mode, and current risk level. Real-time guidance and feedback are provided to the trainee according to this strategy, enabling dynamic perception of the trainee's current specific operational stage and corresponding risk level. When test data is abnormal, the system guides the trainee to perform in-depth reasoning according to the correct analytical path. It also achieves tiered prompts based on trainee level, training mode, and risk level, thus solving the problem of lack of context awareness and in-depth data analysis and reasoning capabilities in intelligent guidance, distinguishing it from existing systems with fixed-step prompts. Next, by collecting operational behavior records and data analysis path records of trainees throughout the entire assignment process, a competency profile containing several competency dimensions is generated based on these records. This allows for accurate identification of trainees' specific competency weaknesses in each professional dimension based on fine-grained, end-to-end behavioral data, thus solving the problem of limited competency evaluation dimensions caused by coarse-grained operational records, unlike existing evaluation methods that only output a general total score. Finally, by comparing the scores of each competency dimension with corresponding thresholds to identify weak competencies, the causes of weaknesses are identified based on the distribution of error types associated with weak competencies and the usage of prompts. Subsequent training tasks are matched and recommended from the training task library based on the weak competencies and their causes. This allows for automatic adjustment of subsequent training content based on different trainees' weak competencies and their causes, thus solving the problem of lack of personalized recommendations due to standardized training content, unlike existing methods that push uniform training content to all trainees. In summary, this invention solves the problems of existing technologies lacking assignment context awareness and data analysis reasoning capabilities, limited competency evaluation dimensions due to coarse-grained operational records, and lack of personalized recommendations and difficulty in solidifying expert experience due to standardized training content.

[0186] Furthermore, in one embodiment, the high-pressure test intelligent guidance and capability profile generation method provided by the present invention further includes:

[0187] Step 1061: Record the trainee's ability profile after multiple training sessions, and calculate the improvement in each ability dimension based on the ability profile after each training session.

[0188] Specifically, firstly, the trainee's ability profile is recorded after multiple training sessions, and then the improvement in each ability dimension is calculated based on the ability profile after each training session. The trainee's ability profile after multiple training sessions is recorded as follows:

[0189] ;

[0190] in, This represents a capability profile after the first training session. This represents the ability profile after the nth training iteration. Based on the ability profiles after each training iteration, the improvement in each ability dimension is calculated, expressed as:

[0191] ;

[0192] in This represents the improvement in the j-th capability dimension. This represents the score for this capability dimension after the nth training iteration. This represents the score for that ability dimension after the first training session. For example, if a trainee scores 60 points in experimental data analysis after the first training session and 75 points after the third training session, the improvement is 75 - 60 = 15 points. Similarly, if a trainee scores 80 points in security risk identification after the first training session and 82 points after the third training session, the improvement is 82 - 80 = 2 points. Then, the improvement for each ability dimension is compared to a preset improvement threshold, and dimensions with improvements below the threshold are marked.

[0193] This step records the ability profiles after multiple training sessions and calculates the improvement in each dimension, enabling the tracking of learners' ability trends and determining the effectiveness of the training.

[0194] Step 1062: When the improvement of any of the aforementioned capability dimensions is lower than a preset threshold, reduce the difficulty level of the training task or add basic knowledge prompts.

[0195] Specifically, firstly, when the improvement in any of the aforementioned capability dimensions falls below a preset threshold, the difficulty level of the training task is reduced or basic knowledge prompts are added. For example, if the preset improvement threshold is 5 points, and a student's improvement in security risk identification ability is 2 points, below the preset threshold of 5 points, it indicates that the security risk identification ability has not improved over a long period. The difficulty level of the student's subsequent training tasks is reduced, for example, by adjusting the security risk identification training from "intermediate" to "beginner" difficulty, or by adding basic knowledge prompts during task execution. As another example, if a student's improvement in experimental data analysis ability is 3 points, below the preset threshold of 5 points, basic knowledge prompts such as "Please check historical trend data first" are added to subsequent data analysis training tasks to help the student establish correct analysis habits. If the improvement in multiple capability dimensions is below the preset threshold, both reducing the difficulty level and adding basic knowledge prompts can be adopted simultaneously. Then, subsequent training tasks are recommended according to the adjusted strategies, and the improvement is recalculated after the next training session, forming a continuous optimization cycle.

[0196] This step records the capability profiles after multiple training sessions and calculates the improvement in each capability dimension. When the capability does not improve for a long period of time, the recommendation strategy is automatically adjusted to form a continuous improvement loop. This not only completes a training evaluation, but also tracks the trend of capability changes and dynamically adjusts the training recommendations to avoid training stagnation.

[0197] To more clearly illustrate the synergistic working relationship between the above-mentioned steps of the present invention, the method of the present invention will be further described below in conjunction with two specific application scenarios.

[0198] Application Scenario 1: Analysis of Dielectric Loss Test Data for Main Transformer

[0199] This application scenario illustrates how the present invention, through the coordinated action of steps 101 to 105, achieves full-process guidance in the high-pressure test data analysis stage, including context awareness, knowledge matching, hierarchical prompts, behavior recording, and capability profiling.

[0200] In this scenario, the equipment type is a main transformer, the test item is a dielectric loss test, the current operation stage is the data analysis stage, the current test data shows that the dielectric loss value is increasing and the insulation resistance is decreasing, the trainee's behavior is to directly judge that the insulation is damp, without checking the historical trend, and without checking the shielding wiring and the ambient humidity.

[0201] First, construct the job context based on the current equipment type, current test item, current operation stage, current operation behavior, and current test data:

[0202] ;

[0203] Where O is the main transformer, P is the dielectric loss test, and Staget For the data analysis phase, X t As dielectric loss increases and insulation resistance decreases, A t Trainees can directly submit defect assessments.

[0204] Then, using relevance calculations to match knowledge units, the system matches the knowledge units for main transformer dielectric loss anomaly analysis, insulation moisture or insulation degradation, the impact of shielding wiring, historical trend comparison, and the impact of environmental humidity. If the trainee is a new employee in learning mode, a direct prompt strategy is generated: "Please first determine if the data exceeds the specification limits, then compare it with the manufacturer's factory values ​​and historical trends, and simultaneously check the impact of shielding wiring and environmental humidity." Real-time guidance and feedback are then provided through text prompts or data panel prompts according to this hierarchical prompt strategy. If the trainee is an experienced employee, a heuristic follow-up question strategy is generated: "Does increased dielectric loss necessarily mean the equipment is damp? What other test factors need to be checked?"

[0205] Then, record whether the trainee has reviewed the procedure limits, manufacturer's data, historical data, data from similar equipment, environmental factors, and wiring factors, and calculate the data analysis path completeness. If the trainee only identifies defects without completing the necessary comparisons, then C... d The score is low, and the test data analysis ability (A7) is penalized; if the defect judgment deviates significantly from the standard cause, then D... cause The score was significantly lower than A8 for the ability to determine the cause of the defect.

[0206] Finally, based on the above behavioral records and the completeness of the data analysis path, a competency profile of the trainees is generated, weak competencies and their causes are identified, and subsequent training tasks are recommended.

[0207] Application Scenario 2: Risk of Failure to Discharge After DC Test

[0208] This application scenario illustrates how the present invention performs risk warnings, step locking, and capability profiling and recording for high-risk operational behaviors.

[0209] In this scenario, the equipment type is high-voltage electrical equipment, the test item is DC high-voltage test, the current operation stage is the test completion and discharge stage, the trainee's behavior is that the trainee has not performed a full discharge and is preparing to dismantle the test line, and the risk type is the risk of electric shock due to residual charge.

[0210] First, the operational context is constructed and matched with the safety procedure knowledge unit "full discharge is required after DC testing". The applicable equipment for this knowledge unit is high-voltage electrical equipment, the applicable test item is DC high-voltage testing, the applicable operation stage is the test completion and discharge stage, the triggering condition is failure to perform full discharge, the risk level is severe, the recommended action is to use a discharge rod to fully discharge and confirm grounding, and the knowledge source is on-site safety procedures.

[0211] Then, since the current risk level has reached the critical threshold, a tiered prompting strategy of "risk warning + step lock" is generated instead of the ordinary prompting method.

[0212] Next, based on the tiered prompting strategy and the current risk level, real-time guided feedback is implemented using a feedback method of "voice prompts + 3D highlighting + step locking". The discharge rod, grounding point and test device terminals are highlighted on the interface. A voice prompt reminds the user that "there is a risk of residual charge. Please complete the discharge and confirm grounding first", and the disconnection step is locked. The user must not continue to operate until the discharge is confirmed.

[0213] Then, the operation behavior records are collected, including one instance of risk omission, one instance of severe risk level, whether prompts were used, and the time spent on error correction, which are used for the subsequent A4 score of security risk identification capability.

[0214] Finally, a competency profile of the trainee is generated, identifying weak competencies and the reasons for these weaknesses, and recommending subsequent training tasks.

[0215] It should be noted that, in the specific implementation of the above steps, those skilled in the art should understand that the present invention is not limited to the above exemplary implementation methods, and various alternative solutions can be used to achieve the same or substantially the same function. Several optional alternative methods are listed below:

[0216] In terms of implementation logic, the above knowledge matching and prompt generation process can be implemented using rule engines, knowledge graphs, retrieval-enhanced generation models, or multi-model fusion. That is, the system does not necessarily have to rely on a specific reasoning model. Any implementation method that can perform knowledge matching and prompt strategy generation based on structured knowledge units and task context can be used as an alternative.

[0217] In terms of data storage, the high-pressure test knowledge base can be stored using relational databases, graph databases, vector databases, or hybrid databases to meet the management and retrieval needs of knowledge units of different sizes.

[0218] Regarding interactive terminals, the collection of work context and the execution of real-time guidance feedback are not limited to head-mounted display devices, but can also be completed through mixed reality terminals, computers, tablets or field auxiliary terminals. The system can adaptively adjust the feedback method according to different terminal types.

[0219] In terms of prompts and feedback, the prompts and feedback methods for real-time guidance are not limited to text, voice and 3D highlighting, but can also take the form of pop-up windows, vibration feedback, step locking, or a combination of the above forms.

[0220] In terms of presenting competency profiles, competency profiles are not limited to a percentage-based scoring system. They can also be presented to trainees in the form of a rating system, radar charts, text reports, or trend graphs to more intuitively reflect competency structure and changing trends.

[0221] In terms of personalized recommendations, recommendations for subsequent training tasks are not limited to rule matching methods; similar case matching, machine learning recommendation models, or hybrid recommendation models can also be used.

[0222] The above-mentioned alternative solutions can also achieve the purpose of this invention. Those skilled in the art can choose a suitable implementation method according to the actual application scenario and deployment conditions, and all of them fall within the protection scope of this invention.

[0223] The above describes a method for intelligent guidance and capability profile generation for high-pressure testing provided in the embodiments of the present invention. The following describes a system for intelligent guidance and capability profile generation for high-pressure testing provided in the embodiments of the present invention.

[0224] Please see Figure 2 The present invention provides a high-pressure test intelligent guidance and capability profile generation system, comprising:

[0225] The data acquisition unit 201 is used to acquire high-pressure test knowledge data and classify it, and decompose the classified knowledge data into structured knowledge units.

[0226] The construction unit 202 is used to divide the high-voltage test operation into several operation stages, collect the operation stage, operation behavior and test data of the trainee in the current operation process in real time, and construct a dynamic operation context in combination with the trainee's ability level;

[0227] The calculation unit 203 is used to calculate the relevance between the task context and each knowledge unit to filter target knowledge units, determine a hierarchical prompting strategy based on the target knowledge unit, the ability level, the current training mode and the current risk level, and provide real-time guidance and feedback to the trainee according to the hierarchical prompting strategy.

[0228] The generation unit 204 is used to collect the operation behavior records and data analysis path records of the trainees during the entire process of the assignment, and generate a capability profile containing several capability dimensions based on the operation behavior records and data analysis path records.

[0229] The matching unit 205 is used to compare the scores of each of the capability dimensions with the corresponding thresholds to identify weak capability items, identify the cause of weakness based on the error type distribution and prompt usage associated with the weak capability items, and match and recommend subsequent training tasks from the training task library based on the weak capability items and the cause of weakness.

[0230] Furthermore, this embodiment of the invention also provides a high-pressure test intelligent guidance and capability profile generation device, the device including a processor and a memory:

[0231] The memory is used to store program code and transmit the program code to the processor;

[0232] The processor is used to execute the steps of the high-pressure test intelligent guidance and capability profile generation method as described in the above method embodiments, according to the instructions in the program code.

[0233] Furthermore, this embodiment of the invention also provides a computer-readable storage medium for storing program code, which is used to execute the high-pressure test intelligent guidance and capability profile generation method described in the above method embodiments.

[0234] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0235] In the embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0236] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0237] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0238] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0239] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent guidance and capability profile generation in high-pressure testing, characterized in that, include: Collect and classify high-pressure test knowledge data, and then break down the classified knowledge data into structured knowledge units; The high-pressure test operation is divided into several operation stages. The operation stage, operation behavior and test data of the trainee in the current operation process are collected in real time, and a dynamic operation context is constructed in combination with the trainee's ability level. The relevance between the task context and each knowledge unit is calculated to filter target knowledge units. A tiered prompting strategy is determined based on the target knowledge unit, the ability level, the current training mode, and the current risk level. Real-time guidance and feedback are then provided to the trainee according to the tiered prompting strategy. Collect student's operational behavior records and data analysis path records throughout the entire assignment process, and generate a competency profile containing several competency dimensions based on the operational behavior records and the data analysis path records; The scores of each capability dimension are compared with the corresponding thresholds to identify weak capabilities. The reasons for the weakness are identified based on the error type distribution and prompt usage associated with the weak capability. Subsequent training tasks are matched and recommended from the training task library based on the weak capability and the reasons for the weakness.

2. The high-voltage test intelligent guidance and capability profile generation method according to claim 1, characterized in that, The process involves collecting and classifying high-pressure test knowledge data, then breaking down the classified knowledge data into structured knowledge units, including: Collect high-voltage testing knowledge data, which includes at least one of the following: regulations and standards, work instructions, instrument data, manufacturer data, historical data, defect cases, expert experience, and risk rules; Add source tags, scope of application tags, credibility tags, and update time tags to each of the knowledge materials to obtain the classified and labeled knowledge materials; The categorized and labeled knowledge data is broken down into knowledge units. Each knowledge unit includes at least the applicable equipment, applicable test items, applicable operation stages, triggering conditions, knowledge content, risk level, judgment basis, recommended actions, and knowledge source.

3. The high-pressure test intelligent guidance and capability profile generation method according to claim 1, characterized in that, The high-voltage testing operation is divided into several operation stages. Real-time data collection of the trainee's current operation stage, operational behavior, and test data is used to construct a dynamic operation context based on the trainee's ability level. This includes: After the high-pressure test operation is pre-divided into several operation stages, the operation stage, operation behavior and test data of the trainee in the current operation process are collected in real time, and the trainee's ability level is determined based on the trainee's historical operation records. A dynamic job context is constructed based on the current equipment type, current test project, job stage, operation behavior, test data, environmental parameters, historical data retrieval status, and capability level.

4. The high-voltage test intelligent guidance and capability profile generation method according to claim 1, characterized in that, The calculation of the relevance between the job context and each of the knowledge units to filter target knowledge units includes: Retrieve the current equipment type, current test item, current job stage, current operation behavior, and current test data from the current job context; For each of the knowledge units, calculate the following: a first relevance sub-item matching the current device type; a second relevance sub-item matching the current test project; a third relevance sub-item matching the current operation stage; a fourth relevance sub-item matching whether the current operation behavior meets the triggering condition; a fifth relevance sub-item matching whether the current test data meets the triggering condition; and a sixth relevance sub-item matching the credibility level of the knowledge unit. The first to sixth relevance sub-items are weighted and summed to obtain the relevance between the job context and each knowledge unit. The knowledge units are sorted according to their relevance, and target knowledge units are selected from the knowledge units based on the sorting results.

5. The high-voltage test intelligent guidance and capability profile generation method according to claim 1, characterized in that, The step of determining a tiered prompting strategy based on the target knowledge unit, the ability level, the current training mode, and the current risk level, and then providing real-time guidance and feedback to the trainee according to the tiered prompting strategy, includes: Obtain the knowledge content of the target knowledge unit, and obtain the trainee's ability level, current training mode, and current risk level; Based on the ability level, the current training mode, and the current risk level, and in conjunction with the knowledge content, a hierarchical prompting strategy is generated. The hierarchical prompting strategy includes a prompting method that matches the current scenario and a prompting message to be output generated based on the knowledge content. Based on the hierarchical prompting strategy, the feedback method is determined according to the current risk level and the current interactive terminal type, and the prompting information to be output is output according to the feedback method.

6. The high-voltage test intelligent guidance and capability profile generation method according to claim 1, characterized in that, The system collects records of trainees' operational behaviors and data analysis paths throughout the entire assignment process. Based on these records, a competency profile is generated, comprising several competency dimensions, including: Collect records of trainees' operational behaviors throughout the entire assignment process. These records must include at least the operational results, error types, and risk levels. Collect data on the completion status of necessary analysis steps by trainees during the data analysis phase, and calculate the data analysis path completeness based on the number of necessary analysis steps completed. Based on the recorded operation behavior and the completeness of the data analysis path, scores for each dimension are calculated according to several preset capability dimensions to generate a capability profile of the trainee.

7. The method for intelligent guidance and capability profile generation in high-voltage testing according to claim 1, characterized in that, The step of comparing the scores of each capability dimension with corresponding thresholds to identify weak capabilities, identifying the causes of weakness based on the error type distribution and prompt usage associated with the weak capabilities, and matching and recommending subsequent training tasks from the training task library based on the weak capabilities and the causes of weakness includes: The scores of each capability dimension are compared with the corresponding pass thresholds, and capability dimensions with scores below the pass thresholds are identified as weak capability items. Obtain the error type distribution and prompt usage frequency associated with the weak capability item; when the frequency of occurrence of the same error type exceeds a frequency threshold within a preset time period, the error type is identified as a weak cause; and / or, when the prompt usage frequency exceeds a prompt frequency threshold, prompt dependency is identified as a weak cause. Based on the weak skills, the reasons for the weakness, and the trainee's historical training records, the matching degree between each training task in the training task library and the trainee is calculated. Subsequent training tasks are recommended based on the matching degree. Each training task in the training task library includes applicable skill dimensions, applicable equipment, applicable test items, difficulty level, training objectives, typical error points, and recommended prerequisites.

8. The high-voltage test intelligent guidance and capability profile generation method according to claim 1, characterized in that, The step of matching and recommending subsequent training tasks from the training task library based on the weak capability items and the reasons for the weaknesses, further includes: Record the trainees' ability profiles after multiple training sessions, and calculate the improvement in each ability dimension based on the ability profiles after each training session; When the improvement in any of the aforementioned capability dimensions is less than a preset threshold, the difficulty level of the training task is reduced or basic knowledge prompts are added.

9. A high-pressure test intelligent guidance and capability profile generation system, characterized in that, include: The data acquisition unit is used to collect and classify high-pressure test knowledge data, and to break down the classified knowledge data into structured knowledge units. The construction unit is used to divide the high-voltage test operation into several operation stages, collect the operation stage, operation behavior and test data of the trainee in the current operation process in real time, and construct a dynamic operation context in combination with the trainee's ability level. The calculation unit is used to calculate the relevance between the task context and each knowledge unit to filter target knowledge units, determine a hierarchical prompting strategy based on the target knowledge unit, the ability level, the current training mode and the current risk level, and provide real-time guidance and feedback to the trainee according to the hierarchical prompting strategy. The generation unit is used to collect the student's operation behavior records and data analysis path records throughout the entire assignment process, and generate a capability profile containing several capability dimensions based on the operation behavior records and data analysis path records. The matching unit is used to compare the scores of each capability dimension with the corresponding thresholds to identify weak capability items, identify the causes of weakness based on the error type distribution and prompt usage associated with the weak capability items, and match and recommend subsequent training tasks from the training task library based on the weak capability items and the causes of weakness.

10. A high-pressure test intelligent guidance and capability profile generation device, characterized in that, The device includes a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the high-pressure test intelligent guidance and capability profile generation method according to any one of the claims 1-8 according to the instructions in the program code.

Images