Classroom simulation method and apparatus for virtual reality

Abstract

The application discloses a kind of virtual reality classroom simulation method and device, belong to the technical field of intelligent education, method includes: obtaining multi-modal physiological signal;Rhythm characteristic parameter is extracted based on multi-modal physiological signal, establishes group cognitive rhythm model, predicts the cognitive rhythm phase evolution trend in future time window;Monitoring the understanding state of current knowledge point of learner group, identify weak knowledge points;Based on the feature of weak knowledge point, the cognitive rhythm phase of current learner group and the phase evolution trend of future phase, determine time anchor interference strategy;In the case where time anchor interference strategy is determined to implement time anchor interference, the subjective time perception of learner is regulated by VR multi-sensory control means;Assess the learning effect after the implementation of time anchor interference strategy, obtain evaluation result, and update the parameters of group cognitive rhythm model based on evaluation result.It can respond in time effectively, improve control efficiency and teaching effect.

Description

Technical Field

[0001] This invention belongs to the technical field of intelligent education, and specifically relates to a virtual reality classroom simulation method and device. Background Technology

[0002] Currently, virtual reality teaching is widely used in online courses. Students can learn knowledge courses in an immersive environment by watching virtual reality classrooms. This not only helps to improve students' learning interest, but also helps to improve their learning efficiency.

[0003] Currently, existing virtual reality classroom teaching systems typically adjust the difficulty, presentation method, or pace of teaching content by collecting learners' interaction data (such as answer accuracy, operation trajectory, and gaze point) to assess their learning status. For example, some patents dynamically adjust the presentation frequency and time intervals of content by calculating the user's level of mastery and combining this with an interval repetition algorithm.

[0004] In the aforementioned technologies, adjustments are made only after a decline in learning status is detected. This "remedial" approach inherently suffers from a lag, resulting in the teaching system being unable to respond promptly and effectively. The system's control is untimely and inefficient, failing to provide preventative intervention before cognitive decline occurs, thus impacting teaching effectiveness. Summary of the Invention

[0005] This application provides a virtual reality classroom simulation method and apparatus that can respond promptly and effectively, improving control efficiency and teaching effectiveness.

[0006] On one hand, embodiments of this application provide a virtual reality classroom simulation method, the method comprising:

[0007] Acquire multimodal physiological signals of learners in a virtual classroom;

[0008] Based on the multimodal physiological signals, rhythmic feature parameters are extracted, a group cognitive rhythm model is established, the current group's cognitive rhythm phase is calculated, and the evolution trend of the cognitive rhythm phase within future time windows is predicted; the cognitive rhythm phase includes at least the attention rise phase, attention plateau phase, attention decline phase, and attention trough phase;

[0009] Monitor learners' understanding of current knowledge points and identify weak knowledge points;

[0010] Based on the characteristics of the weak knowledge points, the cognitive rhythm phase of the current learner group, and the phase evolution trend of the future phase, a time anchor interference strategy is determined.

[0011] When the time anchor interference strategy is determined to be implemented, the learner's subjective time perception is adjusted by VR multi-sensory control methods, so that the subjective dwell time of the weak knowledge points is stretched and / or the subjective dwell time of the mastered knowledge points is compressed; the VR multi-sensory control methods include one or more of visual refresh rate gradient modulation, spatial reference system displacement, multi-sensory delay modulation and dynamic time reference system embedding.

[0012] The learning effect after the implementation of the time anchor interference strategy is evaluated, the evaluation results are obtained, and the parameters of the group cognitive rhythm model are updated based on the evaluation results.

[0013] Optionally, the multimodal physiological signals include heart rate variability signals, skin conductance signals, and eye movement data, and the extraction of rhythmic feature parameters based on the multimodal physiological signals includes:

[0014] Frequency domain analysis was performed on the heart rate variability signal to extract the ratio of low-frequency components to high-frequency components as an indicator of the activity state of the autonomic nervous system.

[0015] The frequency and amplitude of phase-dependent skin conductance responses were extracted from skin conductance signals and used as an indicator of alertness.

[0016] The fluctuation frequency of pupil diameter and the distribution characteristics of fixation duration were extracted from eye-tracking data and used as indicators of attention concentration.

[0017] The indicators of the autonomic nervous system activity state, the alertness index, and the attention concentration index are input into a pre-trained cognitive rhythm phase classification model to obtain the cognitive rhythm phase of the current group.

[0018] Optionally, calculating the cognitive rhythm phase of the current group includes:

[0019] The multimodal physiological signals are subjected to time window sliding sampling, and multidimensional rhythmic features at the population level are extracted within each sliding window; the multidimensional rhythmic features include at least: heart rate variability frequency domain features, skin conductance time domain features, and eye movement spatial features;

[0020] The multidimensional rhythm features are input into the corresponding phase probability evaluation sub-models to obtain the phase probability distribution of each mode independently determined;

[0021] The phase probability distributions of each modality, determined independently, are weighted and fused according to a preset dynamic confidence weight to obtain the cognitive rhythm phase of the current group. The dynamic confidence weight is dynamically adjusted according to the signal-to-noise ratio of each modality's physiological signal within the current time window, with higher fusion weights assigned to modalities with higher signal-to-noise ratios.

[0022] Optionally, the prediction of the cognitive rhythm phase evolution trend within a future time window includes:

[0023] Using the historical phase sequence of N consecutive time windows preceding the current time window as input, the time series prediction model outputs the predicted probability distribution of each phase within the future time window and the predicted value of the phase transition time.

[0024] The predicted value of the phase transition time includes at least: the expected duration of the current phase, the expected arrival time of the next phase, and the transition probability from the current phase to the next phase;

[0025] The time series prediction model also outputs the prediction confidence interval of the predicted value, which is dynamically calculated based on the goodness of fit of the current group cognitive rhythm model and the historical prediction error.

[0026] Based on the predicted confidence interval, the cognitive phase evolution trend is determined.

[0027] Optionally, monitoring the learners' understanding of the current knowledge point and identifying weak knowledge points includes:

[0028] Obtain the individual comprehension level index of each learner in the group at the current knowledge point;

[0029] The statistical characteristics of the group-level understanding state are calculated based on the individual understanding index; the statistical characteristics of the group-level understanding state include at least one of the group understanding mean, group understanding variance, group understanding entropy value, and individual deviation value between the individual understanding index and the group understanding mean; the group understanding entropy value is used to quantify the consistency of the group's solution state in the current knowledge point.

[0030] Based on the statistical characteristics of the understanding state at the group level, the weakness type of the current knowledge point is determined; the weakness type includes at least one of the following: common weak knowledge points when the average group understanding level is lower than the first threshold and the group understanding entropy value is lower than the second threshold, differentiated weak knowledge points when the average group understanding level is lower than the first threshold and the group understanding entropy value is higher than the third threshold, and individual weakness markers corresponding to individual learners whose individual deviation values ​​exceed the preset individual deviation threshold.

[0031] For the common weak knowledge points, a unified time anchor interference strategy is initiated for the group; for the differentiated weak knowledge points, individual differential parameters are superimposed on the unified time anchor interference for the group; for the individual weak markers, individualized interference parameters are generated for the corresponding individual learners.

[0032] Optionally, the step of determining the time anchor interference strategy based on the characteristics of the weak knowledge points, the cognitive rhythm phase of the current learner group, and the phase evolution trend of future phases includes:

[0033] A multidimensional cognitive state space is constructed with the features of cognitive rhythm phase and weak knowledge points as coordinate axes; the multidimensional cognitive state space includes at least the cognitive rhythm phase dimension and the knowledge point cognitive complexity dimension;

[0034] The prediction results of the cognitive rhythm phase and future phase of the current learner group are mapped to the multidimensional cognitive state space to determine the position coordinates and movement trajectory of the current group in the cognitive state space.

[0035] Based on the location coordinates and motion trajectory, the corresponding optimal time anchor interference strategy is queried from the preset cognitive state space decision graph; the cognitive state space decision graph is a decision graph that is obtained in advance through reinforcement learning training and represents the mapping relationship between each region of the cognitive state space and the optimal interference strategy.

[0036] Optionally, the construction of a multidimensional cognitive state space with the features of cognitive rhythm phases and weak knowledge points as coordinate axes includes:

[0037] Construct a virtual classroom simulation environment, in which multiple simulated teaching sessions are conducted with a simulated group of learners;

[0038] In each simulated teaching session, the position coordinates in the cognitive state space are used as the state input, the parameters of each dimension of the time anchor interference strategy are used as the action output, and the learning effect gain value after the simulated teaching is used as the reward signal.

[0039] A deep reinforcement learning algorithm is used to train the policy network, which takes the spatial coordinates of the cognitive state as input and outputs the optimal parameter combination of each dimension of the time anchor interference policy.

[0040] After training is completed, the input-output mapping relationship of the policy network is solidified into the multidimensional cognitive state space.

[0041] Optionally, the evaluation of the learning effect after implementing the time anchor interference strategy, and the resulting evaluation, includes:

[0042] At the first preset time point after the implementation of the time anchor interference strategy, the learner group’s first comprehension index of the weak knowledge points is collected as an immediate evaluation result.

[0043] At a second preset time point after the implementation of the time anchor interference strategy, the learner group's second level of understanding of the weak knowledge points is collected as the delayed evaluation result; the second preset time point is later than the first preset time point.

[0044] Based on the immediate evaluation results and the delayed evaluation results, the comprehensive learning effect evaluation value of the time anchor interference strategy is calculated.

[0045] Optionally, the comprehensive calculation of the overall learning performance evaluation value of the time anchor interference strategy includes:

[0046] Obtain the preset forgetting curve parameters corresponding to the weak knowledge points. The forgetting curve parameters are pre-calibrated based on the knowledge point type and cognitive complexity level of the weak knowledge points.

[0047] Based on the immediate assessment results, the delayed assessment results, and the preset forgetting curve parameters, the actual forgetting rate of the weak knowledge points is calculated.

[0048] The actual forgetting rate is compared with the standard forgetting rate in the preset forgetting curve parameters to obtain the forgetting inhibition effect coefficient;

[0049] Based on the immediate evaluation results and the forgetting inhibition effect coefficient, the comprehensive learning effect evaluation value is calculated.

[0050] On the other hand, embodiments of this application provide a virtual reality classroom simulation method apparatus, the apparatus comprising:

[0051] The acquisition module is used to acquire multimodal physiological signals of learners in the virtual classroom.

[0052] A module is established to extract rhythmic feature parameters based on the multimodal physiological signals, establish a group cognitive rhythm model, calculate the current group's cognitive rhythm phase, and predict the evolution trend of the cognitive rhythm phase within a future time window; the cognitive rhythm phase includes at least the attention rise phase, attention plateau phase, attention decline phase, and attention trough phase;

[0053] The identification module is used to monitor the learners' understanding of the current knowledge points and identify weak knowledge points;

[0054] The determination module is used to determine the time anchor interference strategy based on the characteristics of the weak knowledge points, the cognitive rhythm phase of the current learner group, and the phase evolution trend of the future phase.

[0055] The control module is used to control the learner's subjective time perception through VR multi-sensory manipulation when the time anchor interference strategy is determined to be implemented, so that the subjective dwell time of the weak knowledge points is stretched and / or the subjective dwell time of the mastered knowledge points is compressed; the VR multi-sensory manipulation means include one or more of visual refresh rate gradient modulation, spatial reference system displacement, multi-sensory delay modulation and dynamic time reference system embedding.

[0056] The evaluation module is used to evaluate the learning effect after the implementation of the time anchor interference strategy, obtain the evaluation results, and update the parameters of the group cognitive rhythm model based on the evaluation results.

[0057] In another aspect, embodiments of this application provide an electronic device, the device comprising: a processor and a memory storing computer program instructions;

[0058] When the processor executes the computer program instructions, it implements the virtual reality classroom simulation method as described in the first aspect.

[0059] In another aspect, embodiments of this application provide a computer storage medium storing computer program instructions, which, when executed by a processor, implement the virtual reality classroom simulation method as described in the first aspect.

[0060] In another aspect, embodiments of this application provide a computer program product in which instructions, when executed by a processor of an electronic device, cause the electronic device to perform the virtual reality classroom simulation method as described in the first aspect.

[0061] The virtual reality classroom simulation method and apparatus of this application first acquires multimodal physiological signals of a group of learners in a virtual classroom. Rhythmic feature parameters are extracted from these signals to establish a group cognitive rhythm model, which identifies the cognitive phase evolution trend within a future event window and identifies weak knowledge points. Then, based on the characteristics of the weak knowledge points, the current cognitive rhythm phase of the learner group, and the phase evolution trend of the future phase, a time anchor interference strategy is determined. This strategy actively regulates learners' subjective time perception, enhances their learning efficiency, and achieves preventative intervention in the learning process. It avoids the lag of traditional "post-hoc remediation" models, thereby improving the timeliness, effectiveness, and learning outcomes of teaching in a virtual reality classroom environment. Attached Figure Description

[0062] To more clearly illustrate the technical solutions in the embodiments of this specification 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 recorded in the embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings.

[0063] Figure 1 This is a flowchart illustrating a virtual reality classroom simulation method provided in one embodiment of this application;

[0064] Figure 2This is a schematic diagram of the structure of a virtual reality classroom simulation device provided in another embodiment of this application;

[0065] Figure 3 This is a schematic diagram of the structure of an electronic device provided in another embodiment of this application. Detailed Implementation

[0066] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0067] To facilitate understanding of the embodiments of this application, further explanation and description will be provided below with reference to the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of this application. In the drawings, the dimensions and relative dimensions of components may be exaggerated for clarity and / or descriptive purposes. When exemplary embodiments can be implemented differently, a specific process sequence may be performed in a different order than that described. For example, two consecutively described processes may be performed substantially simultaneously or in the reverse order of their description. Furthermore, the same reference numerals denote the same components.

[0068] The terminology used herein is for the purpose of describing particular embodiments and is not intended to be limiting. As used herein, unless the context clearly indicates otherwise, the singular forms “a” and “the” are intended to include the plural forms as well. Furthermore, when the terms “comprising” and / or “including” and variations thereof are used in this specification, it indicates the presence of the stated features, integrals, steps, operations, parts, components, and / or groups thereof, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, parts, components, and / or groups thereof. It should also be noted that, as used herein, the terms “substantially,” “about,” and other similar terms are used as approximate terms rather than as terms of degree, thus explaining the inherent biases in measurements, calculated values, and / or provided values ​​that would be recognized by one of ordinary skill in the art.

[0069] To address the problems of existing technologies, this application provides a virtual reality classroom simulation method and apparatus. During classroom simulation, multimodal physiological signals of the learner group in the virtual classroom are first acquired. Rhythmic feature parameters are extracted from these signals to establish a group cognitive rhythm model. This model identifies the cognitive phase evolution trend within a future event window and identifies weak knowledge points. Then, based on the characteristics of these weak knowledge points, the current cognitive rhythm phase of the learner group, and the phase evolution trend of future phases, a time anchor interference strategy is determined. This strategy actively regulates learners' subjective time perception, enhancing their learning efficiency and achieving preventative intervention in the learning process. It avoids the lag of traditional "post-hoc remediation" models, thereby improving the timeliness, effectiveness, and learning outcomes of teaching in a virtual reality classroom environment.

[0070] The virtual reality classroom simulation method provided in the embodiments of this application will be introduced first.

[0071] Figure 1 A flowchart illustrating a virtual reality classroom simulation method according to an embodiment of this application is shown. Figure 1 As shown, virtual reality classroom simulation methods may include S101-S106:

[0072] S101, to acquire multimodal physiological signals of learners in a virtual classroom.

[0073] In this embodiment, sensors can be incorporated into a virtual reality headset to measure the learner's heart rate data. Alternatively, learners can be asked to record their perception of fatigue levels during the learning process and input this as a physiological signal. These multimodal physiological signals may include heart rate variability (HRV) signals, skin conductance (GSR) signals, eye movement data (pupil diameter changes, fixation duration, scan rate), electroencephalogram (EEG) signals, and head movement trajectory and posture data.

[0074] S102, based on multimodal physiological signals, extracts rhythmic feature parameters, establishes a group cognitive rhythm model, calculates the current group's cognitive rhythm phase, and predicts the evolution trend of the cognitive rhythm phase within future time windows.

[0075] In this embodiment, the cognitive rhythm phase includes at least the attention rise phase, attention plateau phase, attention decline phase, and attention trough phase. The acquired heart rate data can be averaged, and based on the trend of the average heart rate, the learner's cognitive state can be divided into high and low heart rate ranges, which are then used as cognitive rhythm phases. For predicting future phases, a linear regression model can be used to predict the heart rate trend at subsequent time points based on the average heart rate at past time points.

[0076] S103 monitors the learners' understanding of the current knowledge points and identifies weak knowledge points.

[0077] In this embodiment, learners may be required to take a test using multiple-choice or fill-in-the-blank questions after completing the learning of a knowledge point, and their understanding of the knowledge point may be judged based on the accuracy of their answers. When the accuracy rate of the learners' group is lower than a preset threshold, the knowledge point is identified as a weak knowledge point.

[0078] S104. Based on the characteristics of weak knowledge points, the cognitive rhythm phase of the current learner group, and the phase evolution trend of future phases, determine the time anchor interference strategy.

[0079] In this embodiment, a set of rules can be pre-defined: when a weak knowledge point appears and the group is in a period of attention decline, a time-stretching strategy is triggered; when the group is in a period of attention rise, no interference is performed. The determination of this strategy can be based solely on phase judgment without considering the cognitive complexity of the weak knowledge point.

[0080] S105, when the time anchor interference strategy is determined to be implemented, the learner's subjective time perception is adjusted through VR multi-sensory control methods, so that the subjective dwell time of weak knowledge points is stretched and / or the subjective dwell time of mastered knowledge points is compressed.

[0081] In this embodiment, VR multi-sensory control methods include one or more of visual refresh rate gradient modulation, spatial reference frame displacement, multi-sensory delay modulation, and dynamic time reference frame embedding. When it is necessary to extend the subjective dwell time on weak knowledge points, the visual refresh rate of the virtual scene can be reduced, making the learner perceive the passage of time as slower. When it is necessary to compress the subjective dwell time on mastered knowledge points, the visual refresh rate can be increased. These adjustments can be made at predetermined amplitudes and do not vary with individual learners or the characteristics of knowledge points.

[0082] S106. Evaluate the learning effect after the implementation of the time anchor interference strategy, obtain the evaluation results, and update the parameters of the group cognitive rhythm model based on the evaluation results.

[0083] After implementing the time-anchored interference strategy, a knowledge point test can be conducted again, and the learners' correct answer rate can be recorded as an evaluation result of learning effectiveness. This evaluation result can be used to adjust the parameters in the group cognitive rhythm model in order to adapt to the learners' state in future teaching.

[0084] During classroom simulation, multimodal physiological signals of learners in the virtual classroom are first acquired. Rhythmic characteristic parameters are extracted from these signals to establish a group cognitive rhythm model. This model identifies the cognitive phase evolution trend within future event windows and identifies weak knowledge points. Subsequently, based on the characteristics of weak knowledge points, the current cognitive rhythm phase of the learner group, and the phase evolution trend of future phases, a time anchor interference strategy is determined. This strategy actively regulates learners' subjective time perception, enhances their learning efficiency, and achieves preventative intervention in the learning process. It avoids the lag of traditional "post-remedial" models, thereby improving the timeliness, effectiveness, and learning outcomes of teaching in a virtual reality classroom environment.

[0085] In some other embodiments, S102 may include:

[0086] Frequency domain analysis was performed on the heart rate variability signal to extract the ratio of low-frequency components to high-frequency components as an indicator of the activity state of the autonomic nervous system.

[0087] The frequency and amplitude of phase-dependent skin conductance responses were extracted from skin conductance signals and used as an indicator of alertness.

[0088] The fluctuation frequency of pupil diameter and the distribution characteristics of fixation duration were extracted from eye-tracking data and used as indicators of attention concentration.

[0089] By inputting indicators of autonomic nervous system activity, alertness, and attention concentration into a pre-trained cognitive rhythm phase classification model, the cognitive rhythm phase of the current group can be obtained.

[0090] In this embodiment, the selection of multimodal physiological signals is based on their strong correlation with the learner's cognitive state. Heart rate variability signals can reflect the activity state of the autonomic nervous system, and their variation patterns are closely related to cognitive load, stress, and attention levels. Skin conductance signals directly reflect the arousal level of the sympathetic nervous system and are closely linked to emotional excitement, alertness, and cognitive effort. Eye-tracking data, as a non-invasive indicator, can directly reveal the learner's visual attention and cognitive processing; for example, changes in pupil diameter are related to cognitive load, and fixation duration reflects attention concentration. These signals provide rich information about the learner's cognitive state from different physiological and behavioral dimensions.

[0091] In extracting rhythmic characteristic parameters, frequency domain analysis is performed on heart rate variability signals to transform them from the time domain to the frequency domain, thereby separating the energy distribution of different frequency components. The ratio of low-frequency (LF) to high-frequency (HF) components (LF / HF) is widely used as an indicator of autonomic nervous system balance. An elevated LF / HF ratio typically indicates increased sympathetic activity or decreased parasympathetic activity, corresponding to increased stress, alertness, and cognitive effort in learners during cognitive tasks. By calculating this ratio, the activity state of the learner's autonomic nervous system can be quantified, providing a physiological basis for the recognition of cognitive rhythms.

[0092] For skin conductance signals, this application focuses on extracting the frequency and amplitude of the phase-dependent skin conductance response. The phase-dependent skin conductance response refers to the brief, rapid rise and fall of skin conductance after stimulation; it is an involuntary physiological response closely related to an individual's alertness, attention, and emotional arousal. Its frequency reflects the intensity of the stimulus or the individual's sensitivity to the stimulus, while the amplitude reflects the intensity of physiological arousal. By extracting these features, learners' alertness and emotional arousal levels under specific stimuli can be directly quantified, providing objective indicators for assessing their attentional state and cognitive engagement.

[0093] In eye-tracking data processing, this application extracts the distribution characteristics of pupil diameter fluctuation frequency and fixation duration. Changes in pupil diameter are related to cognitive load and attention allocation; high-frequency fluctuations may indicate high cognitive load or attention switching between different tasks. Fixation duration is a key indicator of attentional concentration; longer fixation duration is generally associated with deep cognitive processing and focused attention, while shorter fixation duration may indicate distraction or rapid information browsing. By analyzing these eye-tracking characteristics, the learner's level of attentional concentration and cognitive load when processing information can be intuitively reflected, providing direct visual behavioral evidence for judging their cognitive rhythm state.

[0094] Finally, the indicators of autonomic nervous system activity, alertness, and attention concentration extracted from different modalities of physiological signals are used as feature vectors and input into a pre-trained cognitive rhythm phase classification model. This model can be built based on machine learning or deep learning algorithms, such as support vector machines or neural networks. It has been thoroughly trained on a large amount of historical learner physiological data and corresponding cognitive rhythm phase labels (such as attention rise, attention plateau, attention decay, and attention trough). Based on the input features, the model predicts the cognitive rhythm phase of the current learner group.

[0095] Through the aforementioned technical solutions, this application clarifies the specific types of multimodal physiological signals used to identify learners' cognitive rhythms, namely heart rate variability signals, skin conductance signals, and eye movement data. For these signals, this application further proposes specific methods for extracting rhythmic feature parameters, including frequency domain analysis of heart rate variability signals to obtain indicators of autonomic nervous system activity, extraction of the frequency and amplitude of phased skin conductance responses from skin conductance signals as indicators of alertness, and extraction of the fluctuation frequency of pupil diameter and the distribution characteristics of fixation duration from eye movement data as indicators of attention concentration. These refined feature parameters, with clear physiological and cognitive significance, can more comprehensively and accurately reflect the real-time cognitive state of learners. Finally, inputting these multimodal features into a pre-trained cognitive rhythm phase classification model can effectively overcome the technical challenges of extracting effective information from complex physiological signals, significantly improve the accuracy and reliability of current group cognitive rhythm phase determination, and provide a solid data foundation and decision-making basis for subsequent personalized teaching intervention strategies.

[0096] In some other embodiments, S102 may further include:

[0097] Multimodal physiological signals are sampled using a sliding time window, and multidimensional rhythmic features at the population level are extracted within each sliding window; the multidimensional rhythmic features include at least: frequency domain features of heart rate variability, time domain features of skin conductance, and spatial features of eye movement;

[0098] The multidimensional rhythm features are input into the corresponding phase probability evaluation sub-models to obtain the phase probability distribution of each mode independently determined.

[0099] The phase probability distributions of each modality, which are determined independently, are weighted and fused according to preset dynamic confidence weights to obtain the cognitive rhythm phase of the current group. The dynamic confidence weights are dynamically adjusted according to the signal-to-noise ratio of each modality's physiological signal within the current time window, with higher fusion weights assigned to modalities with higher signal-to-noise ratios.

[0100] In this embodiment, time-window sliding sampling is performed on multimodal physiological signals to capture the dynamic changes in the cognitive rhythms of the learner group. Specifically, continuously acquired multimodal physiological signals can be divided into a series of time windows with fixed durations (e.g., 5 seconds, 10 seconds, or 30 seconds) and overlap (e.g., 50% overlap). Within each sliding window, multidimensional rhythmic features at the group level are extracted to comprehensively reflect the cognitive state of the entire learner group. These features are not simply averaged from individual data, but can be extracted from individual physiological signals using statistical aggregation methods (such as mean, median, variance, entropy, etc.) or more complex group modeling techniques (such as independent component analysis, principal component analysis, etc.) to characterize the overall physiological response and cognitive trends of the group.

[0101] Multidimensional rhythmic features include at least heart rate variability (HRV) frequency domain features, skin conductance time domain features, and eye-tracking spatial features. HRV frequency domain features reflect important indicators of autonomic nervous system activity and are closely related to cognitive load and attentional state. In addition to the ratio of low-frequency to high-frequency components, the absolute or relative power of ultra-low frequency (ULF), very low frequency (VLF), low frequency (LF), and high frequency (HF), as well as the total power, can be extracted. Skin conductance time domain features primarily reflect sympathetic nervous system activity and are closely linked to an individual's alertness, emotional arousal, and attentional level. In addition to the frequency and amplitude of phase-dependent skin conductance responses, the skin conductance level (SCL), nonspecific skin conductance response (NS-SCR) counts, latency, rise time, and half-recovery time can be extracted. Eye-tracking spatial features directly reflect the learner's visual attentional allocation and cognitive processing. In addition to the fluctuation frequency of pupil diameter and the distribution characteristics of fixation duration, we can also extract fixation point dispersion, saccade amplitude, saccade speed, number of saccades, frequency of fixation area switching, and the proportion of time the fixation point stays in a specific area of ​​interest (AOI).

[0102] Subsequently, the independent information from each modality is used to make a preliminary judgment on the phase of the cognitive rhythm. Each sub-model can be an independent classifier or regression model, such as a support vector machine (SVM), random forest, neural network, or Gaussian mixture model. These models are pre-trained and can output the probability distribution of the learner group being in the attention rise phase, attention plateau phase, attention decline phase, or attention trough phase based on the features of a single modality. Thus, obtaining the phase probability distribution independently determined by each modality means that each physiological modality (such as heart rate variability, skin conductance, and eye movement) will independently provide a probability estimate of the current cognitive rhythm phase.

[0103] Based on this, the phase probability distributions independently determined by each modality are weighted and fused according to preset dynamic confidence weights. The core of this method is to adjust the influence of each modality in the final decision based on its real-time reliability. The dynamic confidence weights are dynamically adjusted based on the signal-to-noise ratio (SNR) of each modality's physiological signal within the current time window, with modalities having higher SNRs assigned higher fusion weights. The SNR can be calculated in various ways; for example, for heart rate variability signals, it can assess the stability of the RR interval sequence or the proportion of artifacts; for skin conductance signals, it can detect baseline drift or motion artifacts; and for eye-tracking data, it can assess the data loss rate or the degree of fixation jitter. When the signal quality of a modality decreases (low SNR), its corresponding weight is reduced, thereby reducing its negative impact on the final cognitive rhythm phase determination; conversely, its weight is increased. This dynamic adjustment mechanism ensures the robustness and accuracy of the fusion results. After weighted fusion, the cognitive rhythm phase of the current group is obtained, which is a comprehensive and more reliable judgment result on the current cognitive state of the learner group (such as the attention rise phase, attention plateau phase, attention decline phase or attention trough phase).

[0104] Through the aforementioned technical solution, this application ensures the capture and comprehensive representation of dynamic changes in the cognitive state of the current group by employing time window sliding sampling and multi-dimensional rhythmic feature extraction at the group level when calculating the cognitive rhythm phase. Furthermore, by introducing a phase probability assessment sub-model to independently determine each modality's physiological signal and dynamically adjusting its fusion weight based on the signal-to-noise ratio of each modality signal within the current time window, adaptive management of the reliability of different modal signals is achieved. This dynamic weighted fusion mechanism significantly improves the accuracy, robustness, and real-time adaptability of the group's cognitive rhythm phase determination, effectively suppresses interference from low-quality signals, and provides a more reliable basis for the subsequent precise application of time anchor interference strategies, thereby optimizing the learning effect of the virtual classroom.

[0105] In some other embodiments, S102 may further include:

[0106] Using the historical phase sequence of N consecutive time windows preceding the current time window as input, the time series prediction model outputs the predicted probability distribution of each phase within the future time window and the predicted value of the phase transition time.

[0107] The predicted values ​​for the phase transition time include at least: the expected duration of the current phase, the expected arrival time of the next phase, and the transition probability from the current phase to the next phase;

[0108] The time series prediction model also outputs the prediction confidence interval of the predicted value. The prediction confidence interval is dynamically calculated based on the goodness of fit of the current group cognitive rhythm model and the historical prediction error.

[0109] Based on the prediction confidence interval, the evolution trend of cognitive phase is determined.

[0110] In this embodiment, simple phase prediction alone may not be sufficient to capture the dynamic details of cognitive rhythm changes, such as the expected duration of each phase, the expected arrival time of the next phase, and the transition probability from the current phase to the next phase. It would also be difficult to assess the reliability of the prediction results, potentially leading to inaccurate or insufficiently adaptive time-anchor interference strategies, thus affecting the optimization of learning outcomes. Therefore, when predicting the evolution trend of cognitive rhythm phases within a future time window, the system first uses the historical phase sequence of N consecutive time windows preceding the current time window as input. This historical phase sequence contains the trajectory of cognitive rhythm phase changes within the learner group over a past period, providing crucial contextual information for predicting future trends.

[0111] Subsequently, the system processes the input historical phase sequence using a time series prediction model to output the predicted probability distribution of each phase within the future time window, as well as the predicted values ​​of the phase transition times. The time series prediction model can be a variety of mature machine learning or deep learning models, such as Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Autoregressive Moving Average (ARIMA), or Hidden Markov Model (HMM). These models can learn complex patterns and dependencies in the time series, thereby predicting future phase states. The output predicted probability distribution represents the likelihood of being in various cognitive rhythm phases at a future point in time, rather than a single deterministic outcome, which better reflects the inherent uncertainty of cognitive states.

[0112] The predicted phase transition time is a crucial component of the prediction results, including at least the expected duration of the current phase, the expected arrival time of the next phase, and the probability of transitioning from the current phase to the next phase. The expected duration of the current phase indicates how long the learner group's current cognitive rhythm phase is expected to last, which is essential for assessing the stability of the current state and when intervention is necessary. The expected arrival time of the next phase provides the specific time point of future cognitive state transitions, allowing the system to plan and prepare appropriate intervention measures in advance. The probability of transitioning from the current phase to the next phase quantifies the likelihood of this transition occurring, providing decision-makers with a basis for risk assessment.

[0113] In addition to outputting predicted values, time series forecasting models also output prediction confidence intervals. The prediction confidence interval represents the range of reliability of the predicted result, i.e., how likely the actual value is to fall within this range. This prediction confidence interval is dynamically calculated based on the goodness of fit of the current group's cognitive rhythm model and historical prediction errors. The model's goodness of fit reflects its ability to interpret historical data; a higher goodness of fit generally means a better understanding of the current group's cognitive rhythm patterns, and the higher the confidence level of the prediction is likely to be. Historical prediction errors record the model's performance in past prediction tasks; a smaller error indicates a more accurate model, and the confidence level of its future predictions will correspondingly increase. Dynamic calculation means that these parameters are updated in real time as new data flows in and model performance changes, ensuring the accuracy and timeliness of the confidence interval.

[0114] Finally, the system determines the cognitive phase evolution trend based on the predicted confidence interval. This means that when judging the future cognitive phase evolution trend, not only the predicted phase and transition time are considered, but also the reliability of the prediction. For example, predictions with a high confidence interval can be considered to have a clearer and more reliable trend, thus allowing for more proactive or precise intervention strategies; while predictions with a low confidence interval may require more conservative, adaptive, or further verification strategies.

[0115] In this embodiment, by using historical phase sequences across N consecutive time windows as input and employing a time series prediction model, the dynamic evolution of collective cognitive rhythms can be captured more accurately. The model outputs the predicted probability distribution of each phase within the future time window, rather than a single deterministic prediction, thus better reflecting the inherent uncertainty of cognitive states. Simultaneously, it clearly predicts the expected duration of the current phase, the expected arrival time of the next phase, and the phase transition probability, providing refined timing points and decision-making basis for subsequent time-anchor interference strategies. More importantly, by outputting the prediction confidence interval of the predicted values ​​and dynamically calculating based on the goodness of fit of the collective cognitive rhythm model and historical prediction errors, the system can quantify the reliability of the prediction results. This allows for more flexible and robust decision-making when determining time-anchor interference strategies, depending on the confidence level of the prediction. For example, more aggressive interventions can be implemented when confidence is high, while more conservative or adaptive strategies can be adopted when confidence is low. This significantly improves the accuracy, effectiveness, and robustness of the time-anchor interference strategy, ultimately optimizing learning outcomes in virtual classrooms.

[0116] In some embodiments, S103 may include:

[0117] Obtain the individual comprehension level index of each learner in the group at the current knowledge point;

[0118] The statistical characteristics of the group's understanding state are calculated based on the individual understanding index. The statistical characteristics of the group's understanding state include at least one of the following: the group's understanding mean, the group's understanding variance, the group's understanding entropy value, and the individual deviation value between the individual understanding index and the group's understanding mean. The group's understanding entropy value is used to quantify the consistency of the group's solution state in the current knowledge point.

[0119] Based on the statistical characteristics of the understanding state at the group level, the weakness type of the current knowledge point is determined; the weakness type includes at least one of the following: common weak knowledge points when the average group understanding level is lower than the first threshold and the group understanding entropy value is lower than the second threshold, differentiated weak knowledge points when the average group understanding level is lower than the first threshold and the group understanding entropy value is higher than the third threshold, and individual weakness markers corresponding to individual learners whose individual deviation value exceeds the preset individual deviation threshold.

[0120] For common weak knowledge points, a unified time anchor interference strategy is initiated for the group; for differentiated weak knowledge points, individual differentiated parameters are superimposed on the unified time anchor interference for the group; for individual weak markers, individualized interference parameters are generated for the corresponding individual learners.

[0121] Specifically, in a virtual reality environment, the system can record the learner's accuracy, reaction time, and number of incorrect attempts when completing interactive tasks related to the knowledge point, or assess them through embedded quizzes and question-and-answer sessions. Furthermore, the cognitive load and attention level reflected by the learner's physiological signals (such as eye-tracking data and electroencephalograms) while learning the knowledge point can also be used as a component of the individual's comprehension index.

[0122] Building upon this, statistical characteristics of group-level comprehension are calculated based on individual comprehension indices. These statistical characteristics provide a macro-level view of the group's overall understanding of the knowledge points. Specifically, the group comprehension mean reflects the average mastery level of the group; the group comprehension variance measures the dispersion of comprehension levels within the group, i.e., the magnitude of fluctuation in the individual comprehension index relative to the mean; the group comprehension entropy quantifies the consistency of the group's solution state for the current knowledge point. Its calculation method involves discretizing the individual comprehension index into several levels, and then calculating the information entropy based on the probability distribution of these levels within the group. A higher entropy value indicates greater differences in comprehension within the group. Individual deviation values ​​are used to identify individual learners whose comprehension levels significantly deviate from the group average.

[0123] Furthermore, based on the statistical characteristics of the group's understanding status, the type of weakness in the current knowledge point is determined. This classification mechanism is key to achieving precise intervention. When the average group understanding level is below a preset first threshold and the group understanding entropy is below a preset second threshold, it indicates that most learners do not fully understand the knowledge point, and this knowledge point is identified as a common weak knowledge point. When the average group understanding level is below the preset first threshold but the group understanding entropy is above the preset third threshold, it indicates that there is a significant differentiation in the understanding of the knowledge point within the group; some learners may have mastered it, while others have difficulty understanding it, and this knowledge point is identified as a differentiated weak knowledge point. In addition, if a learner's individual deviation value exceeds a preset individual deviation threshold, the learner is marked as an individual deviation learner, and the corresponding knowledge point is assigned an individual weakness label. These thresholds (first threshold, second threshold, third threshold, and individual deviation threshold) can be dynamically adjusted based on teaching experience, the difficulty of the knowledge point, or by training the model using historical data.

[0124] Finally, based on the identified weaknesses, corresponding intervention strategies are initiated. For common weak knowledge points, given the widespread comprehension difficulties within the group, the system will activate a unified time-anchor interference strategy for the group. For example, this could involve uniformly extending the subjective dwell time on the knowledge point for all learners, or reinforcing the knowledge point through standardized VR multi-sensory manipulation. For differentiated weak knowledge points, individual-specific parameters are superimposed on the unified time-anchor interference strategy for the group. For instance, the intensity and duration of VR multi-sensory manipulation are fine-tuned based on differences in individual comprehension indices to balance group and individual needs. For individual weakness markers, the system will generate highly individualized interference parameters for the corresponding individual learner, implementing a customized VR multi-sensory manipulation plan to ensure the accuracy and effectiveness of the intervention.

[0125] Through the aforementioned technical solution, this application can meticulously obtain the individual comprehension index of each learner in the group regarding the current knowledge point, and calculate the statistical characteristics of the group-level comprehension status based on this, thereby accurately determining the type of weakness in the current knowledge point. This classification and identification mechanism allows subsequent time-anchor interference strategies to be finely matched according to the specific type of weakness (common weakness, differentiated weakness, or individual weakness). For common weak knowledge points, a unified group interference strategy can be activated to ensure overall learning efficiency; for differentiated weak knowledge points, individual difference parameters are superimposed on the unified group strategy to take into account both group and individual needs; and for individual weakness markers, highly individualized interference parameters can be generated for the corresponding learner. This layered and refined intervention strategy effectively solves the limitations of the "one-size-fits-all" intervention in traditional methods, significantly improving the targeting, effectiveness, and resource utilization efficiency of the time-anchor interference strategy, thereby more accurately regulating learners' subjective time perception and optimizing learning outcomes.

[0126] In some other embodiments, S104 may include:

[0127] Construct a multidimensional cognitive state space with the features of cognitive rhythm phase and weak knowledge points as coordinate axes; the multidimensional cognitive state space includes at least the cognitive rhythm phase dimension and the knowledge point cognitive complexity dimension;

[0128] The prediction results of the cognitive rhythm phase of the current learner group and the future phase are mapped to the multidimensional cognitive state space to determine the position coordinates and movement trajectory of the current group in the cognitive state space.

[0129] Based on location coordinates and motion trajectory, the corresponding optimal time anchor interference strategy is queried from the pre-set cognitive state space decision graph; the cognitive state space decision graph is a decision graph obtained in advance through reinforcement learning training, which represents the mapping relationship between each region of the cognitive state space and the optimal interference strategy.

[0130] Specifically, to more accurately simulate virtual reality classrooms, a multi-dimensional cognitive state space can be constructed, using the characteristics of cognitive rhythm phases and weak knowledge points as coordinate axes. This multi-dimensional cognitive state space includes at least a cognitive rhythm phase dimension and a knowledge point cognitive complexity dimension. The cognitive rhythm phase dimension can be divided according to discrete states such as the learner's attention rise, attention plateau, attention decay, and attention trough, or represented by continuous phase values; for example, mapping a complete cognitive rhythm cycle to the interval between 0 and 2π. The knowledge point cognitive complexity dimension is used to quantify the difficulty of the weak knowledge point itself, the amount of prior knowledge required, the degree of conceptual abstraction, and its relevance to other knowledge points. This complexity can be determined through expert evaluation, historical learning data analysis (such as average mastery time and error rate), or computational methods based on knowledge graphs. For example, a simple concept memory point may have low cognitive complexity, while a knowledge point requiring multi-step reasoning and problem-solving has high cognitive complexity. By constructing such a multi-dimensional space, key factors affecting learning effectiveness can be integrated into a unified mathematical model, providing a foundation for subsequent strategic decisions.

[0131] The system then updates and dynamically tracks the cognitive state of the learner group in real time. The current cognitive rhythm phase of the learner group, such as being in an attention plateau, is mapped to a corresponding position on the cognitive rhythm phase dimension. Simultaneously, the evolution trend of the cognitive rhythm phase within a future time window, such as a prediction of a gradual shift from an attention plateau to an attention decay phase within the next five minutes, forms a trajectory in this multi-dimensional cognitive state space. This trajectory not only reflects the current cognitive state but, more importantly, reveals the dynamic trend of cognitive state change. The characteristics of weak knowledge points, such as their high cognitive complexity, are also mapped to specific coordinates on the knowledge point cognitive complexity dimension. Through this mapping, the system obtains a precise positional coordinate representing the overall cognitive state of the learner group when facing a specific weak knowledge point, as well as a trajectory indicating the short-term evolution direction of this state.

[0132] In this embodiment, the cognitive state space decision graph is a decision graph pre-trained through reinforcement learning that represents the mapping relationship between each region of the cognitive state space and the optimal interference strategy. This decision graph can be a multidimensional lookup table, a decision tree model, or a deep neural network model. During training, the reinforcement learning agent interacts with the learner group in a virtual classroom simulation environment, continuously trying different time-anchor interference strategies and optimizing its strategy based on the learning effect gain value (reward signal). Ultimately, the trained decision graph can provide one or a set of optimal time-anchor interference strategy parameters for each position coordinate and motion trajectory in the cognitive state space. Once the system identifies the current group's cognitive state (position coordinates) and its evolution trend (motion trajectory), it can directly query the decision graph to obtain the most suitable VR multi-sensory control methods (such as visual refresh rate gradient modulation, spatial reference frame displacement, etc.) and their specific parameter settings (such as intensity, duration, and start time).

[0133] In this embodiment, firstly, by constructing a multi-dimensional cognitive state space, complex factors such as the cognitive rhythm phase of the learner group, the characteristics of weak knowledge points, and future phase evolution trends are integrated into a unified framework, resulting in a more comprehensive and in-depth understanding of the learner's state. Secondly, mapping real-time and predicted data onto this space allows for the dynamic capture of the learner group's current cognitive state and its future evolution trends, providing precise contextual information for decision-making. Most importantly, utilizing the cognitive state space decision map obtained in advance through reinforcement learning training, the system can intelligently query and obtain the optimal time anchor interference strategy based on the current position coordinates and movement trajectory. This reinforcement learning-based decision-making mechanism enables the determined interference strategy to fully consider the dynamic interaction of various complex factors, thereby achieving precise control over the learner's subjective time perception, effectively stretching the subjective dwell time of weak knowledge points, or reasonably compressing the subjective dwell time of mastered knowledge points. This not only improves the targeting and effectiveness of time anchor interference but also significantly optimizes learning outcomes, avoiding blind or inefficient interventions, thereby improving the teaching quality and efficiency of virtual reality classroom simulation.

[0134] Furthermore, the construction of this decision graph requires accurately capturing the complex mapping relationship between cognitive states and optimal interference strategies. This presents a challenge in practical applications: how to efficiently and accurately acquire and optimize these mapping relationships, especially in dynamically changing virtual classroom environments where traditional manual rule setting or exhaustive testing is insufficient. A multi-dimensional cognitive state space is constructed, using the features of cognitive rhythm phases and weak knowledge points as coordinate axes, including:

[0135] Construct a virtual classroom simulation environment, and conduct multiple simulated teaching sessions in the simulation environment with a simulated group of learners;

[0136] In each simulated teaching session, the position coordinates in the cognitive state space are used as the state input, the parameters of each dimension of the time anchor interference strategy are used as the action output, and the learning effect gain value after the simulated teaching is used as the reward signal.

[0137] A deep reinforcement learning algorithm is used to train the policy network. The policy network takes the spatial coordinates of the cognitive state as input and outputs the optimal parameter combination of each dimension of the time anchor interference policy.

[0138] After training, the input-output mapping relationship of the policy network is solidified into a multidimensional cognitive state space.

[0139] In this embodiment, constructing a virtual classroom simulation environment refers to creating a digital, interactive learning space that can simulate the physical environment of a real classroom, the presentation of teaching content, and learner behavioral feedback. This environment can precisely control various teaching parameters and learner perceptual inputs, such as the difficulty of knowledge points, the pace of instruction, and parameters of VR multi-sensory control methods. The simulated learner group refers to a collection of virtual learners with different cognitive characteristics, learning habits, and physiological response patterns simulated through a computational model. By executing multiple simulated teaching sessions in the simulation environment, a large amount of teaching interaction data and learning effect data can be efficiently generated, providing rich experience samples for subsequent reinforcement learning training without relying on large-scale real-user experiments, thereby reducing experimental costs and time.

[0140] In each simulated teaching session, the position coordinates in the cognitive state space serve as the state input. This refers to the representation of the agent's (i.e., the policy network) perception of the current state of the environment within the reinforcement learning framework. Specifically, this refers to the position coordinates in the multi-dimensional cognitive state space, which comprehensively reflect the cognitive rhythm phase and characteristics of weak knowledge points within the learner group. The parameters of the time-anchored interference strategy, used as the action output, represent the decisions made by the agent based on the current state. These parameters include specific parameters of the time-anchored interference strategy, such as the amplitude of the visual refresh rate gradient modulation, the displacement of the spatial reference frame, and the duration of multi-sensory delay modulation. The learning effect gain value after simulated teaching serves as the reward signal. This refers to the feedback given by the environment after the agent performs an action, used to evaluate the quality of that action. Here, it is the learning effect gain value after simulated teaching, which quantifies the improvement in the learner's understanding of weak knowledge points or the effect of forgetting suppression after implementing a specific interference strategy.

[0141] Training a policy network using deep reinforcement learning algorithms combines the perceptual capabilities of deep learning with the decision-making abilities of reinforcement learning, enabling it to handle high-dimensional state and action spaces. The policy network is a deep neural network whose input layer receives position coordinates from the cognitive state space. Through multiple nonlinear transformations, the output layer generates parameter combinations for each dimension of the time-anchored perturbation policy. During training, deep reinforcement learning algorithms (e.g., Deep Q-Network (DQN), Asynchronous Advantage Actor-Critic (A3C), Proximal Policy Optimization (PPO)) continuously interact with the simulated environment, adjusting the policy network's weights and biases based on the obtained reward signals. This allows the network to gradually learn the optimal combination of time-anchored perturbation policy parameters that maximizes learning performance gains under different cognitive states.

[0142] After training, the input-output mapping relationship of the policy network is solidified into a multidimensional cognitive state space. This means saving the parameters and structure of the trained and optimized policy network, making it a stable and directly queryable decision model. This implies that once the policy network is trained, it can quickly and accurately output the parameter combination of the corresponding optimal time-anchor interference policy based on any given cognitive state space coordinates. This solidified input-output mapping relationship essentially constitutes a multidimensional cognitive state space decision graph, which represents the correspondence between each region of the cognitive state space and the optimal interference policy, thereby achieving automation and intelligence in complex decision-making logic.

[0143] Through the aforementioned technical solution, this application utilizes a virtual classroom simulation environment to generate a large amount of training data and combines it with a deep reinforcement learning algorithm, enabling the policy network to autonomously learn and discover the optimal time-anchor interference strategy. This method avoids the limitations of manual experience rules and can adapt to a wider range of cognitive states and weak knowledge point characteristics, thereby generating more refined and effective interference strategies. Specifically, by using the spatial coordinates of the cognitive state as the state input, the time-anchor interference strategy parameters as the action output, and the learning effect gain value as the reward signal, the deep reinforcement learning algorithm can iteratively optimize the policy network, enabling it to output the VR multi-sensory control parameter combination that best improves learning effectiveness when facing different learner groups' cognitive rhythm phases and weak knowledge point characteristics. Ultimately, the solidified policy network forms an intelligent decision graph, greatly improving the accuracy and adaptability of the time-anchor interference strategy determination, thereby effectively regulating learners' subjective time perception, optimizing the learning effect of weak knowledge points, and promoting the consolidation of mastered knowledge points, thus comprehensively improving the teaching efficiency and personalization level of the virtual classroom simulation method.

[0144] In some embodiments, S106 may include:

[0145] At the first preset time point after the implementation of the time anchor interference strategy, the learners' first understanding index of the weak knowledge points is collected as an immediate assessment result.

[0146] At the second preset time point after the implementation of the time anchor interference strategy, the second comprehension index of the learners' group on the weak knowledge points is collected as the delayed assessment result; the second preset time point is later than the first preset time point.

[0147] Based on both immediate and delayed evaluation results, the overall learning performance evaluation value of the time anchor interference strategy is calculated.

[0148] In this embodiment, the first and second preset time points are key time nodes for evaluating learning effectiveness. The first preset time point aims to capture the learner's immediate comprehension state after the implementation of the time-anchored interference strategy, reflecting their short-term memory and immediate learning effectiveness. This time point can be set immediately after the interference strategy ends, or within minutes after the strategy ends, to ensure timely reflection of the learner's initial grasp of the knowledge points. The second preset time point is used to capture the learner's comprehension state after a period of time, reflecting their long-term memory and forgetting suppression effects. This time point is later than the first preset time point, for example, it can be set several hours, a day, or several days after the interference strategy ends, depending on the complexity of the weak knowledge points and the expected memory retention period. The selection of these two time points should be optimized based on educational psychology principles and experimental data to ensure effective differentiation between immediate and long-term learning effects.

[0149] The first and second comprehension indices are indicators that quantify learners' understanding of specific weak knowledge points, used to objectively measure learners' learning outcomes at different points in time. These indices can be obtained in various ways, such as through in-class quizzes, Q&A sessions, concept mapping, completion of problem-solving tasks targeting the weak knowledge point, or calculations based on learners' interactive behaviors with knowledge-related content in virtual environments (such as repeated viewing, active exploration, error rates, etc.). These indices can be percentage scores, rating scales, or outputs from more complex cognitive load models. The first comprehension index is used as the immediate assessment result, and the second comprehension index is used as the delayed assessment result.

[0150] After obtaining both immediate and delayed evaluation results, a comprehensive evaluation value for the learning effectiveness of the time-anchored interference strategy needs to be calculated. This evaluation value is a comprehensive indicator used to quantify the overall effectiveness of the time-anchored interference strategy, providing a single, comprehensive measure to compare the effects of different strategies and guide parameter updates in the group cognitive rhythm model. The comprehensive calculation can employ various algorithms. For example, a weighted average of the immediate and delayed evaluation results can be used, with the delayed evaluation results having a higher weight depending on actual needs to emphasize the importance of long-term memory. Alternatively, it can be calculated based on the forgetting curve theory, by comparing the actual forgetting rate with the expected forgetting rate. More complex calculation methods can consider factors such as the learner's initial comprehension level and the difficulty of the knowledge points to provide a more accurate evaluation.

[0151] By collecting learners' comprehension indices of weak knowledge points at both the first and second preset time points after implementing the time-anchored interference strategy, and using these indices as immediate and delayed assessment results, the learning effectiveness can be comprehensively captured in both short-term and long-term dimensions. A comprehensive calculation based on these two different time-point assessment results can more accurately quantify the actual impact of the time-anchored interference strategy on learners' knowledge mastery and memory retention ability, effectively avoiding the bias that may arise from a single immediate assessment. This multi-time-point, comprehensive assessment mechanism provides more reliable and instructive feedback for subsequent parameter updates to the group's cognitive rhythm model, thereby continuously optimizing the overall effectiveness of the virtual classroom simulation method and ensuring that the implemented interference strategy truly promotes learners' long-term consolidation of weak knowledge points.

[0152] In other embodiments, the comprehensive learning performance evaluation value of the time anchor interference strategy is calculated, including:

[0153] Obtain the preset forgetting curve parameters corresponding to the weak knowledge points. The forgetting curve parameters are pre-calibrated based on the knowledge point type and cognitive complexity level of the weak knowledge points.

[0154] Based on the immediate assessment results, the delayed assessment results, and the preset forgetting curve parameters, the actual forgetting rate of weak knowledge points is calculated.

[0155] The actual forgetting rate is compared with the standard forgetting rate in the preset forgetting curve parameters to obtain the forgetting inhibition effect coefficient;

[0156] Based on the immediate assessment results and the forgetting inhibition effect coefficient, a comprehensive learning effect evaluation value is calculated.

[0157] In this embodiment, when evaluating the learning effect after implementing the time-anchor interference strategy, it is first necessary to obtain the preset forgetting curve parameters corresponding to the weak knowledge points. These parameters are pre-calibrated according to the knowledge point type (e.g., factual knowledge, conceptual knowledge, procedural knowledge, etc.) and cognitive complexity level (e.g., memory, understanding, application, analysis, evaluation, creation, etc.) of the weak knowledge point. The preset forgetting curve parameters can be obtained through large-scale learning behavior data analysis, educational psychology experiments, or expert experience, and are used to describe the natural forgetting pattern of knowledge points of this type and complexity level over time without specific intervention. For example, a simple conceptual knowledge point may have a faster initial forgetting rate, while a complex skill knowledge point may have a slower forgetting rate but requires more practice to consolidate.

[0158] Based on this, by combining the immediate assessment results (e.g., the learner's immediate mastery score of the knowledge point) collected at the first preset time point after the implementation of the time-anchor interference strategy and the delayed assessment results (e.g., the learner's retention score after a period of time) collected at the second preset time point, along with the aforementioned preset forgetting curve parameters, the actual forgetting rate of the weak knowledge point is calculated. The actual forgetting rate reflects the actual rate at which the learner forgets the knowledge point under the influence of the time-anchor interference strategy. This can be quantified by comparing the decline rates of the immediate and delayed assessment results and referring to the natural forgetting trend indicated by the preset forgetting curve parameters. For example, a model can be constructed with the immediate assessment results as the initial state and the delayed assessment results as the final state, and the preset forgetting curve parameters can be used to infer the actual forgetting process parameters.

[0159] Subsequently, the calculated actual forgetting rate is compared with the corresponding standard forgetting rate in the preset forgetting curve parameters to obtain the forgetting inhibition effect coefficient. The standard forgetting rate is the expected rate of forgetting under no-intervention or conventional teaching conditions, based on the type of knowledge point and the level of cognitive complexity. The forgetting inhibition effect coefficient is a quantitative indicator that shows the extent to which the time-anchored interference strategy slows down or inhibits the learner's forgetting process. For example, if the actual forgetting rate is much lower than the standard forgetting rate, the forgetting inhibition effect coefficient will be higher, indicating that the strategy is effective. This coefficient can be a ratio (such as the ratio of the standard forgetting rate to the actual forgetting rate) or a difference (such as the standard forgetting rate minus the actual forgetting rate). After normalization, it can intuitively reflect the intervention effect of the strategy on forgetting.

[0160] Finally, based on the immediate assessment results and the obtained forgetting suppression effect coefficient, a comprehensive learning effectiveness evaluation value is calculated. This comprehensive evaluation value not only considers the learner's immediate mastery of the knowledge points, but more importantly, it incorporates the contribution of the time-anchored interference strategy to long-term memory retention. For example, the comprehensive learning effectiveness evaluation value can be designed as a weighted combination of the immediate assessment results and the forgetting suppression effect coefficient, or it can be fused together by a function, so that the evaluation results can comprehensively reflect the learner's immediate performance and long-term memory retention ability after intervention.

[0161] By introducing preset forgetting curve parameters corresponding to weak knowledge points and combining immediate and delayed assessment results, the actual forgetting rate of that knowledge point can be accurately calculated. Comparing this actual forgetting rate with the preset standard forgetting rate quantifies the degree to which the time-anchored interference strategy inhibits the forgetting process, yielding a forgetting inhibition effect coefficient. Finally, the comprehensive learning effect assessment value calculated based on the immediate assessment results and the forgetting inhibition effect coefficient not only reflects the learner's immediate mastery of the knowledge point, but more importantly, it incorporates the strategy's contribution to long-term memory retention, making the assessment results more comprehensive and instructive. This assessment method effectively overcomes the limitations of relying solely on immediate and delayed assessments, providing more accurate and insightful feedback for updating the parameters of the group cognitive rhythm model, thereby continuously optimizing learning intervention strategies in virtual classroom simulations and improving the persistence of learning outcomes.

[0162] Based on the virtual reality classroom simulation method provided in the above embodiments, this application also provides specific implementation methods for a virtual reality classroom simulation device. Please refer to the following embodiments.

[0163] First see Figure 2 The virtual reality classroom simulation device 200 provided in this application embodiment may include:

[0164] Acquisition module 201 is used to acquire multimodal physiological signals of learners in a virtual classroom.

[0165] Module 202 is established to extract rhythmic feature parameters based on multimodal physiological signals, establish a group cognitive rhythm model, calculate the current cognitive rhythm phase of the group, and predict the evolution trend of the cognitive rhythm phase within a future time window; the cognitive rhythm phase includes at least the attention rise phase, attention plateau phase, attention decline phase, and attention trough phase;

[0166] The identification module 203 is used to monitor the learners' understanding of the current knowledge points and identify weak knowledge points;

[0167] Module 204 is used to determine the time anchor interference strategy based on the characteristics of weak knowledge points, the cognitive rhythm phase of the current learner group, and the phase evolution trend of future phases.

[0168] The control module 205 is used to control the learner's subjective time perception through VR multi-sensory control methods when the time anchor interference strategy is determined to be implemented, so that the subjective dwell time of weak knowledge points is stretched and / or the subjective dwell time of mastered knowledge points is compressed; the VR multi-sensory control methods include one or more of visual refresh rate gradient modulation, spatial reference system displacement, multi-sensory delay modulation and dynamic time reference system embedding.

[0169] Evaluation module 206 is used to evaluate the learning effect after the implementation of the time anchor interference strategy, obtain evaluation results, and update the parameters of the group cognitive rhythm model based on the evaluation results.

[0170] Figure 3 A schematic diagram of the hardware structure of the electronic device provided in an embodiment of this application is shown.

[0171] An electronic device may include a processor 301 and a memory 302 storing computer program instructions.

[0172] Specifically, the processor 301 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0173] Memory 302 may include mass storage for data or instructions. For example, and not limitingly, memory 302 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. In one instance, memory 302 may include removable or non-removable (or fixed) media, or memory 302 may be a non-volatile solid-state memory. Memory 302 may be internal or external to an electronic device.

[0174] In one instance, memory 302 may be read-only memory (ROM). In one instance, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0175] Memory 302 may include read-only memory (ROM), random access memory (RAM), disk storage media device, optical storage media device, flash memory device, electrical, optical, or other physical / tangible memory storage device. Therefore, generally, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the virtual reality classroom simulation method according to the first aspect of this disclosure.

[0176] The processor 301 reads and executes computer program instructions stored in the memory 302 to achieve... Figure 1 The illustrated embodiment presents a virtual reality classroom simulation method.

[0177] In one example, the electronic device may also include a communication interface 303 and a bus 304. For example, Figure 3 As shown, the processor 301, memory 302, and communication interface 303 are connected through bus 304 and complete communication with each other.

[0178] The communication interface 303 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0179] Bus 304 includes hardware, software, or both, that couples components of an electronic device together. For example, and not as a limitation, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 304 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, this application contemplates any suitable bus or interconnect.

[0180] The electronic device can execute the virtual reality classroom simulation method in the embodiments of this application, thereby achieving a combination of Figures 1-2 The virtual reality classroom simulation method and apparatus are described.

[0181] Furthermore, in conjunction with the virtual reality classroom simulation methods in the above embodiments, this application embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the virtual reality classroom simulation methods in the above embodiments.

[0182] In an optional embodiment, in conjunction with the virtual reality classroom simulation method in the above embodiments, this application embodiment can provide a computer program product to implement it. The instructions in the computer program product are executed by the processor of the electronic device, enabling the electronic device to implement any of the virtual reality classroom simulation methods in the above embodiments.

[0183] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0184] The functional blocks shown in the above block diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0185] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0186] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0187] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A virtual reality classroom simulation method, characterized in that, include: Acquire multimodal physiological signals of learners in a virtual classroom; Based on the multimodal physiological signals, rhythmic feature parameters are extracted, a group cognitive rhythm model is established, the current group's cognitive rhythm phase is calculated, and the evolution trend of the cognitive rhythm phase within future time windows is predicted; the cognitive rhythm phase includes at least the attention rise phase, attention plateau phase, attention decline phase, and attention trough phase; Monitor learners' understanding of current knowledge points and identify weak knowledge points; Based on the characteristics of the weak knowledge points, the cognitive rhythm phase of the current learner group, and the phase evolution trend of the future phase, a time anchor interference strategy is determined. When the time anchor interference strategy is determined to be implemented, the learner's subjective time perception is adjusted by VR multi-sensory control methods, so that the subjective dwell time of the weak knowledge points is stretched and / or the subjective dwell time of the mastered knowledge points is compressed; the VR multi-sensory control methods include one or more of visual refresh rate gradient modulation, spatial reference system displacement, multi-sensory delay modulation and dynamic time reference system embedding. The learning effect after the implementation of the time anchor interference strategy is evaluated, the evaluation results are obtained, and the parameters of the group cognitive rhythm model are updated based on the evaluation results.

2. The method according to claim 1, characterized in that, The multimodal physiological signals include heart rate variability signals, skin conductance signals, and eye movement data. The extraction of rhythmic feature parameters based on the multimodal physiological signals includes: Frequency domain analysis was performed on the heart rate variability signal to extract the ratio of low-frequency components to high-frequency components as an indicator of the activity state of the autonomic nervous system. The frequency and amplitude of phase-dependent skin conductance responses were extracted from skin conductance signals and used as an indicator of alertness. The fluctuation frequency of pupil diameter and the distribution characteristics of fixation duration were extracted from eye-tracking data and used as indicators of attention concentration. The indicators of the autonomic nervous system activity state, the alertness index, and the attention concentration index are input into a pre-trained cognitive rhythm phase classification model to obtain the cognitive rhythm phase of the current group.

3. The method according to claim 1, characterized in that, The calculation of the cognitive rhythm phase of the current group includes: The multimodal physiological signals are subjected to time window sliding sampling, and multidimensional rhythmic features at the population level are extracted within each sliding window; the multidimensional rhythmic features include at least: heart rate variability frequency domain features, skin conductance time domain features, and eye movement spatial features; The multidimensional rhythm features are input into the corresponding phase probability evaluation sub-models to obtain the phase probability distribution of each mode independently determined; The phase probability distributions of each modality, determined independently, are weighted and fused according to a preset dynamic confidence weight to obtain the cognitive rhythm phase of the current group. The dynamic confidence weight is dynamically adjusted according to the signal-to-noise ratio of each modality's physiological signal within the current time window, with higher fusion weights assigned to modalities with higher signal-to-noise ratios.

4. The method according to claim 1, characterized in that, The predicted trend of cognitive rhythm phase evolution within a future time window includes: Using the historical phase sequence of N consecutive time windows preceding the current time window as input, the time series prediction model outputs the predicted probability distribution of each phase within the future time window and the predicted value of the phase transition time. The predicted value of the phase transition time includes at least: the expected duration of the current phase, the expected arrival time of the next phase, and the transition probability from the current phase to the next phase; The time series prediction model also outputs the prediction confidence interval of the predicted value, which is dynamically calculated based on the goodness of fit of the current group cognitive rhythm model and the historical prediction error. Based on the predicted confidence interval, the cognitive phase evolution trend is determined.

5. The method according to claim 1, characterized in that, The monitoring of learners' understanding of current knowledge points and identification of weak knowledge points includes: Obtain the individual comprehension level index of each learner in the group at the current knowledge point; The statistical characteristics of the group-level understanding state are calculated based on the individual understanding index; the statistical characteristics of the group-level understanding state include at least one of the group understanding mean, group understanding variance, group understanding entropy value, and individual deviation value between the individual understanding index and the group understanding mean; the group understanding entropy value is used to quantify the consistency of the group's solution state in the current knowledge point. Based on the statistical characteristics of the understanding state at the group level, the weakness type of the current knowledge point is determined; the weakness type includes at least one of the following: common weak knowledge points when the average group understanding level is lower than the first threshold and the group understanding entropy value is lower than the second threshold, differentiated weak knowledge points when the average group understanding level is lower than the first threshold and the group understanding entropy value is higher than the third threshold, and individual weakness markers corresponding to individual learners whose individual deviation values ​​exceed the preset individual deviation threshold. For the common weak knowledge points, a unified time anchor interference strategy is initiated for the group; for the differentiated weak knowledge points, individual differential parameters are superimposed on the unified time anchor interference for the group; for the individual weak markers, individualized interference parameters are generated for the corresponding individual learners.

6. The method according to claim 1, characterized in that, The time anchor interference strategy, determined based on the characteristics of the weak knowledge points, the cognitive rhythm phase of the current learner group, and the phase evolution trend of future phases, includes: A multidimensional cognitive state space is constructed with the features of cognitive rhythm phase and weak knowledge points as coordinate axes; the multidimensional cognitive state space includes at least the cognitive rhythm phase dimension and the knowledge point cognitive complexity dimension; The prediction results of the cognitive rhythm phase and future phase of the current learner group are mapped to the multidimensional cognitive state space to determine the position coordinates and movement trajectory of the current group in the cognitive state space. Based on the location coordinates and motion trajectory, the corresponding optimal time anchor interference strategy is queried from the preset cognitive state space decision graph; the cognitive state space decision graph is a decision graph that is obtained in advance through reinforcement learning training and represents the mapping relationship between each region of the cognitive state space and the optimal interference strategy.

7. The method according to claim 6, characterized in that, The construction of a multi-dimensional cognitive state space with the features of cognitive rhythm phases and weak knowledge points as coordinate axes includes: Construct a virtual classroom simulation environment, in which multiple simulated teaching sessions are conducted with a simulated group of learners; In each simulated teaching session, the position coordinates in the cognitive state space are used as the state input, the parameters of each dimension of the time anchor interference strategy are used as the action output, and the learning effect gain value after the simulated teaching is used as the reward signal. A deep reinforcement learning algorithm is used to train the policy network, which takes the spatial coordinates of the cognitive state as input and outputs the optimal parameter combination of each dimension of the time anchor interference policy. After training is completed, the input-output mapping relationship of the policy network is solidified into the multidimensional cognitive state space.

8. The method according to claim 1, characterized in that, The evaluation of the learning effect after implementing the time anchor interference strategy yields evaluation results, including: At the first preset time point after the implementation of the time anchor interference strategy, the learner group’s first comprehension index of the weak knowledge points is collected as an immediate evaluation result. At a second preset time point after the implementation of the time anchor interference strategy, the learner group's second level of understanding of the weak knowledge points is collected as the delayed evaluation result; the second preset time point is later than the first preset time point. Based on the immediate evaluation results and the delayed evaluation results, the comprehensive learning effect evaluation value of the time anchor interference strategy is calculated.

9. The method according to claim 8, characterized in that, The comprehensive learning performance evaluation value of the time anchor interference strategy is calculated, including: Obtain the preset forgetting curve parameters corresponding to the weak knowledge points. The forgetting curve parameters are pre-calibrated based on the knowledge point type and cognitive complexity level of the weak knowledge points. Based on the immediate assessment results, the delayed assessment results, and the preset forgetting curve parameters, the actual forgetting rate of the weak knowledge points is calculated. The actual forgetting rate is compared with the standard forgetting rate in the preset forgetting curve parameters to obtain the forgetting inhibition effect coefficient; Based on the immediate evaluation results and the forgetting inhibition effect coefficient, the comprehensive learning effect evaluation value is calculated.

10. A virtual reality classroom simulation device, characterized in that, The device includes: The acquisition module is used to acquire multimodal physiological signals of learners in the virtual classroom. A module is established to extract rhythmic feature parameters based on the multimodal physiological signals, establish a group cognitive rhythm model, calculate the current group's cognitive rhythm phase, and predict the evolution trend of the cognitive rhythm phase within a future time window; the cognitive rhythm phase includes at least the attention rise phase, attention plateau phase, attention decline phase, and attention trough phase; The identification module is used to monitor the learners' understanding of the current knowledge points and identify weak knowledge points; The determination module is used to determine the time anchor interference strategy based on the characteristics of the weak knowledge points, the cognitive rhythm phase of the current learner group, and the phase evolution trend of the future phase. The control module is used to control the learner's subjective time perception through VR multi-sensory manipulation when the time anchor interference strategy is determined to be implemented, so that the subjective dwell time of the weak knowledge points is stretched and / or the subjective dwell time of the mastered knowledge points is compressed; the VR multi-sensory manipulation means include one or more of visual refresh rate gradient modulation, spatial reference system displacement, multi-sensory delay modulation and dynamic time reference system embedding. The evaluation module is used to evaluate the learning effect after the implementation of the time anchor interference strategy, obtain the evaluation results, and update the parameters of the group cognitive rhythm model based on the evaluation results.

Images