Virtual reality operation process adaptation method for classroom virtual experiment

Abstract

The invention discloses a virtual reality operation process adaptation method for a classroom virtual experiment, and relates to the technical field of virtual classs.The method comprises the following steps that hand posture data and operation data of a target object are obtained; performing feature analysis and template matching on the hand posture data according to a sliding window to generate a gesture candidate set, and performing time mapping on the operation data according to the gesture candidate set to generate a credibility sequence; and performing event anomaly evaluation on the operation data according to operation event segments, and performing time sequence accumulation on an event anomaly evaluation result and the gesture candidate set on the same operation event according to the credibility sequence to generate a risk accumulation curve. According to the scheme, the exception index of the operation event is matched with the preliminary matching fragment according to the timestamp, and the matching result is subjected to weighted calculation in combination with the credibility sequence, so that the risk assessment is ensured to consider the size of the operation exception and the reliability of gesture execution.

Description

Technical Field

[0001] This invention relates to the field of virtual classroom technology, specifically to a method for adapting virtual reality operation processes for virtual classroom experiments. Background Technology

[0002] With the rapid development of virtual reality technology, the application of virtual experiments in the field of education has gradually become widespread. Virtual experiments, by constructing an immersive and interactive experimental environment, enable students to complete experimental operations without the need for real experimental materials and venues. This technological environment can not only reduce experimental costs and improve experimental safety, but also enable the experimental process to be recorded, traceable, and analyzable, thereby assisting teachers in evaluating and providing feedback on students' operational behavior.

[0003] In existing technologies, gesture recognition-based control methods are commonly used to guide virtual experiment operation processes. For example, by capturing the hand movement trajectory and joint angles of the target object during operation, gesture recognition results are generated using sliding windows or template matching to determine the operation type for real-time monitoring and feedback. The advantages of these methods are that they can perform basic recognition and partial guidance of the target object's operation, provide a certain degree of real-time interaction, and play an auxiliary role in experimental safety and operation guidance. However, in processing operations, these methods only make coarse-grained judgments on complete events and lack a data analysis mechanism based on event segments. It is difficult to synchronously accumulate abnormal indicators and gesture candidate credibility sequences of each segment, resulting in insufficient risk assessment accuracy. Furthermore, when identifying risky operations based on the risk assessment results, there is a lack of self-calibration grading based on accumulated deviation, which limits the accuracy of process adaptation. Summary of the Invention

[0004] The purpose of this invention is to provide a virtual reality operation process adaptation method for classroom virtual experiments, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution of the present invention is as follows:

[0006] In a first aspect, the present invention discloses a virtual reality operation process adaptation method for classroom virtual experiments, comprising the following steps:

[0007] Acquire the hand gesture data and operation data of the target object;

[0008] The hand gesture data is subjected to feature analysis and template matching by sliding window to generate a gesture candidate set, and the operation data is time-mapped according to the gesture candidate set to generate a confidence sequence;

[0009] The operation data is segmented by operation event for event anomaly assessment. Based on the confidence sequence, the event anomaly assessment results and the gesture candidate set are accumulated in time series on the same operation event to generate a risk accumulation curve. Furthermore, based on the risk accumulation curve, the gesture candidate set is subjected to inverted accumulation and sorting optimization to generate a risk weight sequence.

[0010] The operation data is statistically analyzed using a sliding window to determine the load characteristics. Based on the risk weight sequence, the load characteristics and the gesture candidate set are used to perform risk interaction calculations to generate a cognitive load score. The cognitive load score is then compared with a self-calibration grading threshold to generate a set of load-triggered events.

[0011] The self-calibration grading threshold is obtained by performing distribution statistics on the cognitive load score;

[0012] Based on the credibility sequence, the set of load triggering events is corrected for deviations to generate operation adaptation data.

[0013] Secondly, this invention discloses a virtual reality operation process adaptation system for classroom virtual experiments, comprising:

[0014] The data acquisition module is used to acquire the hand gesture data and operation data of the target object;

[0015] The credibility assessment module is used to perform feature analysis and template matching on the hand gesture data by sliding window, generate a gesture candidate set, and perform time mapping on the operation data based on the gesture candidate set to generate a credibility sequence.

[0016] The risk assessment module is used to perform event anomaly assessment on the operation data by segmenting it according to operation events, and to perform time-series accumulation of the event anomaly assessment results and the gesture candidate set on the same operation event according to the credibility sequence to generate a risk accumulation curve. Furthermore, the module performs inverted accumulation and sorting optimization on the gesture candidate set according to the risk accumulation curve to generate a risk weight sequence.

[0017] The load detection module is used to statistically analyze the load characteristics of the operation data using a sliding window, and to perform risk interaction calculations on the load characteristics and the gesture candidate set according to the risk weight sequence to generate a cognitive load score. Then, the cognitive load score is compared with a self-calibration grading threshold to generate a set of load triggering events.

[0018] An operation adaptation processing module is used to perform deviation correction on the load triggering event set according to the credibility sequence and generate operation adaptation data.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0020] 1. This solution matches the abnormal indicators of operational events with the initial matching segments by timestamp, and combines the confidence sequence to weight the matching results. This ensures that the risk assessment considers both the magnitude of the operational anomaly and the reliability of the gesture execution. The initial risk value is corrected by interval penalty and summarized in chronological order to form a risk accumulation curve. By mapping the time interval with the confidence assessment results of the initial matching segments, the impact of adjacent initial matching segments on the overall risk is dynamically adjusted, thus achieving accurate quantification of operational risk.

[0021] 2. This scheme statistically analyzes the load characteristics of operational data using a sliding window and maps them to the time window of the initial matching segments in the gesture candidate set. This allows for precise association between time-related operational loads and each initial matching segment. By combining risk weight sequences and confidence sequences for risk interaction calculation and using confidence-weighted normalization, the generated cognitive load score not only reflects the potential changes in cognitive load during the operation but also reflects the risk sensitivity and execution reliability of each initial matching segment during execution. This enables dynamic evaluation of the operational load of the target object during the virtual experiment. Attached Figure Description

[0022] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts. Wherein:

[0023] Figure 1 A flowchart illustrating the steps of the virtual reality operation process adaptation method for classroom virtual experiments according to the present invention;

[0024] Figure 2 A schematic diagram of the process for generating a risk accumulation curve provided by the present invention;

[0025] Figure 3 A schematic diagram of the process for generating risk weight sequences provided by the present invention;

[0026] Figure 4 This is a schematic diagram of the process for generating a cognitive load score provided by the present invention;

[0027] Figure 5 This is a schematic diagram of the module functions of the virtual reality operation process adaptation system for classroom virtual experiments provided by the present invention. Detailed Implementation

[0028] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0029] Application Overview:

[0030] In the process of adapting traditional virtual experiment operation procedures, existing technologies only make coarse-grained judgments on complete operation events and lack a data parsing mechanism based on event segments. This makes it difficult to achieve synchronous accumulation of abnormal indicators and gesture candidate credibility sequences in each segment, resulting in insufficient accuracy of risk assessment. At the same time, the risk operation identification link lacks a self-calibration and grading mechanism based on accumulated deviation, which limits the accuracy of process adaptation and thus affects the accuracy of operation monitoring and feedback.

[0031] For example, in a titration scenario of a virtual chemical experiment, when the target object performs burette control actions, the system captures the hand movement trajectory and joint angle through gesture recognition. However, it can only perform basic recognition of the overall titration event and is unable to quantitatively analyze segmented operational events such as changes in drip rate and droplet intervals. Specifically, when the target object experiences local drip rate anomalies during titration, the system lacks the ability to accumulate segmented anomaly indicators and gesture reliability sequences over time, making it difficult to generate continuous risk assessment curves. Furthermore, risk identification relies on preset fixed thresholds and fails to dynamically calibrate grading standards based on historical operational data. This leads to an increased misjudgment rate of abnormal operations and affects the timeliness of experimental guidance.

[0032] If the above problems are not addressed, insufficient risk assessment accuracy will lead to inaccurate operation adaptation data generation, making it difficult for the virtual experiment system to effectively identify segmented operation risks, which may cause experimental process interruptions or failure of safety warnings. At the same time, the lack of a self-calibration and grading mechanism will result in a decrease in the system's adaptability to different operating environments or user groups, and a deviation between the cognitive load assessment results and actual operational needs, thereby reducing the reliability of virtual experiment teaching and hindering the optimized application of immersive experimental environments.

[0033] After introducing the basic concept of the present invention, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0034] Example 1:

[0035] Please see Figure 1 The virtual reality operation process adaptation method for classroom virtual experiments includes the following steps:

[0036] Acquire the hand gesture data and operation data of the target object;

[0037] The hand gesture data is subjected to feature analysis and template matching by sliding window to generate a gesture candidate set, and the operation data is time-mapped according to the gesture candidate set to generate a confidence sequence;

[0038] The operation data is segmented by operation event for event anomaly assessment. Based on the confidence sequence, the event anomaly assessment results and the gesture candidate set are accumulated in time series on the same operation event to generate a risk accumulation curve. Furthermore, based on the risk accumulation curve, the gesture candidate set is subjected to inverted accumulation and sorting optimization to generate a risk weight sequence.

[0039] The operation data is statistically analyzed using a sliding window to determine the load characteristics. Based on the risk weight sequence, the load characteristics and the gesture candidate set are used to perform risk interaction calculations to generate a cognitive load score. The cognitive load score is then compared with a self-calibration grading threshold to generate a set of load-triggered events.

[0040] The self-calibration grading threshold is obtained by performing distribution statistics on the cognitive load score;

[0041] Based on the credibility sequence, the set of load triggering events is corrected for deviations to generate operation adaptation data.

[0042] Among them, hand gesture data refers to a set of digital information used to describe the motion state and joint configuration of the target object's hand in three-dimensional space;

[0043] Operational data refers to the structured record data generated in classroom virtual experiments when the target object performs operations in a virtual environment, which can be used to analyze the execution of actions, the sequence of events, and the duration.

[0044] Feature analysis refers to the process of quantitatively describing hand gesture data at each moment or within a sliding window in a virtual experiment from multiple dimensions.

[0045] Template matching refers to the process of comparing the hand gesture feature sequence extracted within the sliding window with the gesture templates preset by the system.

[0046] Gesture candidate set refers to the set of candidate operation segments generated by the system after extracting features from hand gesture data within a sliding time window and matching them with preset gesture templates;

[0047] A reliability sequence is a data sequence used to quantify the reliability of execution at each operational time period or critical node.

[0048] Event anomaly assessment refers to the process of quantitatively calculating the deviation, delay, or error during the execution of each operational event (such as key press, grab, move, place, etc.).

[0049] Temporal accumulation refers to the chronological accumulation and synthesis of the abnormal characteristics of operational events, the matching information of the gesture candidate set, and the credibility, in order to quantify the dynamic evolution of risks during the operation process.

[0050] The risk accumulation curve is used to represent the cumulative risk level estimated by the system during the execution of each operation event and gesture candidate segment;

[0051] Inverted cumulative and sorting optimization refers to the process of calculating the cumulative risk value of each preliminary matching segment in chronological order based on the generated gesture candidate set, combined with the risk exposure and execution credibility of each preliminary matching segment in the operation event, and optimizing the order of the preliminary matching segments through an inverted strategy.

[0052] Risk weight sequence refers to a data structure used to quantify the potential error risk of each gesture candidate segment during operation execution;

[0053] Load characteristics refer to quantitative indicators that characterize the difficulty of performing operational tasks and the consumption of cognitive resources by a target object in a virtual experiment through statistical analysis and feature extraction of operational data.

[0054] Risk interaction computation refers to the process of jointly analyzing the four dimensions of information—time, credibility, risk, and cognitive load—of the preliminary matched segments.

[0055] Cognitive load score refers to the intensity of cognitive stress experienced by the target subject due to task complexity, operational difficulty, and operational risk at each initial matching segment;

[0056] Self-calibrated grading thresholds refer to multi-level load limits automatically generated based on the cognitive load score distribution within the current session.

[0057] The set of load-triggered events refers to the set of all high, medium, and low load trigger points determined by the system during the virtual experiment based on the cognitive load score of the target object's operation and the self-calibration threshold.

[0058] Deviation correction refers to the process of real-time identification, quantification, and correction of deviations from the expected path or operational target caused by the target object's gesture operation or system response during the execution of a virtual classroom experiment.

[0059] Operational adaptation data refers to the data set generated in a virtual experiment system by processing and associating existing data such as hand posture data, operation data, gesture candidate sets, credibility sequences, risk weight sequences, and cognitive load scores. This data is used to adapt the target object's operational behavior in real time, correct deviations, adjust operational procedures, and trigger prompts or intervention strategies.

[0060] This solution achieves real-time quantification and adaptive correction of the deviation between the actual operational intention of the target object and the system's target operation process during virtual classroom experiments by performing multi-level correlation processing of hand gesture data and operation data in the time and event dimensions. Among them, sliding window feature analysis and template matching improve the stability of gesture recognition, enabling the credibility sequence to accurately reflect the reliability of gestures over time. The risk accumulation curve generated by event anomaly assessment combined with the credibility sequence can reveal the potential error trend of the target object in complex experimental steps at a fine-grained level, and form the ability to prioritize the identification of key risk gestures through inverted accumulation and sorting optimization. On this basis, the interactive calculation of risk weight sequence and load features can dynamically estimate the cognitive pressure of the target object during operation, so that the cognitive load score can reflect both the complexity of the action and the risk sensitivity, while the self-calibration threshold ensures that the score is adapted to different target object characteristics and different experimental tasks. Finally, the deviation correction of the credibility sequence for load-triggered events makes the operation adaptation data real-time and personalized, which can effectively reduce the occurrence rate of misoperation, improve the consistency of action understanding and system response in the virtual reality classroom environment, and significantly enhance the adaptability and stability of the virtual experiment operation process in multiple target objects and multiple scenarios.

[0061] The above describes a complete solution for adapting virtual reality operation processes to classroom virtual experiments. The following section describes how to obtain the hand gesture data and operation data of the target object, specifically including:

[0062] The hand posture data of the target object is acquired through a hand tracking sensor; the hand posture data includes, but is not limited to, position vector, joint angle vector, velocity and acceleration, and timestamp, etc.

[0063] Operational data of the target object is acquired through an inertial measurement unit (IMU); the operational data includes, but is not limited to, device confidence level of the operation event, timestamp, pause duration, error flag, success flag, and operation interval.

[0064] The above describes the acquisition of hand gesture data and operation data of the target object. The following describes the feature analysis and template matching of the hand gesture data using a sliding window to generate a gesture candidate set, specifically including:

[0065] The hand gesture data is analyzed by sliding window, and the feature analysis results of each sliding window are compared with a preset gesture template signature set for similarity measurement. It is determined whether the similarity measurement result is greater than a preset discrimination threshold. If so, the hand gesture data in the corresponding sliding window is marked as a preliminary matching segment.

[0066] The confidence level of the preliminary matching segments is evaluated, and then the preliminary matching segments are merged into windows based on the confidence level evaluation results and the start and end timestamps of the preliminary matching segments to generate a gesture candidate set.

[0067] The preset gesture template signature set refers to a set of standardized gesture fragment feature vectors pre-stored by the system. It is obtained by collecting hand posture data from the system's sensors and extracting features through training or recording standard gestures during the experimental phase.

[0068] Similarity measurement results refer to the data structure used to measure the degree of matching between the sliding window features of hand gesture data and the signature of a pre-set gesture template;

[0069] The preset discrimination threshold is a numerical limit used to judge the degree of matching between the hand gesture features in the sliding window and the preset gesture template signature set. The threshold is automatically adjusted based on the real-time similarity distribution of the sliding window matching in actual sessions. For example, it is set to the third quantile (75th percentile) of the matching scores of the past B windows, where B represents a positive integer between 5 and 10.

[0070] Preliminary matching segments refer to the continuous time segments in the hand gesture data sliding window that are determined to possibly correspond to a certain gesture after feature analysis and similarity measurement with the preset gesture template signature set;

[0071] The confidence assessment result refers to the data structure used to measure the reliability of the initial matching fragment and the gesture template, as well as the executability of the fragment in the time series.

[0072] The above content will be described in detail below:

[0073] Continuous hand posture data is captured sequentially in a sliding window of fixed time length (e.g., 200 milliseconds). Within each sliding window, fields such as position, joint angle vector, and velocity are read. Statistical and transformation processing is performed on the hand posture data within the sliding window, including but not limited to calculating the mean, variance, velocity vector magnitude, and rate of change of adjacent joint angles, generating a feature vector for each sliding window, and using it as the corresponding feature analysis result.

[0074] The feature analysis results of each sliding window are compared with a preset set of gesture template signatures to perform a similarity measurement. The specific calculation formula is as follows:

[0075] ;

[0076] In the formula, Represents a sliding window Feature analysis results and gesture template signatures within The similarity measurement results The result of feature analysis is shown in the first... The weighting coefficients of each feature field. This represents the total number of feature fields. Represents a sliding window The results of the feature analysis are as follows: The numerical values ​​of each feature field (such as position, angle, speed, etc.). Indicates gesture template signature The Middle The signature value of each feature field. The similarity function can be represented by Euclidean distance transformation or cosine similarity function. All the above data have been normalized during the calculation.

[0077] If the similarity measurement result is greater than the preset discrimination threshold, the hand gesture data in the corresponding sliding window is marked as a preliminary matching segment.

[0078] The confidence level of the preliminary matching fragment is assessed, and the specific calculation formula is as follows:

[0079] ;

[0080] In the formula, Indicates preliminary matching fragments The confidence assessment results Indicates preliminary matching fragments The maximum value of the corresponding similarity measurement result. Indicates preliminary matching fragments The error flag count corresponds to the operation event count within the sliding window. Indicates preliminary matching fragments The success count corresponds to the operation event count within the sliding window. Indicates preliminary matching fragments The average pause time within the corresponding sliding window. , , and These represent the corresponding weighting coefficients. All the data above have been normalized during the calculation.

[0081] For each preliminary matching segment, its time window information and corresponding confidence assessment result are obtained. Then, adjacent or overlapping preliminary matching segments on the time axis are merged. The merging conditions are: the time interval between two adjacent preliminary matching segments is less than a preset time interval threshold and the confidence assessment results of the two preliminary matching segments are both greater than the preset confidence threshold. The start and end times of the merged preliminary matching segment are taken as the earliest start and latest end of the time window of the two preliminary matching segments. The confidence assessment results are calculated by weighted average. All preliminary matching segments that meet the merging conditions are merged to generate a gesture candidate set.

[0082] Read the start and end time windows and corresponding confidence evaluation results of each merged preliminary matching segment in the gesture candidate set, compare the timestamp events contained in the operation data with the merged preliminary matching segments, and determine the range of the merged preliminary matching segments in which each operation event falls.

[0083] The device confidence scores of all operational events within each merged preliminary matching segment are weighted and averaged, and then combined with the confidence score evaluation results corresponding to the merged preliminary matching segments to generate a continuous confidence sequence.

[0084] For each operation event, fields such as pause duration, error flag count, success flag, and event interval are extracted. The mean and variance are calculated based on the operation event, and then the relative anomaly factor is calculated. The formula is: relative anomaly factor = (current pause duration − mean) / variance. At the same time, the error flag count and event interval are normalized, and the normalization results are combined with the relative anomaly factor through weighted synthesis to generate the event anomaly assessment result.

[0085] This scheme utilizes sliding window feature analysis to achieve real-time capture of fine-grained dynamic hand gesture changes. By performing similarity measurement with a pre-set gesture template signature set, irrelevant actions can be effectively filtered out, thereby improving the accuracy of gesture screening. Furthermore, confidence evaluation is introduced into the initial matching segments, and windows are merged based on timestamps, enabling the system to automatically eliminate short-term misjudgments caused by local jitter, enhancing the continuity and reliability of the initial matching segments. This results in a final gesture candidate set with both high matching degree and high stability, providing a clear and robust data foundation for the interactive action recognition and process adaptation of subsequent classroom virtual experiments, thereby significantly improving the operation recognition accuracy and interaction smoothness in the virtual reality environment.

[0086] The above describes the feature analysis and template matching of the hand gesture data using a sliding window to generate a gesture candidate set. The following describes the time-series accumulation of the event anomaly assessment results and the gesture candidate set on the same operation event based on the confidence sequence, generating a risk accumulation curve. Please refer to [link / reference]. Figure 2 , Figure 2 This is a schematic diagram of the process for generating a risk accumulation curve provided in an embodiment of this application. Generating a risk accumulation curve specifically includes:

[0087] The event anomaly assessment results are matched with the timestamp of the operation event and the preliminary matching fragments of the gesture candidate set, and the matching results are weighted according to the confidence sequence to generate the initial risk value of the corresponding preliminary matching fragments;

[0088] The initial risk value is corrected based on the interval penalty and summarized in chronological order to generate a risk accumulation curve;

[0089] The interval penalty is obtained by calculating the time interval between adjacent preliminary matching segments and the preliminary matching segment, and mapping the time interval to the confidence evaluation result of the preliminary matching segment.

[0090] The initial risk value refers to the degree of risk that each gesture candidate segment may deviate or err when performing the corresponding operation event;

[0091] The interval penalty refers to the weighted impact of the time interval between two adjacent initial matching segments on the overall risk accumulation;

[0092] The time interval refers to the time difference between two adjacent preliminary matching segments.

[0093] The above content will be described in detail below:

[0094] Read the event anomaly evaluation result corresponding to each operation event in the operation data, and at the same time read the start and end time window information of the preliminary matching segment merged in the gesture candidate set, and map each operation event to the preliminary matching segment that overlaps with its time window according to the timestamp.

[0095] The initial risk value of the corresponding merged preliminary matching segment is obtained by multiplying the event anomaly assessment results of each operation event corresponding to the merged preliminary matching segment by the confidence level and averaging them.

[0096] The risk accumulation curve is generated by multiplying the interval penalty by the initial risk value and summarizing the multiplication results in chronological order.

[0097] The interval penalty is calculated by mapping the time interval between adjacent preliminary matching segments and the preliminary matching segment to the confidence evaluation result of the preliminary matching segment. The specific calculation formula is as follows:

[0098] ;

[0099] In the formula, This represents the initial matching fragment after merging. Compared to the initial matching fragment after the previous merging Interval penalty, This represents the initial matching fragment after merging. The initial matching fragment after merging with the previous one The time interval, This represents the initial matching fragment after merging. The confidence assessment results are as follows; all data above have been normalized during calculation.

[0100] This scheme achieves the quantification and dynamic reflection of operational risks by performing time-series accumulation and weighting processing on event anomaly assessment results and gesture candidate sets. Specifically, by matching the anomaly indicators of operational events with preliminary matching segments by timestamp and weighting the matching results with a confidence sequence, an initial risk value can be generated for each preliminary matching segment. This ensures that the risk assessment considers both the magnitude of the operational anomaly and the reliability of the gesture execution. Subsequently, the initial risk value is corrected through interval penalties and summarized in chronological order to form a risk accumulation curve, which can reflect the continuity and accumulation effect of risks during the operation. At the same time, by mapping the time interval with the confidence assessment results of the preliminary matching segments, the impact of adjacent preliminary matching segments on the overall risk is dynamically adjusted, achieving accurate quantification, real-time monitoring, and traceability of operational risks. This provides a reliable data foundation for subsequent path optimization, cognitive load regulation, and real-time correction.

[0101] The above describes the time-series accumulation of the event anomaly assessment results and the gesture candidate set based on the credibility sequence on the same operation event to generate a risk accumulation curve. The following describes the inverted accumulation and sorting optimization of the gesture candidate set based on the risk accumulation curve to generate a risk weight sequence. Please refer to [link / reference]. Figure 3 , Figure 3 This is a schematic diagram of the process for generating a risk weight sequence provided in an embodiment of this application. Generating the risk weight sequence specifically includes:

[0102] Using the time interval as a continuity constraint, the confidence assessment result as a validity constraint, and minimizing the peak value of the corrected initial risk value in the risk accumulation curve as the objective, the preliminary matching segments in the gesture candidate set are reordered from back to front according to the timestamp.

[0103] A risk weight sequence is generated by weighted fusion of the reordered gesture candidate set and the risk accumulation curve.

[0104] The above content will be described in detail below:

[0105] The merged initial matching segments are selected as the reordering execution targets if the time interval between adjacent merged initial matching segments is less than a preset time interval threshold and the confidence assessment results of both merged initial matching segments are greater than a preset confidence threshold. With the goal of minimizing the peak value of the corrected initial risk value in the risk accumulation curve, the objects to be reordered are reordered from back to front according to their timestamps. The specific calculation formula is as follows:

[0106] ;

[0107] In the formula, This represents the peak value of the corrected initial risk value after reordering. This represents all possible permutations of the objects to be reordered. This indicates the total number of objects to be reordered. Indicates reordering Ranked in The corrected initial risk value of the merged preliminary matching fragments. This represents the initial matching fragment after merging. The confidence assessment results are as follows; all the above data have been normalized during the calculation.

[0108] After obtaining the reordered gesture candidate set, the timestamps of each merged preliminary matching segment are used as the index. Then, the recalculated and corrected initial risk value at the corresponding time point in the risk accumulation curve is obtained, and the difference between the original corrected initial risk value and the original risk value is calculated to obtain the risk increment data. According to a unified time axis, each merged preliminary matching segment and the risk increment data in the reordered gesture candidate set are aligned to form paired data items. A weighted value is calculated for each paired data item, where the weighting factor is composed of a preset proportional coefficient (e.g., all are 0.5). All weighted values ​​are accumulated and integrated in chronological order to output a continuous risk weight sequence.

[0109] This scheme optimizes the gesture candidate set by backward accumulation and sorting based on timestamps. While maintaining operational continuity, the system can achieve refined quantification of risk information by integrating the confidence assessment results of the initial matching segments with the initial risk value. By using the weighted fusion of the reordered gesture candidate set and the risk accumulation curve, an accurate risk weight sequence can be generated. This allows the risk level of each initial matching segment in path selection and execution priority to be quantified and controllable. It helps subsequent path generation, cognitive load assessment, and real-time correction to prioritize intervention for high-risk or low-confidence operation nodes, thereby optimizing the operational safety, accuracy, and system response decision-making during virtual experiments.

[0110] The above describes the process of performing inverted cumulative sorting and optimization on the gesture candidate set based on the risk accumulation curve to generate a risk weight sequence. The following describes the statistical analysis of the operational data using a sliding window, and the calculation of the risk interaction between the load features and the gesture candidate set based on the risk weight sequence to generate a cognitive load score. Please refer to [link / reference]. Figure 4 , Figure 4 This is a schematic diagram of the process for generating a cognitive load score provided in an embodiment of this application. Generating a cognitive load score specifically includes:

[0111] The operation data is statistically analyzed using a sliding window, and the load features are mapped according to the time window of each preliminary matching segment in the reordered gesture candidate set to extract a set of load features covering the time range of the preliminary matching segments.

[0112] The risk interaction calculation is performed on the load feature set based on the risk weight sequence and the confidence sequence, and the risk interaction calculation result is weighted and normalized based on the confidence assessment result of the corresponding preliminary matching segment to generate a cognitive load score.

[0113] Among them, the load feature set refers to the set of statistical features extracted from the operation data within each sliding time window that characterizes the operation load in multiple dimensions.

[0114] Weighted normalization refers to the data processing process in which, when comprehensively evaluating the results of risk interaction calculations, the corresponding risk interaction calculation results are weighted according to the credibility of each preliminary matching segment, and the weighted values ​​are mapped to a uniform numerical range.

[0115] The above content will be described in detail below:

[0116] The operation data within each sliding window is aggregated in chronological order, and the load characteristics such as instruction frequency, instruction switching count, cursor trajectory jitter, and viewpoint dwell time distribution within the window are statistically analyzed.

[0117] Based on the start and end times of the time window corresponding to each merged preliminary matching segment in the reordered gesture candidate set, a time index interval is established. Within this time index interval, the load features are retrieved one by one, and all load features whose timestamps fall within this time index interval are screened out. They are then aggregated in chronological order to extract a set of load features covering the time range of the merged preliminary matching segments.

[0118] The risk interaction calculation is performed on the load feature set based on the risk weight sequence and the confidence sequence, and the risk interaction calculation result is weighted and normalized based on the confidence assessment result of the corresponding merged preliminary matching segments to generate a cognitive load score. The specific calculation formula is as follows:

[0119] ;

[0120] In the formula, This represents the initial matching fragment after merging. Cognitive load score, This represents the initial matching fragment after merging. The confidence assessment results This represents the total number of elements in the load characteristic set. Indicates the first The value of each load characteristic, Indicates the first Risk weights for each load feature and its corresponding timestamp Indicates the first The reliability of the timestamp corresponding to each load characteristic; all the above data have been normalized during the calculation.

[0121] The self-calibrated grading threshold is obtained by performing distribution statistics on the cognitive load scores:

[0122] The cognitive load scores are subjected to feature statistics, including calculating the mean and standard deviation of the cognitive load scores, and then constructing a self-calibrated grading threshold that includes three levels of thresholds:

[0123] The first threshold is obtained by adding a first multiple (e.g., 0.75 times) of the standard deviation of the cognitive load scores to the mean of the cognitive load scores;

[0124] The second threshold is obtained by adding a second multiple (e.g., 0.45 times) of the standard deviation of the cognitive load scores to the mean of the cognitive load scores;

[0125] The third threshold is obtained by subtracting the standard deviation of the cognitive load scores by a third factor (e.g., 0.25) from the mean cognitive load score;

[0126] Among them, the first multiple is greater than the second multiple, and the second multiple is greater than the third multiple;

[0127] The cognitive load score is compared with the self-calibration grading threshold to generate a set of load-triggered events:

[0128] If the cognitive load score is greater than the first threshold, the merged preliminary matching segment corresponding to the cognitive load score is marked as high load trigger. If it is between the second threshold and the first threshold, the merged preliminary matching segment corresponding to the cognitive load score is marked as medium load trigger. If it is between the third threshold and the second threshold, the merged preliminary matching segment corresponding to the cognitive load score is marked as low load trigger.

[0129] This scheme statistically analyzes the load characteristics of operational data using a sliding window and maps them to the time window of the initial matching segments in the gesture candidate set. This allows for precise association between time-related operational loads and each initial matching segment. By combining risk weight sequences and confidence sequences for risk interaction calculation and using confidence-weighted normalization, the generated cognitive load score not only reflects potential cognitive load changes during the operation but also demonstrates the risk sensitivity and execution reliability of each initial matching segment. This enables a dynamic and quantifiable assessment of the operational load of the target object during the virtual experiment, providing accurate data for subsequent path optimization, prompting strategies, and risk intervention.

[0130] The above describes the statistical analysis of the operational data using a sliding window to identify load characteristics, and the calculation of risk interactions between the load characteristics and the gesture candidate set based on the risk weight sequence to generate a cognitive load score. The following describes the bias correction of the load-triggered event set based on the credibility sequence to generate operational adaptation data, specifically including:

[0131] The credibility sequence, the set of load-triggered events, and the risk weight sequence are indexed and mapped by timestamps to generate the final path;

[0132] The final path is compared with the hand gesture data to generate a deviation detection event;

[0133] The cognitive load score is used to perform corrective analysis on the deviation detection events, generate corrective instructions, and then the corrective instructions and the deviation detection events are combined in a structured manner to generate operational adaptation data.

[0134] Among them, index mapping refers to the data processing process of establishing a one-to-one or many-to-one correspondence between data from different sources in chronological order based on the timestamps of various types of time series data;

[0135] The final path refers to the complete path sequence generated after multi-stage data processing by the system, which includes the execution order, time window, execution strategy, and reference association fields of each operation node;

[0136] Deviation detection events refer to a set of data in which the system detects in real time the spatial, temporal, or behavioral deviations between the target object's hand posture and the expected nodes of the final path during the execution of a virtual experiment, and records the type, intensity, and related time information of the deviation in a structured manner.

[0137] Corrective analysis refers to the process of quantitatively assessing the type, severity, and potential impact of operational deviations after they are detected, based on the timestamps of each deviation detection event and related data such as path, gesture, risk, and cognitive load, and determining the optimal corrective strategy or operational instructions.

[0138] Corrective instructions refer to operational intervention signals generated in response to real-time deviation detection events during virtual classroom experiments;

[0139] Structured composition refers to a data processing method that organizes data in a unified manner according to a clear data structure and logical relationship.

[0140] The above content will be described in detail below:

[0141] The credibility of each merged preliminary matching segment is read from the credibility sequence using the timestamp as the primary key, and the risk weight at the same time position is retrieved synchronously from the risk weight sequence. The two are then weighted segment by segment according to a preset weighting rule (e.g., the weighting factor of credibility is 0.7 and the weighting factor of risk weight is 0.3) to form a weighted scoring table. The weighted scoring table is then sorted from high to low according to the score value, and the top A merged preliminary matching segments are selected and marked as path nodes based on the sorting results.

[0142] Where A represents a positive integer between 10 and 15;

[0143] Using the time window of each load triggering event in the load triggering event set as the retrieval condition, after removing the merged preliminary matching segments corresponding to the path nodes, the time range of the segments is compared one by one from the remaining merged preliminary matching segments. The merged preliminary matching segments that overlap with any time window are selected and recorded as candidate nodes. The path nodes and the candidate nodes are re-indexed and merged according to the timestamp to form a path candidate set.

[0144] Constrained by the interval penalty and the confidence evaluation result, the path candidate set is... Perform planning and solving to generate the final path. The specific calculation formula is as follows:

[0145]

[0146] ;

[0147] In the formula, Indicates the possible operation execution path. This represents the total number of nodes in the candidate path set. Represents a node The rating, Represents a node and nodes The corresponding gap penalty between the merged initial matching segments, This represents the weighting coefficient of the interval penalty. Represents a node The confidence assessment results This represents the preset confidence threshold. Represents a node Cognitive load score, This represents the preset upper limit of cognitive load. All the above data have been normalized during the calculation.

[0148] The difference between the final path and the hand posture data is calculated to generate a comparison result. It is determined whether the comparison result is greater than a preset filtering threshold. If so, the comparison result, the corresponding final path, and the hand posture data are integrated to generate a deviation detection event.

[0149] The preset screening threshold is composed of the confidence assessment results and credibility of the same merged preliminary matching segments according to a preset weighting coefficient (e.g., both are 0.5).

[0150] Using the final path as a reference, and based on the comparison results and the cognitive load score, combined with the self-calibration grading threshold, the deviation detection events are stratified and conditionally judged to generate correction instructions, including:

[0151] The comparison results of each node corresponding to the final path are used to generate deviation stratification detection thresholds in a synchronous manner according to the method of generating self-calibration grading thresholds, and the same method is used to judge and generate high deviation, medium deviation and low deviation.

[0152] When the deviation detection event corresponds to a high deviation at the node in the final path, a "force rollback and reboot" instruction is generated.

[0153] If the deviation detection event corresponds to a medium deviation node in the final path and the preliminary matching segment after merging corresponding to the node is triggered by high load, then a "delay and prompt" instruction will be generated.

[0154] If the deviation detection event corresponds to a low deviation node in the final path, a "smoothly replace with alternative node" instruction is generated.

[0155] The correction instructions and the deviation detection events are combined in a structured manner to generate operation adaptation data.

[0156] This solution indexes and maps the credibility sequence, load triggering event set, and risk weight sequence by timestamp, enabling precise association of the execution reliability, potential risks, and cognitive load of each initially matched segment. This generates a highly consistent and continuous final path, making the operation sequence traceable in both time and logic. Furthermore, the final path is compared with hand gesture data to identify deviation detection events in real time, ensuring that abnormal operations can be captured immediately. Subsequently, based on the cognitive load score, the deviation detection events are analyzed for correction, generating targeted correction instructions. The correction instructions and deviation detection events are then structurally combined to form operation adaptation data, achieving dynamic control and closed-loop optimization of the virtual experiment operation process.

[0157] The above describes the deviation correction of the load triggering event set based on the credibility sequence to generate operation adaptation data. The following describes the indexing and mapping of the credibility sequence, the load triggering event set, and the risk weight sequence by timestamp to generate the final path, specifically including:

[0158] Based on the credibility sequence and the risk weight sequence, the preliminary matching segments are weighted and sorted, and the top A segments are selected as path nodes. At the same time, the preliminary matching segments that overlap with the time window of the load triggering event set are retrieved as candidate nodes. The selected path nodes and the candidate nodes are integrated to obtain a path candidate set.

[0159] Using the interval penalty and the confidence evaluation result as constraints, the path candidate set is planned and solved to generate the final path.

[0160] Weighted sorting refers to the data processing process of sorting the initially matched segments from high to low based on their comprehensive scores on the time axis.

[0161] A path node is a structured data unit that represents an executable operation or a preliminary matching segment on the timeline.

[0162] Alternate nodes refer to preliminary matching segments that were not selected as path nodes in the initial matching segments, but overlap with the time window of the load triggering event set;

[0163] The path candidate set refers to the set of candidate nodes that can be selected for path planning in the virtual experiment operation process, based on existing preliminary matching fragments, credibility, risk weight, and cognitive load information.

[0164] Planning and solving refers to the data processing process that, based on existing path candidate sets, confidence assessment results, risk weights, and cognitive load, combines nodes into a time series path by establishing constraints and objective functions, so that the path generates a unique or optimal final path under the conditions of continuity, confidence, risk, and load balance.

[0165] This part has already been described in detail above, so I will not repeat it here.

[0166] This scheme achieves precise sorting and filtering of preliminary matching segments by indexing and mapping the credibility sequence, load triggering event set, and risk weight sequence by timestamp. This ensures that path nodes not only consider the execution reliability of preliminary matching segments but also take into account preventive intervention during high-risk and high-load periods. By integrating the sorted path nodes with the candidate preliminary matching segments within the load triggering time window to generate a path candidate set, and introducing interval penalties and confidence constraints during the planning and solving process, the scheme achieves optimized control over node continuity and execution credibility. As a result, the generated final path maintains the logical order of operations while minimizing execution risks, thereby improving the safety, reliability, and adaptability of the virtual classroom experiment.

[0167] The above describes the indexing and mapping of the credibility sequence, the load triggering event set, and the risk weight sequence by timestamp to generate the final path. The following describes comparing the final path with the hand gesture data to generate a deviation detection event, specifically including:

[0168] The final path is compared with the hand gesture data, and it is determined whether the comparison result is greater than a preset filtering threshold. If so, the comparison result, the corresponding final path, and the hand gesture data are integrated to generate a deviation detection event.

[0169] The preset screening threshold is obtained by fusing the confidence assessment results and the confidence sequence.

[0170] The comparison result refers to the quantitative value of the difference between the expected hand position / posture of each node in the final path and the real-time hand posture data in the spatial and posture dimensions.

[0171] The preset filtering threshold refers to the numerical limit used to determine whether the deviation between the hand gesture data and the final path needs to trigger a deviation detection event.

[0172] This part has already been described in detail above, so I will not repeat it here.

[0173] This solution achieves accurate identification of deviations from expected operation of the target object by comparing the final path with hand gesture data in real time and judging the deviation based on a preset screening threshold. The process of generating deviation detection events integrates the comparison results, final path, and hand gesture data into a structured event record, enabling the system to capture operational anomalies in real time and provide traceable data evidence. At the same time, the screening threshold is dynamically calculated by fusing the confidence assessment results and the credibility sequence, making the deviation judgment adaptive to different operation quality and target object characteristics, thereby improving the accuracy, reliability, and response speed of deviation detection in virtual experiments and realizing real-time monitoring and closed-loop management of the operation process.

[0174] The above describes comparing the final path with the hand gesture data to generate a deviation detection event. The following describes the corrective analysis of the deviation detection event based on the cognitive load score to generate corrective instructions, specifically including:

[0175] Using the final path as a reference, and based on the comparison results and the cognitive load score, the deviation detection event is stratified and judged according to the self-calibration grading threshold, and a correction instruction is generated.

[0176] Among them, the hierarchical condition judgment refers to a data processing rule that, in response to deviation events detected in real time during classroom virtual experiments, combines cognitive load scores and self-calibration thresholds to perform hierarchical processing of the generation of correction instructions.

[0177] This part has already been described in detail above, so I will not repeat it here.

[0178] This scheme uses cognitive load scoring to quantify the load status of each path node and combines it with self-calibration grading thresholds to perform hierarchical condition judgment on deviation detection events. This enables the generated correction instructions to automatically select the optimal intervention strategy for different deviation types and operational load levels. At the same time, it ensures that the correction instructions strictly correspond to the node information and time window of the final path, realizing real-time adaptive adjustment and closed-loop control of the target object operation, thereby improving the operational safety, real-time performance, and path execution accuracy of the virtual experiment.

[0179] Example 2:

[0180] Please see Figure 5 A virtual reality operation process adaptation system for classroom virtual experiments, including:

[0181] The data acquisition module is used to acquire the hand gesture data and operation data of the target object;

[0182] The credibility assessment module is used to perform feature analysis and template matching on the hand gesture data by sliding window, generate a gesture candidate set, and perform time mapping on the operation data based on the gesture candidate set to generate a credibility sequence.

[0183] The risk assessment module is used to perform event anomaly assessment on the operation data by segmenting it according to operation events, and to perform time-series accumulation of the event anomaly assessment results and the gesture candidate set on the same operation event according to the credibility sequence to generate a risk accumulation curve. Furthermore, the module performs inverted accumulation and sorting optimization on the gesture candidate set according to the risk accumulation curve to generate a risk weight sequence.

[0184] The load detection module is used to statistically analyze the load characteristics of the operation data using a sliding window, and to perform risk interaction calculations on the load characteristics and the gesture candidate set according to the risk weight sequence to generate a cognitive load score. Then, the cognitive load score is compared with a self-calibration grading threshold to generate a set of load triggering events.

[0185] An operation adaptation processing module is used to perform deviation correction on the load triggering event set according to the credibility sequence and generate operation adaptation data.

[0186] This embodiment has the same technical effects as Embodiment 1.

[0187] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. The data mentioned in this application have undergone normalization and other preprocessing to unify dimensions during formula calculations.

[0188] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for adapting virtual reality operation procedures for classroom virtual experiments, characterized in that, Includes the following steps: Acquire the hand gesture data and operation data of the target object; The hand gesture data is subjected to feature analysis and template matching by sliding window to generate a gesture candidate set, and the operation data is time-mapped according to the gesture candidate set to generate a confidence sequence; The operation data is segmented by operation event for event anomaly assessment. Based on the confidence sequence, the event anomaly assessment results and the gesture candidate set are accumulated in time series on the same operation event to generate a risk accumulation curve. Furthermore, based on the risk accumulation curve, the gesture candidate set is subjected to inverted accumulation and sorting optimization to generate a risk weight sequence. The operation data is statistically analyzed using a sliding window to determine the load characteristics. Based on the risk weight sequence, the load characteristics and the gesture candidate set are used to perform risk interaction calculations to generate a cognitive load score. The cognitive load score is then compared with a self-calibration grading threshold to generate a set of load-triggered events. The self-calibration grading threshold is obtained by performing distribution statistics on the cognitive load score; Based on the credibility sequence, the set of load triggering events is corrected for deviations to generate operation adaptation data.

2. The virtual reality operation process adaptation method for classroom virtual experiments according to claim 1, characterized in that: The hand gesture data is subjected to feature analysis and template matching using a sliding window to generate a gesture candidate set, specifically including: The hand gesture data is analyzed by sliding window, and the feature analysis results of each sliding window are compared with a preset gesture template signature set for similarity measurement. It is determined whether the similarity measurement result is greater than a preset discrimination threshold. If so, the hand gesture data in the corresponding sliding window is marked as a preliminary matching segment. The confidence level of the preliminary matching segments is evaluated, and then the preliminary matching segments are merged into windows based on the confidence level evaluation results and the start and end timestamps of the preliminary matching segments to generate a gesture candidate set.

3. The virtual reality operation process adaptation method for classroom virtual experiments according to claim 2, characterized in that: Based on the confidence sequence, the event anomaly assessment results and the gesture candidate set are time-series accumulated over the same operation event to generate a risk accumulation curve, specifically including: The event anomaly assessment results are matched with the timestamp of the operation event and the preliminary matching fragments of the gesture candidate set, and the matching results are weighted according to the confidence sequence to generate the initial risk value of the corresponding preliminary matching fragments; The initial risk value is corrected based on the interval penalty and summarized in chronological order to generate a risk accumulation curve; The interval penalty is obtained by calculating the time interval between adjacent preliminary matching segments and the preliminary matching segment, and mapping the time interval to the confidence evaluation result of the preliminary matching segment.

4. The virtual reality operation process adaptation method for classroom virtual experiments according to claim 3, characterized in that: The risk weight sequence is generated by performing inverted cumulative sorting and optimization on the gesture candidate set based on the risk accumulation curve, specifically including: Using the time interval as a continuity constraint, the confidence assessment result as a validity constraint, and minimizing the peak value of the corrected initial risk value in the risk accumulation curve as the objective, the preliminary matching segments in the gesture candidate set are reordered from back to front according to the timestamp. A risk weight sequence is generated by weighted fusion of the reordered gesture candidate set and the risk accumulation curve.

5. The virtual reality operation process adaptation method for classroom virtual experiments according to claim 2, characterized in that: The cognitive load score is generated by statistically analyzing the operational data using a sliding window method and performing risk interaction calculations on the load features and the gesture candidate set based on the risk weight sequence. Specifically, this includes: The operation data is statistically analyzed using a sliding window, and the load features are mapped according to the time window of each preliminary matching segment in the reordered gesture candidate set to extract a set of load features covering the time range of the preliminary matching segments. The risk interaction calculation is performed on the load feature set based on the risk weight sequence and the confidence sequence, and the risk interaction calculation result is weighted and normalized based on the confidence assessment result of the corresponding preliminary matching segment to generate a cognitive load score.

6. The virtual reality operation process adaptation method for classroom virtual experiments according to claim 4, characterized in that: The deviation correction of the load triggering event set based on the confidence sequence, and the generation of operation adaptation data specifically include: The credibility sequence, the set of load-triggered events, and the risk weight sequence are indexed and mapped by timestamps to generate the final path; The final path is compared with the hand gesture data to generate a deviation detection event; The cognitive load score is used to perform corrective analysis on the deviation detection events, generate corrective instructions, and then the corrective instructions and the deviation detection events are combined in a structured manner to generate operational adaptation data.

7. The virtual reality operation process adaptation method for classroom virtual experiments according to claim 6, characterized in that: The process of indexing and mapping the credibility sequence, the set of load-triggered events, and the risk weight sequence by timestamp to generate the final path specifically includes: Based on the credibility sequence and the risk weight sequence, the preliminary matching segments are weighted and sorted, and the top A segments are selected as path nodes. At the same time, the preliminary matching segments that overlap with the time window of the load triggering event set are retrieved as candidate nodes. The selected path nodes and the candidate nodes are integrated to obtain a path candidate set. Using the interval penalty and the confidence evaluation result as constraints, the path candidate set is planned and solved to generate the final path.

8. The virtual reality operation process adaptation method for classroom virtual experiments according to claim 6, characterized in that: The process of comparing the final path with the hand gesture data to generate a deviation detection event specifically includes: The final path is compared with the hand gesture data, and it is determined whether the comparison result is greater than a preset filtering threshold. If so, the comparison result, the corresponding final path, and the hand gesture data are integrated to generate a deviation detection event. The preset screening threshold is obtained by fusing the confidence assessment results and the confidence sequence.

9. The virtual reality operation process adaptation method for classroom virtual experiments according to claim 8, characterized in that: Based on the cognitive load score, a corrective analysis is performed on the deviation detection events to generate corrective instructions, specifically including: Using the final path as a reference, and based on the comparison results and the cognitive load score, the deviation detection event is stratified and judged according to the self-calibration grading threshold, and a correction instruction is generated.

10. A virtual reality operation process adaptation system for classroom virtual experiments, characterized in that, include: The data acquisition module is used to acquire the hand gesture data and operation data of the target object; The credibility assessment module is used to perform feature analysis and template matching on the hand gesture data by sliding window, generate a gesture candidate set, and perform time mapping on the operation data based on the gesture candidate set to generate a credibility sequence. The risk assessment module is used to perform event anomaly assessment on the operation data by segmenting it according to operation events, and to perform time-series accumulation of the event anomaly assessment results and the gesture candidate set on the same operation event according to the credibility sequence to generate a risk accumulation curve. Furthermore, the module performs inverted accumulation and sorting optimization on the gesture candidate set according to the risk accumulation curve to generate a risk weight sequence. The load detection module is used to statistically analyze the load characteristics of the operation data using a sliding window, and to perform risk interaction calculations on the load characteristics and the gesture candidate set according to the risk weight sequence to generate a cognitive load score. Then, the cognitive load score is compared with a self-calibration grading threshold to generate a set of load triggering events. An operation adaptation processing module is used to perform deviation correction on the load triggering event set according to the credibility sequence and generate operation adaptation data.

Images