A psychotherapy system and method integrating virtual reality scenarios

Abstract

This invention discloses a system and method for psychotherapy that integrates virtual reality scenes, belonging to the interdisciplinary field of psychotherapy and virtual reality. It addresses the problems of fixed scenes, delayed adjustment, weak interactivity, and lack of cultural adaptability in traditional psychotherapy. The method includes using non-contact biosensors to capture facial blood oxygen fluctuations, combining this with a hemodynamic model to remove artifacts, and then quantifying these fluctuations into emotional energy values ​​based on an individual emotional sample set and an incremental learning algorithm to generate scene seeds containing parameters such as color temperature. A virtual physics engine is then driven to reconstruct elements such as light and shadow, and fluids, forming a physically-level dynamic scene. Inertial sensors capture user movements and convert them into virtual pressure gradients, constructing a coupled feedback loop between movement and scene. Psychological trends are predicted through hierarchical Fourier analysis of the seed sequence. This invention achieves real-time adaptation and proactive intervention between psychological states and virtual scenes, enhancing user immersion and cultural and emotional resonance, and improving the accuracy and effectiveness of psychotherapy.

Description

Technical Field

[0001] This invention relates to psychotherapy techniques in the field of medical and health care, and more specifically, to a system and method for psychotherapy that integrates virtual reality scenarios. Background Technology

[0002] Current virtual reality psychotherapy systems have significant limitations in practical applications. Most systems use static, predefined modes for their virtual scene content and parameters. While an initial scene is generated based on the user's historical diagnostic data before system operation, these scene parameters are difficult to substantially adjust throughout the treatment process. Even some systems that claim dynamic adjustment capabilities, such as the Chinese patent disclosure of a psychological catharsis interaction method and system based on virtual reality and artificial intelligence technology (application number 202410884339.6), although mentioning real-time acquisition of user physiological signals and scene adjustment, rely heavily on user feedback data or simple switching based on preset thresholds, failing to achieve immediate synchronization with the user's emotional fluctuations.

[0003] A Chinese patent application also discloses a VR psychotherapy system, application number 202510729977.5. This document regulates physiological state by introducing multiple body-body interactions combined with physiological indicator feedback. Essentially, it reduces the reliance on virtual scenes, which reflects the inadequacy of the current technology's ability to dynamically adjust scenes. This mode, which relies on historical data or non-real-time adjustment, makes the treatment content always different from the user's current psychological needs, directly weakening the therapeutic effect.

[0004] These problems stem from the lack of an effective dynamic mapping mechanism in the system, which fails to smoothly translate the user's real-time psychological state into adjustable parameters for the virtual scene. Current technological development often focuses on optimizing the visual presentation and auditory effects of the scene, aiming to enhance immersion, but avoids the core contradiction of real-time matching between scene parameters and the user's psychological state. The root of this contradiction lies in the flawed way of perceiving the user's state. Existing systems often employ delayed and discrete perception. Even with physiological monitoring components like those in the two patents mentioned above, capable of collecting data such as heart rate and blood oxygen saturation, data processing is mostly reactive, such as summarizing physiological signals over a period before triggering adjustments, rather than continuously capturing and predictively integrating the user's state. This results in a persistent time lag between perception and scene adjustment, preventing the formation of a closed-loop response.

[0005] In virtual reality psychotherapy, the pursuit of immersive and visually realistic scenes often exacerbates the mismatch between the therapeutic content and the user's real-time psychological state. To achieve highly realistic scene presentation, the system needs to pre-define a large number of underlying parameters, including changes in light and shadow, fluid movement trajectories, and particle distribution density. While the fixed design of these parameters ensures the continuity and realism of the scene, it sacrifices the flexibility of adjustment. When a user experiences emotional fluctuations, such as a sudden increase in anxiety, the predefined scene parameters cannot be adjusted in time. Instead, the fixed bright light or rapidly moving scene elements may stimulate the user's nerves, further amplifying psychological discomfort. The two patents mentioned above also fail to resolve this contradiction. The former replaces scene adjustment with physical interaction, while the latter lacks parameter-level instantaneous response in scene adjustment. It is evident that within the existing technological framework, the conflict between scene realism and real-time adaptability is difficult to reconcile, becoming a key issue restricting the improvement of therapeutic effects. Summary of the Invention

[0006] To address the problems existing in the prior art, the present invention aims to provide a system and method for psychotherapy that integrates virtual reality scenarios, enabling real-time adaptation and proactive intervention of psychological states with virtual scenarios, enhancing user immersion and cultural and emotional resonance, and improving the accuracy and effectiveness of psychotherapy.

[0007] To solve the above problems, the present invention adopts the following technical solution.

[0008] Firstly, a method for psychotherapy that integrates virtual reality scenarios includes the following steps:

[0009] Step S1: Collect physiological signals reflecting the user's psychological state through multimodal biosensors, convert the physiological signals into quantified emotional energy values, and generate a parameter sequence to drive the virtual scene based on the emotional energy values. The parameter sequence includes parameters that affect the visual and physical properties of the scene.

[0010] Step S2: Use parameter sequences to control the physics engine in the virtual reality scene in real time, and dynamically adjust the scene's lighting, color, geometry and particle system motion state to build a dynamic virtual environment that is linked to the user's emotions.

[0011] Step S3: Construct a virtual pressure field in a dynamic virtual environment, capture user limb movements through sensors and calculate the perturbation effect on the virtual pressure field, and input the perturbation effect as a feedback quantity into step S1 to adjust the generation of parameter sequence.

[0012] Step S4: Perform time series analysis on the parameter sequence to predict the user's emotional change trend in the future preset time period. Based on the predicted emotional change trend, pre-adjust the parameters of the virtual scene before the corresponding emotional change occurs.

[0013] Step S5: Encode the preset cultural symbol elements into a parameter set that can be called by the physics engine, and control the timing of the fusion of the parameter set in the virtual scene and at least one visual performance intensity parameter according to the predicted emotional change trend.

[0014] Further, step S1 specifically includes:

[0015] Step S11: Filter and remove artifacts from the raw physiological signals from the multi-channel biosensor to extract the effective signal components related to emotional fluctuations.

[0016] Step S12: Based on the preset psychophysiological correspondence model, the effective signal components are calculated as emotional energy values ​​that characterize the current psychological arousal level.

[0017] Step S13: Based on the different numerical ranges of the emotional energy value, call different nonlinear mapping functions to generate a parameter sequence, where the parameters include color temperature parameters, spatial curvature parameters, and particle motion entropy parameters.

[0018] Furthermore, sub-step S12 also includes the following sub-steps:

[0019] Step S121: Before establishing a model corresponding to the frequency of general physiological signal fluctuations and the degree of psychological arousal, collect physiological signal data of users in a preset calm state to obtain the arousal benchmark threshold, and collect physiological signal data corresponding to different emotions actively reported by users and label them as individual emotion sample sets.

[0020] Step S122: Based on the arousal baseline threshold and individual emotion sample set initialized in step S121, the model is initialized, and the user's physiological signal fluctuations and emotional feedback data are collected in real time. The arousal weights corresponding to different fluctuation frequencies in the model are adjusted through an incremental learning algorithm.

[0021] Further, step S2 specifically includes:

[0022] Step S21: Decouple the parameter sequence into independent control commands that control the light source, mesh model and particle emitter respectively;

[0023] In step S22, the physics engine executes independent control commands to synchronously adjust lighting effects, object deformation, and particle motion, and establishes physical linkages between these elements.

[0024] Step S23 also maps the emotional energy value to physical field parameters and inputs the physical field parameters into step S22 to globally influence the physical behavior of scene elements.

[0025] Furthermore, step S22 also includes a load optimization step:

[0026] Step S221: Based on the priority of the currently activated independent control command, dynamically adjust the calculation weight of the physical linkage relationship between different scene elements.

[0027] Step S222: Monitor the performance load of the rendering system in real time. When the load exceeds the preset threshold, automatically reduce the calculation accuracy and / or parameter adjustment frequency of the secondary physical linkage relationship to ensure that the rendering frame rate is not lower than the preset threshold.

[0028] Furthermore, step S3 specifically includes:

[0029] Step S31: Based on the real-time geometric structure and material properties of the dynamic virtual environment, define the range of action and the initial pressure reference value of the virtual pressure field.

[0030] Step S32: Analyze the speed, direction and amplitude characteristics of the user's actions, calculate the impulse generated by the virtual pressure field, and quantify the impulse into pressure gradient change value;

[0031] Step S33: Based on the pressure gradient change value and the current scene state, calculate the feedback amount used to correct the parameter sequence, thereby closing the real-time interaction loop from action to scene.

[0032] Furthermore, step S32 also includes an action validity determination step:

[0033] Step S321: Filter out invalid or minor actions using the spatiotemporal feature matrix of the actions, and extract only valid action feature data that meets the preset conditions;

[0034] Step S322: When calculating the impulse, a weighting coefficient based on the spatiotemporal context is introduced to more accurately evaluate the actual perturbation effect of the action on the virtual field.

[0035] Furthermore, step S4 specifically includes:

[0036] Step S41: Perform multi-scale analysis based on frequency domain transformation on the parameter sequence to extract short-term and long-term fluctuation components that characterize the user's emotional cycle.

[0037] Step S42: Using a probabilistic prediction model trained on historical data, the extracted periodic components are extrapolated, and a probabilistic prediction result of the trend of emotional energy value changes within a preset time period is output.

[0038] Step S43: Based on the probabilistic prediction results, preload and prepare the corresponding scene compensation parameter package, and automatically activate the compensation effect when specific triggering conditions are met.

[0039] Furthermore, step S5 specifically includes:

[0040] Step S51: Deconstruct the preset Chinese cultural elements into parameters that can be called by the physics engine. The parameters include fluid dynamics parameters, motion trajectory parameters, and geometric structure parameters.

[0041] Step S52: Link the probabilistic prediction results with the symbolic field activation logic, and set the threshold conditions for the emotional energy value and the rules for the regulation of visual expression of different cultural elements in their manifestation and evolution.

[0042] Step S53: According to the activation logic and visual performance control rules set in step S52, during the dynamic scene reconstruction process in step S2, the deconstructed cultural parameters are called and integrated. The visual performance control rules are reflected by adjusting at least one parameter among the transparency, brightness, saturation or size of the cultural elements.

[0043] Secondly, a psychotherapy system integrating virtual reality scenarios, applied to the aforementioned method of psychotherapy integrating virtual reality scenarios, includes:

[0044] The emotion mapping module is used to collect physiological signals reflecting the user's psychological state through multimodal biosensors, convert the physiological signals into quantified emotion energy values, and generate a parameter sequence to drive the virtual scene based on the emotion energy values. The parameter sequence contains parameters that affect the visual and physical properties of the scene.

[0045] The scene control module is used to control the physics engine in the virtual reality scene in real time using parameter sequences, and dynamically adjust the scene's lighting, color, geometry and particle system motion state to build a dynamic virtual environment that is linked to the user's emotions.

[0046] The motion feedback module is used to construct a virtual stress field in a dynamic virtual environment. It captures the user's body movements through sensors and calculates the perturbation effect on the virtual stress field. The perturbation effect is then input as a feedback quantity into the emotion mapping module to adjust the generation of parameter sequences.

[0047] The emotion prediction module is used to perform time-series analysis on parameter sequences to predict the trend of user emotion changes in the future within a preset time period. Based on the predicted emotion change trend, the parameters of the virtual scene are pre-adjusted before the corresponding emotion change occurs.

[0048] The symbol fusion module is used to encode preset cultural symbol elements into a parameter set that can be called by the physics engine. Based on the predicted emotional change trend, it controls the timing of the fusion of the parameter set in the virtual scene and at least one visual performance intensity parameter.

[0049] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0050] (1) This solution uses non-contact biosensors to collect facial blood oxygen fluctuations, combines hemodynamic models to eliminate artifacts, and establishes a personalized mapping model through individual emotion sample sets and incremental learning algorithms to ensure the accuracy of emotion energy value quantification. Then, using a motion coupling feedback loop, the solution captures user actions through inertial sensors and converts them into virtual pressure gradients, dynamically correcting scene parameter seeds. This allows the virtual scene to not only be reconstructed in real time according to psychological state, but also to respond to user actions to form an interactive closed loop. This end-to-end dynamic adaptation avoids the problems of fixed scenes and lagging adjustment in traditional psychotherapy, allowing users to gain a stronger sense of immersion in a virtual environment that closely matches their own state. The injection of low-frequency sound waves during negative emotion prediction can achieve proactive intervention, greatly improving the timeliness and effectiveness of emotion regulation.

[0051] (2) This solution deconstructs Chinese cultural elements such as ink painting and calligraphy into parameters that can be recognized by the physical engine. Then, based on the trend of psychological state, it sets the activation threshold and intensity correlation logic of the symbol field. Finally, through the linkage adjustment of cultural element parameters and scene physical parameters, the cultural symbols and scene elements such as light and shadow and fluids form a unified atmosphere. This design breaks through the limitation of existing VR psychotherapy relying on Western visual elements. Chinese cultural symbols can evoke deep emotional identification from users. For example, the tranquility brought by traditional ink painting and the calm atmosphere of the scene can not only consolidate the effect of emotional regulation, but also reduce users' resistance to treatment. It is especially suitable for users with a cognitive foundation of Chinese culture, and significantly improves the compliance and long-term effect of treatment. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0053] Figure 1 This is a flowchart of a psychotherapy method integrating virtual reality scenes according to the present invention;

[0054] Figure 2 This is a flowchart illustrating the various modules of a psychotherapy system that integrates virtual reality scenarios according to the present invention. Detailed Implementation

[0055] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0056] Please see Figures 1 to 2 A psychotherapy method that integrates virtual reality scenarios includes the following steps:

[0057] Step S1: Collect physiological signals reflecting the user's psychological state through multimodal biosensors, convert the physiological signals into quantified emotional energy values, and generate a parameter sequence to drive the virtual scene based on the emotional energy values. The parameter sequence includes parameters that affect the visual and physical properties of the scene.

[0058] Step S2: Use parameter sequences to control the physics engine in the virtual reality scene in real time, and dynamically adjust the scene's lighting, color, geometry and particle system motion state to build a dynamic virtual environment that is linked to the user's emotions.

[0059] Step S3: Construct a virtual pressure field in a dynamic virtual environment, capture user limb movements through sensors and calculate the perturbation effect on the virtual pressure field, and input the perturbation effect as a feedback quantity into step S1 to adjust the generation of the parameter sequence.

[0060] Step S4: Perform time series analysis on the parameter sequence to predict the user's emotional change trend in the future preset time period. Based on the predicted emotional change trend, pre-adjust the parameters of the virtual scene before the corresponding emotional change occurs.

[0061] Step S5: Encode the preset cultural symbol elements into a parameter set that can be called by the physics engine, and control the timing of the fusion of the parameter set in the virtual scene and at least one visual performance intensity parameter according to the predicted emotional change trend.

[0062] Step S1 further includes the following steps:

[0063] Step S11: Filter and remove artifacts from the raw physiological signals from the multi-channel biosensor to extract the effective signal components related to emotional fluctuations.

[0064] Step S12: Based on the preset psychophysiological correspondence model, the effective signal components are calculated as emotional energy values ​​that characterize the current psychological arousal level.

[0065] Step S13: Based on the different numerical ranges of the emotional energy value, call different nonlinear mapping functions to generate a parameter sequence, where the parameters include color temperature parameters, spatial curvature parameters, and particle motion entropy parameters.

[0066] Sub-step S12 also includes the following sub-steps:

[0067] Step S121: In the initial stage, collect the user's physiological baseline data in a calm state and labeled data under specific emotions to establish an individualized benchmark and sample library.

[0068] Step S122: During the application process, the parameters and weights of the corresponding psychophysiological model are dynamically optimized by continuously combining the user's real-time physiological data and subjective feedback through machine learning algorithms.

[0069] In this embodiment, a multimodal biosensor is first used as the data acquisition device. This includes a near-infrared camera with ambient light compensation for acquiring blood oxygen saturation signals, a wrist-worn photoelectric heart rate sensor for acquiring heart rate variability (HRV) signals, and a fingertip-type skin conductance response (GSR) sensor for acquiring skin conductance activity signals. Sampling is performed in zones targeting blood flow-sensitive areas of the user's face, such as the cheeks and forehead, where blood flow is dense and highly correlated with emotional changes. This zoned design aims to improve the targeting of signal acquisition and avoid the potential for insufficient signal representativeness caused by sampling from a single area. During the acquisition process, physiological noise, such as breathing, heartbeat interference, and changes in ambient light, can introduce signal artifacts. Therefore, a multi-source signal fusion processing model is needed. This model, based on Kalman filtering and blind source separation algorithms, analyzes the acquired raw signals and removes artifacts by identifying abnormal components in the signal that do not conform to hemodynamic characteristics. Ultimately, the multimodal physiological fluctuation components directly related to psychological arousal are retained.

[0070] Before establishing a general model corresponding to the frequency of multimodal physiological signal fluctuations and psychological arousal, multimodal physiological signal data of users in a preset calm state, such as sitting still or in a relaxed state, is first collected. Based on this data, the arousal baseline threshold for the user is determined. This threshold serves as a reference standard for subsequently judging the direction of emotional fluctuations, such as being higher or lower than the calm state. At the same time, users are guided to actively report different typical emotions, such as anxiety, pleasure, and calmness, and multimodal physiological signal data of the corresponding emotional states are collected simultaneously. The correspondence between signal data and emotion types is labeled as an individual emotion sample set. The establishment of the individual sample set can effectively reduce the individual difference error of the general model among different users and improve the model's adaptability to specific users. The model is dynamically optimized as follows: Initializing the model uses the acquired arousal baseline threshold and individual emotional sample set as initial inputs. In subsequent practical applications, real-time collection of multimodal physiological signal data and user-initiated emotional state information is used as incremental samples input to the model. An incremental learning algorithm continuously adjusts the arousal weights corresponding to different signal features in the model. This algorithm allows the model to adaptively update according to dynamic changes in user physiological characteristics, such as long-term emotional state trends and subtle changes in physiological function, avoiding long-term accuracy decline due to fixed parameters. Finally, the optimized model accurately quantifies the real-time collected multimodal physiological changes into an emotional energy value that characterizes the intensity of psychological state. The emotional energy value, as a comprehensive indicator of user psychological arousal, is a scalar obtained by linearly weighting and fusing multimodal physiological signals. Its calculation formula is: Emotional energy value E equals the product of weight coefficient α and blood oxygen saturation O2, plus the weight coefficient β and the arousal index. The linear weighted fusion model, which combines the product of heart rate variability (HRV) and the weighting coefficient γ with the skin conductance response (GSR), aims to integrate signals reflecting different physiological dimensions into a unified quantitative value of emotional state. Blood oxygen saturation is typically expressed as a percentage and is related to the body's metabolism and arousal level; heart rate variability is expressed in milliseconds and reflects the activity state of the autonomic nervous system; and skin conductance response is expressed in micro-Siemens and is related to the degree of emotional arousal. Since the original physical dimensions of each physiological signal are different, direct weighted summation would result in a problem of inconsistent dimensions in mathematics. Therefore, before actual calculation, the original signals need to be normalized and preprocessed. That is, based on the individual baseline or a preset normal range, the signal values ​​are converted into dimensionless relative values ​​or standardized scores to ensure the mathematical rigor of the formula. The weighting coefficients α, β, and γ are dimensionless parameters that represent the relative contribution of the corresponding physiological signal to the final emotional energy value. Their values ​​are obtained by training an individual emotional sample set through regression analysis or machine learning methods, with the goal of minimizing the model's prediction error.This linear weighted sum has the advantages of simple structure and strong interpretability, which facilitates real-time calculation. Furthermore, the weight coefficients can be personalized for different users to adapt to the differences in individual physiological response characteristics.

[0071] in, It is expressed as an emotional energy value, which is a scalar representing the overall intensity of an emotional state. The value range is usually a normalized value, such as 0-1 or 0-100.

[0072] Blood oxygen saturation is expressed as a percentage (%), indicating the oxygen content in the blood. The normal range is approximately 95%-100%.

[0073] It is expressed as heart rate variability, usually in milliseconds (ms), which represents the change in heart rate interval. Common indicators include SDNN, and the normal range is about 20-100 ms.

[0074] It is expressed as skin conductance response, usually measured in microsiemens (μS), which represents the change in skin conductivity. The normal range varies from person to person and is usually 0.1-20 μS.

[0075] The weighting coefficient for blood oxygen saturation is dimensionless. The degree of contribution to emotional energy value;

[0076] It is represented as the weighting coefficient of heart rate variability, dimensionless, and indicates the degree of contribution of HRV to the emotional energy value;

[0077] It is represented by the weighting coefficient of the skin conductance response, which is dimensionless and indicates the degree of contribution of GSR to the emotional energy value.

[0078] After quantifying the emotional energy value, scene parameter seeds are generated. Based on the specific range of the output emotional energy value, such as low, medium, and high emotional energy ranges, differentiated nonlinear functions are used for parameter mapping. The reason for choosing nonlinear functions is that the relationship between the user's psychological state and the basic color temperature, spatial curvature, and particle motion entropy values ​​of the virtual scene parameters is not linear. For example, in the low emotional energy range, corresponding to a calm or depressed state, a small change in the emotional energy value may require a large adjustment of the scene parameters to achieve obvious scene feedback. In the high emotional energy range, corresponding to an excited or anxious state, the adjustment range of the scene parameters needs to be controlled to avoid overstimulation. The differentiated design is reflected in the use of different forms of nonlinear functions for different emotional energy ranges to ensure that the mapping relationship between the emotional energy value and the scene parameters in each range can accurately match the changing needs of the psychological state. Meanwhile, during the mapping process, differentiated weights are assigned to three types of scene parameters: base color temperature, spatial curvature, and particle motion entropy. The weight allocation is determined based on the degree of influence of each parameter on the user's psychological perception. For example, the effect of base color temperature on mood regulation may be stronger than that of particle motion entropy, so a higher weight is assigned in the corresponding range. Finally, through the above differentiated nonlinear mapping and weight allocation, scene parameter seeds containing information on the three types of scene parameters are generated.

[0079] In some embodiments of the present invention, step S2 further includes the following steps:

[0080] Step S21: Decouple the parameter sequence into independent control commands that control the light source, mesh model and particle emitter respectively;

[0081] In step S22, the physics engine executes independent control commands to synchronously adjust lighting effects, object deformation, and particle motion, and establishes physical linkages between these elements.

[0082] Step S23 also maps the emotional energy value to physical field parameters and inputs the physical field parameters into step S22 to globally influence the physical behavior of scene elements.

[0083] Sub-step S22 also includes the following sub-steps:

[0084] Step S221: Based on the priority of the currently activated independent control command, dynamically adjust the calculation weight of the physical linkage relationship between different scene elements.

[0085] Step S222: Monitor the performance load of the rendering system in real time. When the load exceeds the preset threshold, automatically reduce the calculation accuracy and / or parameter adjustment frequency of the secondary physical linkage relationship to ensure that the rendering frame rate is not lower than the preset threshold.

[0086] In this embodiment, the primary task is to decouple and convert scene parameter seeds into commands. This serves as the connection between psychological state parameters and the virtual physics engine. Since scene parameter seeds contain three related parameters—color temperature, spatial curvature, and particle motion entropy—and the virtual physics engine's control of scene elements relies on independent executable commands, the seeds must first be decoupled. This separates the three types of parameters from a mixed state into independent parameter variables, preventing mutual interference between parameters that could lead to engine command recognition errors. Subsequently, commands are mapped for each independent parameter: the color temperature parameter essentially reflects the spectral characteristics of the light source, so it is converted into a light source spectrum control command. By adjusting the proportion of different wavelengths of light in the spectrum, the color temperature is dynamically changed. The spatial curvature parameter determines the spatial form of the virtual scene, and the spatial structure of the virtual scene is constructed by a mesh model. Therefore, it is converted into a scene mesh vertex offset command, adjusting the curvature and spatial dimensionality of the scene by changing the spatial coordinates of the mesh vertices. The particle motion entropy value characterizes the disorder of particle motion; the higher the entropy value, the more random the particle motion. Accordingly, it is converted into a particle velocity randomization command, achieving precise control of motion disorder by adjusting the random fluctuation range of particle velocity. Through this process, abstract scene parameter seeds are transformed into specific operation instructions that the physics engine can directly execute.

[0087] Based on the aforementioned independent commands, real-time modulation of scene physical attributes is performed, and coupling relationships between elements are established to ensure the integrity and realism of the virtual scene. During the physical attribute modulation stage, the physical parameters of the three main elements—light and shadow, fluid, and particle systems—are adjusted according to corresponding commands: For the light and shadow system, the wavelength distribution, intensity, and illumination angle of light are adjusted according to the light source spectrum control command to achieve changes in scene brightness and color atmosphere; for the fluid system, the flow boundary and path of the fluid are adjusted in conjunction with the scene mesh vertex offset command to adapt the fluid shape to the curvature of the scene space; for the particle system, the initial velocity, direction of motion, and collision rules of particles are controlled according to the particle velocity randomization command to present a motion state matching the particle motion entropy value. On this basis, it is necessary to further establish coupling relationships between elements, because elements in a virtual scene do not exist in isolation. For example, changes in light and shadow affect the visual reflection effect of fluids, and particle motion also interacts with fluids, such as particles floating or sinking in fluids. The establishment of coupling relationships is achieved by defining the association rules of changes in the physical parameters of each element, ensuring that other related elements can respond synchronously when one element is adjusted, avoiding scene logic breaks caused by independent changes in each element, and enhancing the user's immersion in the virtual scene.

[0088] By optimizing coupling relationships through dynamic weight allocation, scene adjustments can better align with the needs of the current psychological state. Since independent output commands have different priorities—for example, when emotional energy values ​​change significantly, color temperature adjustment may have a stronger guiding effect on user emotions than particle motion entropy adjustment—the light source spectrum control command corresponding to color temperature has a higher priority. Therefore, dynamic weights need to be allocated to the coupling relationships between scene elements based on command priority. Specifically, first, the currently active high-priority command is determined, and then the weights of coupling pairs associated with that command, such as the coupling between light and shadow and fluids, or the coupling between light and shadow and particle systems, are increased. The weight adjustment formula is: in As the initial value, it represents the default weight of the coupling pair when there are no external influences. Based on instruction priority coefficient and emotional energy normalization value The product is used to adjust weights at runtime; higher priority instructions... When the value is large or the emotional energy is high, the adjustment range is larger, ensuring that the scene adjustment is more in line with the current psychological needs. The formula uses linear addition, which is simple, efficient, and easy to calculate in real time, with the final weight... It is the sum of the basic weights and the adjustment terms, which makes the weights change dynamically with the system state.

[0089] Represented as the adjusted new weights, it is a scalar representing the updated weight values ​​of the coupling pair;

[0090] Represented as the basic weight, it is a scalar that represents the initial weight of the coupling pair, and is usually set during system initialization;

[0091] Represented as an instruction priority coefficient, it is a dimensionless coefficient that indicates the priority of the currently active instruction; the larger the value, the higher the priority.

[0092] It represents the normalized value of the current emotional energy value. It is a dimensionless scalar with a range of [0,1], representing the intensity of the user's emotional energy. 0 represents the lowest and 1 represents the highest.

[0093] The above operations allow these coupling pairs to dominate the scene adjustment process, making the interactions between their elements more significant; while coupling pairs corresponding to low-priority instructions have their weight reduced, minimizing their impact on the overall scene changes. This dynamic weight allocation mechanism ensures that the focus of scene adjustment always aligns with the needs corresponding to the current emotional energy value, avoiding excessive adjustments to irrelevant elements that might distract the user and improving the targeted nature of psychological adjustment.

[0094] The smoothness of scene operation is ensured by degrading coupling relationships. The number and complexity of element coupling relationships in a virtual scene directly affect the computational load of the physics engine. Therefore, it is necessary to monitor the load status of the physics engine in real time. By collecting indicators such as CPU utilization, frame rate, and memory usage, it is determined whether the load exceeds a preset threshold, such as CPU utilization > 85% or frame rate < 30fps. When the load is too high, the coupling relationship degradation mechanism is automatically triggered: First, the coupling pairs with the strongest correlation to the emotional energy value are selected. Specifically, the Pearson correlation coefficient between each coupling pair and the emotional energy value is calculated, and the top K coupling pairs with the highest correlation coefficients are retained, for example, the top 30%. These coupling pairs are retained to ensure that the core scene feedback is not lost. At the same time, the parameter adjustment step size of the retained coupling pairs is adjusted. For example, the original parameter adjustment every 10 milliseconds is changed to adjustment every 20 milliseconds. By reducing the frequency of parameter calculation and update, the engine's computational load is reduced. This achieves a balance between engine load and scene smoothness while ensuring the core scene effect.

[0095] By transforming the electromagnetic wave spectrum distribution parameters, the adaptability of the scene to the psychological state is enhanced. As an indicator for quantifying the user's psychological state, the change of emotional energy value needs to be conveyed to the user through more refined light environment adjustments. The electromagnetic wave spectrum distribution directly determines the physical properties of light and the visual perception effect. For example, high emotional energy value usually corresponds to an excited and active psychological state, which can be matched with a spectrum containing more high-frequency electromagnetic waves, such as red light and orange light; low emotional energy value corresponds to a calm and soothing state, which can be matched with a spectrum containing more low-frequency electromagnetic waves, such as blue light and green light. Therefore, an algorithm corresponding energy value and spectral power density is adopted to directly convert the specific value of emotional energy value into the power density of different frequency components in the electromagnetic wave spectrum. The algorithm uses a preset mapping rule to establish a quantitative correlation between the increase or decrease of emotional energy value and the change in power density of electromagnetic waves at a specific frequency. For example, for every 1 unit increase in emotional energy value, the power density of the red light band 620-750nm increases by 5%. This transforms the abstract quantitative index of psychological state into physical parameters that can directly control the scene's light environment, making the changes in the scene's light environment more accurately match the user's current psychological state, and further enhancing the virtual scene's guiding and regulating effect on emotions.

[0096] In some embodiments of the present invention, step S3 further includes the following steps:

[0097] Step S31: Based on the real-time geometric structure and material properties of the dynamic virtual environment, define the range of action and the initial pressure reference value of the virtual pressure field.

[0098] Step S32: Analyze the speed, direction and amplitude characteristics of the user's actions, calculate the impulse generated by the virtual pressure field, and quantify the impulse into pressure gradient change value;

[0099] Step S33: Based on the pressure gradient change value and the current scene state, calculate the feedback amount used to correct the parameter sequence, thereby closing the real-time interaction loop from action to scene.

[0100] Step S32 further includes the following steps:

[0101] Step S321: Filter out invalid or minor actions using the spatiotemporal feature matrix of the actions, and extract only valid action feature data that meets the preset conditions;

[0102] Step S322: When calculating the impulse, a weighting coefficient based on the spatiotemporal context is introduced to more accurately evaluate the actual perturbation effect of the action on the virtual field.

[0103] In this embodiment, the virtual scene, after processing, exhibits specific physical properties, including spatial structure, element distribution, and dynamic state. These characteristics directly determine the environment in which the pressure field exists. To ensure that the pressure field accurately reflects the physical interaction patterns within the scene, the initial boundary conditions of the pressure field need to be set according to the parameters of the current physical-level dynamic scene. Specific setting methods include: defining the effective range of the pressure field using a bounding box or the convex hull of the scene mesh based on the scene's geometry; and setting the initial pressure reference value of the pressure field based on the density distribution of scene elements, such as through density field data from a particle system or fluid simulator, for example, setting a higher reference pressure in densely populated areas and a lower reference pressure in sparsely populated areas. Through this adaptive setting, the pressure field becomes an intermediary connecting user actions and the virtual scene, providing a reasonable physical basis for subsequent calculations of scene disturbances caused by actions.

[0104] Inertial sensors, as motion acquisition devices, can record the motion data of a user's limbs in three-dimensional space in real time. This data includes kinematic characteristics such as acceleration and angular velocity, directly reflecting the intensity, direction, and speed of the action. After acquiring the raw motion data, it needs to be processed in conjunction with the set pressure field boundary conditions. The boundary conditions define the physical rules of the pressure field. For example, the impact of an action at the edge of the scene may be weaker than that in the center. Therefore, when calculating the impulse of the action on the pressure field, the kinematic characteristics of the action must be combined with the spatial constraints in the boundary conditions to ensure that the impulse calculation conforms to the physical logic of the pressure field. The impulse, as a physical quantity that measures the effect of the action on the pressure field, is related to the force and duration of the action. Finally, it is quantified into a virtual pressure gradient through a preset conversion rule, which is the rate of change of pressure distribution in the scene space. This gradient value intuitively reflects the intensity and direction of the pressure field disturbance caused by the action.

[0105] Multi-axis data collected by inertial sensors typically covers three spatial axes: X, Y, and Z, containing motion information of the user's limbs in different directions. However, this data may contain unintentional movements, such as slight tremors or environmental interference signals. To eliminate invalid information, a spatiotemporal feature matrix of the action needs to be constructed in conjunction with the pressure field boundary conditions. This matrix integrates the continuous characteristics of the action in the time dimension, such as the duration from start to finish, and the trajectory characteristics in the spatial dimension, such as the position of the action within the scene. By analyzing the regularity of the data in the matrix, such as whether it conforms to the movement patterns of conscious human actions, the validity of the action is determined. For actions determined to be valid, their core kinematic characteristics are retained; their acceleration reflects the change in the force of the action, and their angular velocity reflects the rotational trend of the action. This ensures that subsequent impulse calculations are based on real and meaningful action information, avoiding erroneous scene responses caused by invalid actions.

[0106] Based on the obtained effective kinematic characteristics of the action, the initial calculation of impulse needs to consider the acceleration of the action, the duration of action, and the preset equivalent mass parameters in the virtual scene to simulate the inertial relationship between the action and the scene. However, a single impulse calculation may ignore the differences in the spatiotemporal dimensions of the action. For example, a fast and short action and a slow and continuous action have different perturbation patterns on the pressure field. Therefore, a spatiotemporal weighting coefficient needs to be introduced for optimization: the time weighting coefficient adjusts the impulse weight according to the duration of the action, and actions with longer durations may be given higher weights; the spatial weighting coefficient adjusts the weight according to the position of the action in the pressure field, such as the position near the core or edge of the scene, so that the impulse more closely matches the boundary characteristics of the pressure field. The optimized impulse can more accurately reflect the actual perturbation effect of the action on the pressure field, and is then converted into a virtual pressure gradient through a mapping algorithm to ensure that the magnitude and distribution of the gradient value are consistent with the real impact of the action.

[0107] The pressure field perturbation value is quantified from the pressure gradient, reflecting the intensity of the user's action on the scene. The current scene particle density reflects the distribution of scene elements; areas with high particle density are typically more sensitive to pressure perturbations. Feedback weights are dynamically adjusted based on these two parameters: if the perturbation value is large and the particle density is high, it indicates a significant impact of the action on the scene, and the feedback weight is increased accordingly, making the action's correction of scene parameters stronger; conversely, the weight is reduced to avoid over-correction leading to scene instability. The adjusted feedback weights are used to correct the generated scene parameter seeds. For example, when the action is intense, corresponding to a high perturbation value, the parameter component of scene particle density is increased through weight adjustment, making the scene exhibit a more obvious response. This complete link from action acquisition to parameter correction forms a closed coupling field between user actions and the virtual scene, ensuring that the scene can respond to user actions in real time and accurately. Simultaneously, the feedback of actions influences the generation of subsequent scenes, achieving deep interaction between the user and the virtual environment in psychotherapy.

[0108] In some embodiments of the present invention, step S4 further includes the following steps:

[0109] Step S41: Perform multi-scale analysis based on frequency domain transformation on the parameter sequence to extract short-term and long-term fluctuation components that characterize the user's emotional cycle.

[0110] Step S42: Using a probabilistic prediction model trained on historical data, the extracted periodic components are extrapolated, and a probabilistic prediction result of the trend of emotional energy value changes within a preset time period is output.

[0111] Step S43: Based on the probabilistic prediction results, preload and prepare the corresponding scene compensation parameter package, and automatically activate the compensation effect when specific triggering conditions are met.

[0112] In this embodiment, the output psychological and scene seed sequence is essentially a time-series data that changes dynamically over time. It includes both fluctuations in the user's psychological state and synchronous changes in virtual scene parameters such as color temperature and spatial curvature. Changes in the user's psychological state often exhibit implicit periodic patterns. For example, short-term emotional fluctuations may show short-term cycles of a few minutes depending on breathing and concentration levels, while long-term emotional trends may show medium-term cycles of tens of minutes depending on the treatment progress. Fourier analysis decomposes the time-domain sequence signal into multiple sinusoidal components in the frequency domain. Each component corresponds to a specific frequency and amplitude. This decomposition removes random noise from the sequence, highlighting the regular periodic components. The hierarchical design divides the periodic components into short-term and medium-term categories based on frequency range: typically, high-frequency components correspond to short-term cycles (e.g., 0.1-1Hz, corresponding to cycles of 1-10 minutes), reflecting the user's immediate emotional fluctuations; low-frequency components correspond to medium-term cycles (e.g., 0.01-0.1Hz, corresponding to cycles of 10-100 minutes), reflecting the overall trend of the user's emotions. After extracting the two types of periodic components, the peak times of amplitude within each period are further marked. The peak times correspond to the highest values ​​of emotional energy values ​​or scene parameters in the sequence. The time intervals and frequency of these times directly reflect the core information of the periodic pattern.

[0113] Because users' psychological states are influenced by various factors such as physiology and environment, their changes, while exhibiting periodic patterns, are not entirely deterministic events. Therefore, using a probabilistic prediction model rather than a deterministic model can more objectively reflect the uncertainty of trends. The model's input data consists of extracted short-term and medium-term cyclical features, including the length of the two types of cycles (e.g., a short-term cycle of 5 minutes and a medium-term cycle of 30 minutes), peak interval time (e.g., the interval between two consecutive short-term peaks of 4-6 minutes), peak intensity (e.g., the emotional energy value at the peak moment is 20%-30% higher than the average), and the current sequence's position within the cycle (e.g., currently in the rising phase of the short-term cycle). The model learns the correlation between these features and the occurrence of subsequent emotional energy peaks in historical data to establish a prediction function. For example, when the model identifies a feature combination where the current short-term cycle is in the rising phase and the medium-term cycle is 10 minutes before the peak, it will call upon the frequency of peak occurrences under this combination in historical data to calculate the probability of emotional energy peaks occurring at different time points within the next 10 minutes, a preset duration. Ultimately, the model outputs the trend result with the highest probability. For example, the probability of an emotional energy peak occurring within the next 5 minutes is 85%, and the probability of a peak occurring within the next 10 minutes is 60%. This result not only clarifies the trend direction, such as whether a peak will occur, but also quantifies the credibility of the trend through probability values.

[0114] If compensation parameters are calculated and loaded in real time only when a trend occurs, the computation time may cause a delay in compensation activation, missing the optimal opportunity for emotion guidance. Preloading, however, allows parameters to be stored in a cache and directly invoked when conditions are triggered, significantly shortening response time. The selection of compensation parameters directly matches the high-probability trend results: for example, when the trend result indicates a high probability of a negative emotion energy peak within the next 5 minutes, preloaded parameters include low-frequency sound wave generation parameters, such as a frequency of 20-50Hz and an intensity of 40-50 dB, which aligns with the original text's requirement for negative emotion pre-intervention; color temperature adjustment parameters, such as adjusting the color temperature from a cool 3000K to a warm 5000K to alleviate anxiety; and parameters reducing particle motion entropy, such as reducing particle velocity randomization by 20%, to make the scene more stable. If the trend result indicates that emotions are trending towards calm, preloaded parameters maintain scene stability, such as keeping the color temperature and particle motion entropy at their current state. While preloading parameters, it is necessary to set clear compensation activation trigger conditions to avoid parameters being activated erroneously or prematurely. Trigger conditions usually combine real-time monitoring data and prediction results. For example, when the real-time emotional energy value reaches a preset threshold, such as 15% higher than the baseline value of the calm state, and the current time is within the predicted peak window period, such as within the next 5 minutes, when both conditions are met, the system automatically triggers the execution of compensation parameters, so that the scene adjustment accurately matches the key nodes of emotional changes, ensuring the compensation effect while avoiding unnecessary scene disturbances that affect the user's immersive experience.

[0115] In some embodiments of the present invention, step S5 further includes the following steps:

[0116] Step S51: Deconstruct the preset Chinese cultural elements into parameters that can be called by the physics engine. The parameters include fluid dynamics parameters, motion trajectory parameters, and geometric structure parameters.

[0117] Step S52: Link the probabilistic prediction results with the symbolic field activation logic, and set the threshold conditions for the emotional energy value and the rules for the regulation of visual expression of different cultural elements in their manifestation and evolution.

[0118] Step S53: According to the activation logic and visual performance control rules set in step S52, during the dynamic scene reconstruction process in step S2, the deconstructed cultural parameters are called and integrated. The visual performance control rules are reflected by adjusting at least one parameter among the transparency, brightness, saturation or size of the cultural elements.

[0119] In this embodiment, the technical adaptation between abstract cultural symbols and the virtual physics engine is achieved by deconstructing the physical properties of Chinese cultural elements. Chinese cultural elements, such as traditional ink painting, calligraphy brushstrokes, paper-cutting patterns, and classical designs, mostly exist in abstract or figurative visual forms. However, the virtual physics engine can only recognize and process parameters with clearly defined physical properties. Specific deconstruction and mapping methods include:

[0120] The diffusion effect of ink wash elements is mapped to the diffusion radius and viscosity parameters of a fluid system; the trajectory and thickness variation of calligraphic brushstrokes are mapped to the motion trajectory and particle radius parameters of a particle system; the outline and hollow structure of paper-cut patterns are mapped to the topological structure parameters of the scene mesh and the transmittance parameters of the lighting system; the repetition and gradation patterns of classical patterns, such as cloud patterns and meander patterns, are mapped to the tiling parameters of the texture sampler and the interpolation parameters of the gradient. Similarly, if unprocessed cultural elements such as light and shadow spectra, fluid motion parameters, and model mesh parameters are directly introduced into the scene, the engine cannot achieve dynamic rendering and interaction. Therefore, the physical deconstruction of cultural elements must be completed first. Specifically, for different types of Chinese cultural elements, their visual characteristics and morphological patterns are extracted and mapped to physical parameters that the engine can recognize. For example, when deconstructing traditional ink painting elements, the ink's diffusion range is mapped to the diffusion radius parameter of a fluid system, the ink's density is mapped to the grayscale value and transparency parameters of a light source spectrum, and the direction of brushstrokes is mapped to the offset trajectory parameters of scene mesh vertices. When deconstructing calligraphic brushstrokes, the thickness of the strokes is mapped to the cross-sectional radius parameter of the line model, and the ink's gradient is mapped to the optical reflectance coefficient distribution parameter of the material. When deconstructing paper-cutting patterns, the pattern outline is mapped to the topological structure parameters of the model mesh, and the hollowed-out areas are mapped to the transmittance parameters of the light and shadow system. Through this deconstruction process, abstract Chinese cultural symbols are transformed into quantifiable parameters with clear physical meaning, enabling the physics engine to dynamically generate and render cultural elements by calling these parameters.

[0121] The output psychological state trend, such as calming down, rising emotional fluctuations, or stable emotions, directly determines the need for introducing cultural elements. For example, when the predicted emotional state is calming down, introducing soothing cultural elements, such as ink wash landscapes, can enhance the sense of calm. When the predicted emotional fluctuations are large, complex cultural elements are temporarily not introduced to avoid interfering with emotional regulation. The setting of the activation threshold is based on this need. Its essence is to define the psychological state conditions for the activation and rendering of cultural elements: if the trend result is calming down, a lower activation threshold is set, that is, when the real-time emotional energy value approaches the baseline threshold for a calm state, the symbolic field can be activated, allowing the cultural elements to gradually appear; if the trend result is rising emotional fluctuations, a higher activation threshold is set, and the symbolic field is only allowed to activate when the emotional energy value falls back to a relatively stable range, avoiding cultural elements from distracting the user's attention during periods of intense emotional fluctuation. At the same time, it is necessary to establish a logical correlation between the activation threshold and the intensity of cultural elements to avoid abrupt experiences caused by inappropriate intensity when cultural elements are activated: when the activation threshold is low and the corresponding emotion tends to be calm, the intensity of cultural elements should be set to be low, such as slowing down the ink diffusion speed and increasing the transparency of ink, so that the elements can be integrated into the scene in a soft way; when the activation threshold is high and the corresponding emotion changes from fluctuation to stability, the intensity of cultural elements should be set to be medium, such as speeding up the color development speed of calligraphy strokes and increasing the clarity of outlines, so as to guide the emotion to be stable with clearer cultural symbols, but the intensity should always be controlled within the range of the core elements of the scene, such as light and shadow and fluids, to ensure the harmony of the overall atmosphere of the scene.

[0122] By adjusting the parameters of cultural elements in conjunction with the physical parameters of the scene, a deep integration of cultural symbols and virtual scenes can be achieved, ultimately enhancing the emotional resonance effect. If the parameters of cultural elements and the physical parameters of the scene are adjusted independently, it is easy to cause a disconnect between the cultural elements and the scene atmosphere. For example, warm-toned ink painting may appear in a cool-toned lighting scene, or slowly spreading ink painting may appear in a fast-flowing fluid scene, which will destroy the user's immersion. Therefore, it is necessary to establish a linkage adjustment mechanism for the two types of parameters based on the activation state of the symbol field, such as active, inactive, and activation intensity level. The specific adjustment logic should follow the principle of atmosphere coordination: when the symbol field is active and the cultural element is of the ink painting type, the physical parameters of the scene should be adjusted synchronously to adapt to the visual atmosphere of ink painting. For example, the color temperature parameter of the lighting system should be adjusted from the original cool tone, such as 3000K, to a bluish neutral tone, such as 4500K, so that the lighting and shadow are consistent with the tone of ink painting; the flow speed parameter of the fluid system should be reduced to match the spreading speed of ink painting, simulating the natural diffusion effect of ink painting in the fluid; and the particle density parameter of the particle system should be reduced to avoid particles obscuring ink painting details. When the activation intensity of the symbolic field increases, such as when ink wash transitions from fading to clarity, the adjustment range of related scene parameters is simultaneously increased. For example, the grayscale value of light and shadow is further increased to enhance the layering of the ink wash. When the activation intensity of the symbolic field decreases, such as when ink wash gradually fades away, the scene parameters are simultaneously reverted to their initial state. This coordinated adjustment ensures that cultural elements do not exist in isolation within the scene, but rather form a unified visual atmosphere with core scene elements such as light and shadow, fluids, and particle systems. When users perceive cultural elements, they experience a dual psychological identification due to the familiar cultural symbols and the comfortable scene atmosphere, thereby strengthening the emotional resonance effect. For example, the sense of tranquility evoked by traditional ink wash elements, combined with the calm atmosphere of the scene, can more effectively solidify the user's calm emotions and assist in achieving the goals of psychotherapy.

[0123] A psychotherapy system integrating virtual reality scenarios, applied to the aforementioned psychotherapy system integrating virtual reality scenarios, comprising:

[0124] The emotion mapping module is used to collect physiological signals reflecting the user's psychological state through multimodal biosensors, convert the physiological signals into quantified emotion energy values, and generate a parameter sequence to drive the virtual scene based on the emotion energy values. The parameter sequence contains parameters that affect the visual and physical properties of the scene.

[0125] The scene control module is used to control the physics engine in the virtual reality scene in real time using parameter sequences, and dynamically adjust the scene's lighting, color, geometry and particle system motion state to build a dynamic virtual environment that is linked to the user's emotions.

[0126] The motion feedback module is used to construct a virtual stress field in a dynamic virtual environment. It captures the user's body movements through sensors and calculates the perturbation effect on the virtual stress field. The perturbation effect is then input as a feedback quantity into the emotion mapping module to adjust the generation of parameter sequences.

[0127] The emotion prediction module is used to perform time-series analysis on parameter sequences to predict the trend of user emotion changes in the future within a preset time period. Based on the predicted emotion change trend, the parameters of the virtual scene are pre-adjusted before the corresponding emotion change occurs.

[0128] The symbol fusion module is used to encode preset cultural symbol elements into a parameter set that can be called by the physics engine. Based on the predicted emotional change trend, it controls the timing of the fusion of the parameter set in the virtual scene and at least one visual performance intensity parameter.

[0129] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and its improved concepts, should be covered within the scope of protection of the present invention.

Claims

1. A method for psychotherapy that integrates virtual reality scenarios, characterized in that, Includes the following steps: Step S1: Collect physiological signals reflecting the user's psychological state through multimodal biosensors, convert the physiological signals into quantified emotional energy values, and generate a parameter sequence to drive the virtual scene based on the emotional energy values. The parameter sequence includes parameters that affect the visual and physical properties of the scene. Step S2: Use parameter sequences to control the physics engine in the virtual reality scene in real time, and dynamically adjust the scene's lighting, color, geometry and particle system motion state to build a dynamic virtual environment that is linked to the user's emotions. Step S3: Construct a virtual pressure field in a dynamic virtual environment, capture user limb movements through sensors and calculate the perturbation effect on the virtual pressure field, and input the perturbation effect as a feedback quantity into step S1 to adjust the generation of parameter sequence. Step S4: Perform time series analysis on the parameter sequence to predict the user's emotional change trend in the future preset time period. Based on the predicted emotional change trend, pre-adjust the parameters of the virtual scene before the corresponding emotional change occurs. Step S5: Encode the preset cultural symbol elements into a parameter set that can be called by the physics engine, and control the timing of the fusion of the parameter set in the virtual scene and at least one visual performance intensity parameter according to the predicted emotional change trend.

2. The method according to claim 1, characterized in that, Step S1 specifically includes: Step S11: Filter and remove artifacts from the raw physiological signals from the multi-channel biosensor to extract the effective signal components related to emotional fluctuations. Step S12: Based on the preset psychophysiological correspondence model, the effective signal components are calculated as emotional energy values ​​that characterize the current psychological arousal level. Step S13: Based on the different numerical ranges of the emotional energy value, call different nonlinear mapping functions to generate a parameter sequence, where the parameters include color temperature parameters, spatial curvature parameters, and particle motion entropy parameters.

3. The method according to claim 2, characterized in that, Step S12 also includes a model personalization step: Step S121: In the initial stage, collect the user's physiological baseline data in a calm state and labeled data under specific emotions to establish an individualized benchmark and sample library. Step S122: During the application process, the parameters and weights of the corresponding psychophysiological model are dynamically optimized by continuously combining the user's real-time physiological data and subjective feedback through machine learning algorithms.

4. The method according to claim 2, characterized in that, Step S2 specifically includes: Step S21: Decouple the parameter sequence into independent control commands that control the light source, mesh model and particle emitter respectively; In step S22, the physics engine executes independent control commands to synchronously adjust lighting effects, object deformation, and particle motion, and establishes physical linkages between these elements. Step S23 also maps the emotional energy value to physical field parameters and inputs the physical field parameters into step S22 to globally influence the physical behavior of scene elements.

5. The method according to claim 4, characterized in that, Step S22 further includes a load optimization step: Step S221: Based on the priority of the currently activated independent control command, dynamically adjust the calculation weight of the physical linkage relationship between different scene elements. Step S222: Monitor the performance load of the rendering system in real time. When the load exceeds the preset threshold, automatically reduce the calculation accuracy and / or parameter adjustment frequency of the secondary physical linkage relationship to ensure that the rendering frame rate is not lower than the preset threshold.

6. The method according to claim 4, characterized in that, Step S3 specifically includes: Step S31: Based on the real-time geometric structure and material properties of the dynamic virtual environment, define the range of action and the initial pressure reference value of the virtual pressure field. Step S32: Analyze the speed, direction and amplitude characteristics of the user's actions, calculate the impulse generated by the virtual pressure field, and quantify the impulse into pressure gradient change value; Step S33: Based on the pressure gradient change value and the current scene state, calculate the feedback amount used to correct the parameter sequence, thereby closing the real-time interaction loop from action to scene.

7. The method according to claim 6, characterized in that, Step S32 further includes an action validity judgment step: Step S321: Filter out invalid or minor actions using the spatiotemporal feature matrix of the actions, and extract only valid action feature data that meets the preset conditions; Step S322: When calculating the impulse, a weighting coefficient based on the spatiotemporal context is introduced to more accurately evaluate the actual perturbation effect of the action on the virtual field.

8. The method according to claim 1 or 2, characterized in that, Step S4 specifically includes: Step S41: Perform multi-scale analysis based on frequency domain transformation on the parameter sequence to extract short-term and long-term fluctuation components that characterize the user's emotional cycle. Step S42: Using a probabilistic prediction model trained based on historical data, the extracted short-term and long-term fluctuation components are extrapolated, and a probabilistic prediction result of the trend of emotional energy value change within a preset period is output. Step S43: Based on the probabilistic prediction results, preload and prepare the corresponding scene compensation parameter package, and automatically activate the compensation effect when the real-time emotional energy value reaches the preset threshold and is within the predicted emotional energy peak window period.

9. The method according to claim 8, characterized in that, Step S5 specifically includes: Step S51: Deconstruct the preset Chinese cultural elements into parameters that can be called by the physics engine. The parameters include fluid dynamics parameters, motion trajectory parameters, and geometric structure parameters. Step S52: Link the probabilistic prediction results with the symbolic field activation logic, and set the threshold conditions for the emotional energy value and the rules for the regulation of visual expression of different cultural elements in their manifestation and evolution. Step S53: According to the activation logic and visual performance control rules set in step S52, during the dynamic scene reconstruction process in step S2, the deconstructed cultural parameters are called and integrated. The visual performance control rules are reflected by adjusting at least one parameter among the transparency, brightness, saturation or size of the cultural elements.

10. A psychotherapy system integrating virtual reality scenarios, applied to the method according to any one of claims 1-9, characterized in that, include: The emotion mapping module is used to collect physiological signals reflecting the user's psychological state through multimodal biosensors, convert the physiological signals into quantified emotion energy values, and generate a parameter sequence to drive the virtual scene based on the emotion energy values. The parameter sequence contains parameters that affect the visual and physical properties of the scene. The scene control module is used to control the physics engine in the virtual reality scene in real time using parameter sequences, and dynamically adjust the scene's lighting, color, geometry and particle system motion state to build a dynamic virtual environment that is linked to the user's emotions. The motion feedback module is used to construct a virtual stress field in a dynamic virtual environment. It captures the user's body movements through sensors and calculates the perturbation effect on the virtual stress field. The perturbation effect is then input as a feedback quantity into the emotion mapping module to adjust the generation of parameter sequences. The emotion prediction module is used to perform time-series analysis on parameter sequences to predict the trend of user emotion changes in the future within a preset time period. Based on the predicted emotion change trend, the parameters of the virtual scene are pre-adjusted before the corresponding emotion change occurs. The symbol fusion module is used to encode preset cultural symbol elements into a parameter set that can be called by the physics engine. Based on the predicted emotional change trend, it controls the timing of the fusion of the parameter set in the virtual scene and at least one visual performance intensity parameter.

Images