An Emotion Regulation Clothing System and Control Method for Students

Abstract

This invention discloses an emotion regulation clothing system and its control method for students. The system integrates a multimodal physiological signal acquisition module, a multi-dimensional tactile feedback module, an intelligent control module, and an intelligent cognitive prompting module, using students' everyday clothing as a carrier. It achieves high-accuracy anxiety state classification by combining a personally calibrated hierarchical multimodal fusion algorithm with the recognition of nervous small movements. The system adaptively matches and dynamically adjusts the tactile feedback strategy based on anxiety level and usage scenario. Simultaneously, it provides cognitive behavioral guidance through an electronic ink cognitive prompting area and an erasable recording area. This invention realizes a complete closed loop of emotion management—monitoring, assessment, feedback, and regulation—and can be used covertly in all scenarios such as classrooms and examination rooms. It quickly alleviates immediate physiological symptoms of anxiety and intervenes at the cognitive root, helping students build long-term emotion management skills.

Description

Technical Field

[0001] This invention relates to the field of smart clothing and human-computer interaction technology, specifically to an emotion regulation clothing system and its control method for students. Background Technology

[0002] Currently, students are at a critical stage of physical and mental development, especially junior high school, high school, and university students aged 13-22. They face multiple challenges, including academic competition, pressure to enter higher education, social adaptation, and self-identity issues. Anxiety has become the most common mental health problem in this group. Prolonged anxiety not only leads to academic problems such as poor concentration, memory loss, and reduced learning efficiency, but also triggers physical reactions such as palpitations, chest tightness, tremors, and sweating, seriously affecting their physical and mental health and personality development.

[0003] Currently, emotion regulation solutions for students mainly fall into the following categories, but all of them have obvious limitations: Drug intervention: It mainly regulates neurotransmitter levels through anti-anxiety drugs, but it has side effects such as drowsiness, dizziness, and gastrointestinal discomfort. It has very low acceptance among students and parents and is only suitable for clinical treatment of patients with severe anxiety.

[0004] Psychological counseling: Cognitive behavioral therapy is conducted under the guidance of professional psychological counselors, but it has problems such as high cost, uneven distribution of resources, and long appointment cycle.

[0005] Mobile app tools: These tools help users regulate their emotions through functions such as mindfulness meditation, guided breathing, and emotional diaries. However, users need to actively open and use them, and they cannot provide immediate intervention when emotions erupt. Furthermore, they are completely unusable in high-frequency anxiety scenarios for students, such as classrooms, exam rooms, and lectures.

[0006] Existing smart clothing focuses primarily on monitoring sports and health, but lacks targeted design for specific student scenarios.

[0007] Therefore, there is an urgent need for a drug-free emotion regulation solution specifically designed for students, which can be used discreetly in all scenarios such as classrooms and examination rooms, and integrates high-accuracy emotion recognition, personalized tactile adjustment and cognitive behavior guidance, so as to achieve immediate relief and long-term improvement of students' anxiety. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention discloses an emotion regulation clothing system and its control method for students, in order to solve the problems mentioned in the background art.

[0009] To achieve the above objectives, the present invention provides the following technical solution: an emotion regulation clothing system for students, comprising a clothing body, wherein the clothing body is everyday clothing worn by students, and all electronic modules adopt a concealed integrated design; The multimodal physiological signal acquisition module includes a heart rate sensor, a skin conductance sensor, a temperature sensor, and a triaxial accelerometer, which are respectively positioned on the garment body at the corresponding locations on the neck, armpit, chest, and wrist. The multi-dimensional tactile feedback module includes a micro vibration motor array, an inflatable airbag array, shape memory alloy wires, and a heating unit, which are respectively set on the garment body at the corresponding positions of the human back, abdomen, neck, and arms; The intelligent control module is electrically connected to the multimodal physiological signal acquisition module and the multidimensional tactile feedback module, and has a built-in personal calibration hierarchical multimodal fusion algorithm, adaptive tactile feedback strategy algorithm and lightweight convolutional neural network model. The intelligent cognitive prompt module includes an electronic ink cognitive prompt area and a pressure-detectable rewritable recording area, which is set on the chest or cuff of the garment. The wireless communication module is electrically connected to the intelligent control module and is used to interact with external intelligent devices to record emotion data, analyze trends, and update cognitive training content.

[0010] This invention also provides a control method for an emotion regulation clothing system for students, comprising the following steps: S1: Through the multimodal physiological signal acquisition module on the clothing worn by the student, multiple physiological signals and limb movement signals of the wearer are collected simultaneously. S2: Adaptive preprocessing and multi-scale feature extraction are performed on the collected multi-source signals to construct an emotion feature vector containing time domain, frequency domain, and time-frequency domain features; S3: Based on the emotional feature vector, the wearer's dynamic anxiety index is calculated using a personally calibrated hierarchical multimodal fusion algorithm, and the emotional state is graded and judged in combination with a preset multi-level anxiety threshold. S4: When it is determined that the wearer is in an anxious state, the tactile feedback strategy is adaptively matched and dynamically adjusted according to the anxiety level, current scene mode and historical adjustment effect data to provide the wearer with multi-dimensional coordinated tactile stimulation. S5: Synchronously initiate the cognitive behavior guidance process, intelligently push corresponding cognitive reconstruction information according to the type of anxiety, and complete the closed-loop management of emotions through physical interaction.

[0011] Preferably, the specific steps for adaptive preprocessing and multi-scale feature extraction of multi-source signals in step S2 include: S21: Perform wavelet threshold denoising and baseline drift correction on the original physiological signal to remove motion artifacts and power frequency interference. The calculation formula for wavelet denoising is as follows: , In the formula, The signal value after denoising. For the first j Scale No. k Wavelet coefficients of position, For the first j Adaptive threshold for scale For wavelet basis functions, For indicator functions, satisfying =1, =0, J The maximum decomposition scale; S22: The denoised signal is segmented using a sliding time window, with both the window length and the sliding step size being preset range values; S23: Extract multi-scale features from each signal segment, including mean, standard deviation, root mean square and peak factor in the time domain, power spectral density and power spectral entropy of the preset frequency band in the frequency domain, and wavelet packet energy entropy in the time and frequency domain. S24: Standardize the extracted features to construct a D-dimensional emotion feature vector. .

[0012] Preferably, the specific steps in step S3 for calculating the dynamic anxiety index using a personally calibrated hierarchical multimodal fusion algorithm include: S31: Perform preliminary emotion probability prediction for each single-modal feature subset to obtain the anxiety probability of the m-th modality. The calculation formula is: , In the formula, It is the Sigmoid activation function. For the first m A feature subset of each modality For the first m The weight vector of each modality For the first m Bias terms for each modality; S32: Introduction of Personal Calibration Factors α The single-mode probability is corrected, and the corrected probability is: , in This represents the individual's baseline anxiety probability for this modality. For the first m The calibration coefficients for each mode, and 0 ≤ ≤1; where the individual's baseline anxiety probability With calibration coefficient The initial values ​​are all derived from the personal calibration initialization process during the first use of the system; calibration coefficients The system will automatically update every 7 days based on users' historical adjustment performance data. The update logic is as follows: if the anxiety prediction accuracy for a certain modality is higher than 85% for 7 consecutive days, then... Increase by 0.05 (maximum 1.0); if below 60%, then Reduce by 0.05 (lower limit 0.5).

[0013] S33: The modified single-modal probabilities are fused at the decision level using DS evidence theory to obtain the final dynamic anxiety index A, calculated as follows: , In the formula, For the first m Global credibility weights for each modality, and satisfying , M The total number of modes; S34: Anxiety Index A Compared with the preset three-level anxiety threshold, it is divided into four levels: normal state, mild anxiety, moderate anxiety and severe anxiety.

[0014] Preferably, the specific steps of adaptive matching and dynamic adjustment of the haptic feedback strategy in step S4 include: S41: Based on the current anxiety level and the preset scenario mode, match an initial adjustment plan from the haptic feedback strategy library. The scenario modes include exam mode, classroom mode, home mode, and exercise mode. S42: During the adjustment process, physiological feedback signals from the wearer are collected in real time, and adjustment effect evaluation indicators are calculated. E The calculation formula is: , In the formula, To adjust the initial anxiety level, To adjust to t Anxiety level at any moment; S43: When evaluating the adjustment effect index E When the haptic feedback intensity, frequency, and duration are less than the preset effect threshold, the adjustment formula is as follows: , In the formula, for t The intensity of tactile stimulation at any given moment. To preset the effect threshold, k This is an adjustment coefficient, and 0 <k <1; The adjustment range of tactile stimulation intensity is strictly limited to the comfortable intensity range determined in the personal calibration initialization process to avoid ineffective stimulation that is too strong or too weak; When multiple modes are activated at the same time, parameter adjustment follows the priority order specified in this strategy, and the sum of the intensities of all modes does not exceed the upper limit of the total comfortable intensity of personal calibration.

[0015] S44: When evaluating the adjustment effect index E The haptic feedback will automatically stop after the effect threshold is greater than or equal to the preset effect threshold and the preset duration is reached.

[0016] Preferably, the multi-dimensional collaborative tactile feedback mentioned in step S4 includes any combination of soothing mode, breathing guidance mode, wrapping mode, and warmth mode, wherein the specific control steps of the soothing mode include: S441: Constructing the spatial coordinate matrix of the back tactile feedback unit: , in N This represents the total number of haptic feedback units. S442: Generates an asymmetric stroking path, with the path direction from the top left to the bottom right. The parametric equation of the stroking path is: , In the formula, ( , (This refers to the coordinates of the starting point.) and They are respectively x shaft and y The stroking rate in the axial direction, and ; S443: Calculate the activation intensity of each unit based on the distance between the stroking path and each tactile feedback unit. The activation intensity formula is: , In the formula, For the first i Each haptic feedback unit in t Activation intensity at any given moment s This is the Gaussian kernel width parameter.

[0017] Preferably, the specific control steps for the breathing guidance mode in step S4 include: S451: Call the personal resting respiratory rate collected during the personal calibration initialization process. Calculate an individual's resting respiratory cycle If a user does not recalibrate for 7 consecutive days, the system will automatically update once during the daily rest period. This ensures the accuracy of breathing guidance.

[0018] S452: Adaptively adjusts the breathing guidance cycle based on the individual's resting respiratory rate. The calculation formula is: , In the formula, For an individual's resting respiratory cycle, c The leading coefficient is 0 < c <0.5, A Current anxiety index; S453: According to the preset ratio of inhalation: breath-holding: exhalation time, the guidance cycle is divided into three stages, and the intensity of stimulation of the abdominal tactile feedback unit is controlled respectively: the intensity increases linearly during the inhalation stage, remains unchanged during the breath-holding stage, and decreases linearly during the exhalation stage. S454: During the guidance process, the wearer's actual breathing rate is detected in real time. When the deviation between the actual breathing rate and the guidance rate exceeds a preset threshold, the guidance rate is automatically fine-tuned to achieve synchronization.

[0019] Preferably, it also includes a deep learning-based step for recognizing and automatically triggering subtle nervous movements, specifically including: S0: Real-time acquisition of the wearer's limb movement signals via a triaxial accelerometer, with a sampling frequency within a preset range; S01: Perform frame segmentation processing on the motion signal, extract the time domain features, frequency domain features and energy features of each frame signal, and construct the motion feature vector; S02: Input the action feature vector into a pre-trained lightweight convolutional neural network model to identify tension-related small actions, including but not limited to leg shaking, hand picking, restlessness, and nail biting, and output the confidence score of the corresponding action. S03: When the confidence level of a certain type of anxiety-inducing small movement is greater than a preset confidence threshold, and the duration is greater than a preset duration threshold, a joint judgment is made by simultaneously combining the current multimodal physiological signals. Anxiety trigger is determined only if any of the following conditions are met: 1) Current skin conductance signal is significantly higher than the individual's baseline value by ≥15%; 2) Current heart rate or heart rate variability (HRV) deviates from the resting baseline by ≥10%; 3) Dynamic Anxiety Index A ≥ 0.3 (threshold for mild anxiety).

[0020] After the judgment is passed, the matching tactile feedback mode is automatically activated based on the anxiety characteristics corresponding to the action type; the output layer of the lightweight convolutional neural network uses the Softmax activation function. c The formula for calculating the confidence level of a class of actions is: , In the formula, For the firstc The network output value of the action type, C This represents the total number of action types; the personal fine-tuning parameters of the lightweight convolutional neural network model come from the personal calibration initialization process; the system will continuously record false positives and false negatives in action recognition, and automatically perform incremental fine-tuning of the model every 15 days.

[0021] Preferably, the specific steps of the cognitive behavior guidance process in step S5 include: S51: Based on the anxiety type and anxiety level determined in step S3, retrieve the cognitive distortion type with the highest matching degree and the corresponding rational rebuttal statement from the cognitive reconstruction information database; S52: Control the electronic ink cognitive prompt area on the garment body, first display the cognitive distortion type indicator, and in response to the wearer's flipping or touch operation, switch to display the corresponding rational rebuttal statement; S53: Detects the wearer's writing and erasing actions in a rewritable recording area using a pressure sensor, and calculates an indicator of emotional release level. R The calculation formula is: , In the formula, S The percentage of the erased area relative to the total recorded area. v The average speed of the erasing motion, t For the duration of the erasure action, These are the weighting coefficients, and ; S54: Indicator of Emotional Release Level R Correlation analysis with changes in physiological signals is performed to update the personal cognitive regulation effect model for subsequent intelligent recommendation of cognitive information.

[0022] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention achieves highly accurate emotion recognition with low false alarms. It adopts a personally calibrated hierarchical multimodal fusion algorithm, which simultaneously integrates heart rate, skin conductance, body temperature and limb movement signals. It introduces individual baseline correction and DS evidence theory decision-making, and combines lightweight CNN for nervous small movements recognition, which improves the accuracy of anxiety recognition to over 90% and reduces the false alarm rate to below 5%, accurately capturing students' instantaneous anxiety state.

[0023] 2. This invention enables personalized and scenario-based real-time adjustment, integrating four tactile feedback methods: vibration, airbags, heating, and shape memory alloy, to simulate a natural soothing touch; it supports automatic matching of feedback combinations based on anxiety level and real-time dynamic adjustment of stimulation parameters; it presets three scenario modes: exam, classroom, and home, to achieve completely discreet and non-intrusive intervention and quickly relieve somatic symptoms of anxiety.

[0024] 3. This invention achieves long-term management through a dual closed loop of physiological and cognitive processes: For the first time, a complete closed loop of monitoring, evaluation, feedback, and regulation is constructed in smart clothing. Emotional states are accurately identified through multimodal signal monitoring and dynamic anxiety assessment. Immediate intervention is achieved through multi-dimensional tactile feedback and cognitive prompts. The regulation strategy is then dynamically adjusted based on physiological feedback and interactive data. Combined with long-term data tracking and cognitive training via an APP, the invention achieves a unity of immediate relief and long-term improvement in emotional capabilities. Attached Figure Description

[0025] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0026] In the attached diagram: Figure 1 This is a simplified structural framework diagram of an emotion regulation clothing system for students according to the present invention; Figure 2 This is a flowchart of a control method for an emotion regulation clothing system for students, as described in this invention. Detailed Implementation

[0027] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0028] Example 1: This example provides an emotion regulation clothing system for students, the overall structure of which is as follows. Figure 1 As shown, the system uses a close-fitting vest-style garment as its carrier, integrating a multimodal physiological signal acquisition module, a multi-dimensional tactile feedback module, an intelligent control module, an intelligent cognitive prompting module, and a wireless communication module. All electronic modules are encapsulated in an ultra-thin, flexible package with a thickness of no more than 3mm, hidden within the garment's layers. The garment's appearance is completely identical to a regular student sports vest, without any protrusions or markings, providing complete concealment.

[0029] The garment itself is made of a cotton-spandex blend fabric, which is elastic and breathable, making it suitable for students' daily wear. The garment features multiple detachable modular pockets, each corresponding to the installation location of a different electronic module. The pockets use concealed zippers; simply unzip to remove all electronic modules for easy garment cleaning and module upgrades / replacements.

[0030] Multimodal physiological signal acquisition module: The multimodal physiological signal acquisition module includes the following sensors, all of which are electrically connected to the intelligent control module via flexible conductive fabric: 1) Photoelectric heart rate sensor: placed on the left side of the chest of the garment corresponding to the human heart, with a sampling frequency of 100-500Hz, used to collect heart rate and heart rate variability (HRV) signals; 2) Skin conductance sensor: Located under the left and right armpits of the garment, using silver chloride electrodes, with a sampling frequency of 10-100Hz, used to collect changes in skin conductivity; 3) Triaxial accelerometer: Located at the center of the back waist of the garment, with a sampling frequency of 50-200Hz, it is used to collect limb movement acceleration signals and identify small, tense movements; 4) Temperature sensor: Located at the back of the neck of the garment, with a sampling frequency of 1-10Hz, used to collect changes in body surface temperature.

[0031] Multi-dimensional haptic feedback module: The multi-dimensional haptic feedback module includes the following units, all of which are independently controlled by the intelligent control module: 1) Back vibration motor array: It consists of 16 miniature flat vibration motors with a diameter of 5mm and a thickness of 2mm, arranged in a 4×4 matrix on both sides of the spine on the back of the garment body. The distance between adjacent motors is 3-5cm, which is used to achieve the gentle touch of the soothing mode. 2) Abdominal inflatable airbag: Made of ultra-thin TPU material, it is placed in the center of the abdomen of the garment, with a maximum inflation thickness of no more than 1cm and an inflation pressure range of 0.5-5kPa, used to achieve breathing guidance mode and wrapping mode; 3) Neck heating unit: A graphene heating film is used and placed at the back of the neck of the garment. The heating temperature range is 36-38℃ and the temperature control accuracy is ±0.5℃. It is used to achieve a warm heating mode. 4) Wrist shape memory alloy wire: Located inside the cuff of the garment, it consists of multiple nickel-titanium alloy wires with a diameter of 0.1mm. When energized, it can shrink and deform to provide gentle wrapping pressure.

[0032] Intelligent control module: The intelligent control module is located in a detachable pocket on the left side of the garment's waist, measuring 5cm × 3cm × 0.8cm, and includes: 1) Microprocessor: A low-power microcontroller with an ARM Cortex-M4F core, a main frequency of 48-168MHz, and 512KB Flash and 128KB RAM, used to run all control algorithms and data processing programs; 2) Power unit: It adopts a 3.7V, 500mAh rechargeable lithium polymer battery, supports USB-C interface charging, and can be used continuously for 7-14 days on a single charge; 3) Power management chip: responsible for battery charging and discharging management and power distribution to each module, supporting multiple low-power modes; 4) Storage unit: 8MB of SPIFlash is used to store personal baseline data, emotional history records and algorithm parameters.

[0033] Intelligent cognitive prompting module: The intelligent cognitive prompting module is located on the left side of the chest of the garment and includes: 1) Electronic Ink Cognitive Prompt Area: It adopts a 1.54-inch black and white electronic ink screen with a resolution of 200×200. It supports double-sided display. The first side displays the cognitive distortion type indicator, and the second side displays the corresponding rational rebuttal statement. The display content can be switched by flipping the screen. 2) Rewritable and erasable recording area with pressure detection: Made of erasable silicone material, with an area of ​​5cm×5cm, and an 8×8 pressure sensor array integrated below, used to detect the position, force and speed of writing and erasing actions.

[0034] Wireless communication module: The wireless communication module adopts Bluetooth 5.0 low power module, which is integrated into the intelligent control module. It is used to conduct wireless data interaction with external intelligent devices such as smartphones and tablets. The transmission rate is 1Mbps and the transmission distance does not exceed 10m.

[0035] The overall system operation process is as follows: After students don the clothing, the system first initializes through personal calibration, establishing individual baselines such as heart rate, skin conductance, respiration, and tactile tolerance. During operation, the multimodal physiological signal acquisition module collects heart rate, skin conductance, temperature, and limb movement data in real time. After preprocessing and feature extraction by the intelligent control module, an anxiety index is calculated and graded using a hierarchical multimodal fusion algorithm. Simultaneously, three-axis acceleration signals and a lightweight CNN are used to identify tense small movements, and physiological signals are combined to determine whether an anxiety state is present. Once anxiety is determined, the multidimensional tactile feedback module activates breathing guidance, warmth, soothing, and wrapping modes according to the scenario and anxiety level, operating collaboratively in a priority manner without conflict or interference. Simultaneously, the intelligent cognitive prompting module pushes cognitive guidance information through the e-ink screen and erasable recording area. The wireless communication module uploads data to smart devices for recording, analysis, and long-term management. The entire system forms a complete closed loop of emotion management: acquisition, identification, judgment, adjustment, cognition, and feedback.

[0036] Example 2: Based on the above system, this example provides an emotion regulation and control method for students, the complete process of which is as follows: Figure 2 As shown, the specific steps are as follows: Step S1: Synchronous acquisition of multimodal signals The system performs personal calibration initialization upon first use, which includes: 1) Preparation and environmental verification: After wearing the clothing to confirm that the module is connected normally, wait for 3 minutes to prepare. The system will automatically detect environmental interference and start data acquisition after the signal stabilizes.

[0037] 2) Physiological baseline acquisition: Heart rate, skin conductance, temperature, and acceleration signals were acquired during a 5-minute resting period. After noise reduction, individual baseline characteristic values ​​and baseline anxiety probabilities for each modality were calculated. .

[0038] 3) Respiratory baseline calibration: Extracting an individual's resting respiratory rate through heart rate variability. Verify respiratory guidance response capability and initialize guidance coefficients. c .

[0039] 4) Small motion feature calibration: Guide the simulation of four typical stressful small motions, make personal fine-tuning to the lightweight CNN model, and initialize the action confidence threshold.

[0040] 5) Tactile tolerance calibration: Test four types of tactile feedback step by step from the lowest intensity and record the upper and lower limits of the student's comfort intensity.

[0041] 6) Parameter saving and verification: Save all calibration parameters and initialize personal calibration coefficients. With global weight The initial values ​​come from the calibration process. The calibration is updated adaptively every 7 days based on the adjustment effect; if the anxiety index A < 0.3 after 1 minute of rest verification, the calibration is successful.

[0042] After calibration, once the system is powered on, the multimodal physiological signal acquisition module synchronously acquires the wearer's heart rate, skin conductance, acceleration, and temperature signals at a preset sampling frequency, and transmits the raw digital signals to the intelligent control module. Simultaneously, the triaxial accelerometer continuously acquires the wearer's limb movement signals for real-time identification of subtle, tense movements.

[0043] Step S2: Adaptive preprocessing and multi-scale feature extraction The intelligent control module performs adaptive preprocessing and multi-scale feature extraction on the acquired raw signals, specifically including: 1. Wavelet Thresholding Denoising and Baseline Drift Correction: The original signal is decomposed into 3-5 levels using a db4 wavelet basis. A soft thresholding function is applied to the wavelet coefficients of each level for denoising, removing motion artifacts and 50Hz power frequency interference. The calculation formula for wavelet denoising is as follows: , In the formula, The signal value after denoising. For the first j Scale No. k Wavelet coefficients of position, For the first j Adaptive threshold for scale For wavelet basis functions, For indicator functions, satisfying =1, =0, J The maximum decomposition scale has a value range of 3-5.

[0044] 2. Signal segmentation: The denoised signal is segmented using a sliding time window with a window length of 1-10 seconds and a sliding step size of 0.1-1 seconds.

[0045] 3. Multi-scale feature extraction: Extract time-domain features (mean, standard deviation, root mean square, peak factor), frequency-domain features (power spectral density in the low-frequency band of 0.04-0.15Hz, power spectral density in the high-frequency band of 0.15-0.4Hz, power spectral entropy), and time-frequency-domain features (wavelet packet energy entropy) from each signal segment.

[0046] 4. Feature Standardization: All extracted features are Z-score standardized to construct a 24-dimensional emotion feature vector. .

[0047] Step S3: Personally calibrated hierarchical multimodal emotion assessment The intelligent control module calculates the wearer's dynamic anxiety index based on the extracted emotional feature vector using a personally calibrated hierarchical multimodal fusion algorithm, and performs emotional state classification judgment, specifically including: 1. Preliminary Single-Mode Prediction: The emotional feature vector is divided into four subsets according to mode: heart rate mode, skin conductance mode, acceleration mode, and temperature mode. These subsets are then input into a pre-trained logistic regression model to obtain the anxiety probability for each mode. The calculation formula is: In the formula, It is the Sigmoid activation function. For the first m A feature subset of each modality For the first m The weight vector of each modality For the first m Bias terms for each modality.

[0048] 2. Personal Calibration: Introducing a personal calibration factor The single-mode probability is corrected, and the corrected probability is: , in The individual baseline anxiety probability for this modality is determined by the system during 5-10 minutes of resting data collection when the wearer first uses the device. For the first m The calibration coefficients for each modality range from 0.5 to 1 and are adaptively updated by the system based on the wearer's historical adjustment effect data.

[0049] 3. Decision-level fusion: The DS evidence theory is used to perform decision-level fusion on the modified four unimodal probabilities to obtain the final dynamic anxiety index A, calculated as follows: , In the formula, For the first m Global credibility weights for each modality, and satisfying In this embodiment (Heart rate) (skin conductance) (acceleration) (temperature).

[0050] 4. Emotional State Grading: The anxiety index A is compared with a preset three-level anxiety threshold, and divided into four levels: 1) Normal state: A<0.3; corresponding to no obvious anxiety symptoms, physiological indicators are stable within the individual baseline range, and do not affect daily study and life; 2) Mild anxiety: 0.3≤A<0.6; This corresponds to mild tension and worry, and may include mild physical reactions such as increased heart rate and sweaty palms, but attention and learning ability are basically normal. 3) Moderate anxiety: 0.6≤A<0.8; corresponding to obvious anxiety, symptoms such as palpitations, chest tightness, and difficulty concentrating, and a significant decrease in learning efficiency; 4) Severe anxiety: A≥0.8; corresponds to intense anxiety or panic attacks, with severe physical and cognitive symptoms such as hand tremors, difficulty breathing, and confused thinking, making it impossible to carry out normal learning activities.

[0051] When the wearer's dynamic anxiety index A is determined to be less than 0.3, which is the normal state threshold and does not reach the preset mild anxiety trigger threshold, the system performs the following operations: Continuous Cyclic Monitoring: The system maintains the real-time operation of the multimodal physiological signal acquisition module, and executes the signal acquisition, preprocessing, feature extraction, and emotion judgment process in steps S1 to S3 cyclically according to the preset sampling frequencies (heart rate 100-500Hz, skin conductance 10-100Hz, acceleration 50-200Hz, and temperature 1-10Hz) and sliding window parameters (window length 1-10 seconds and sliding step size 0.1-1 seconds), achieving uninterrupted tracking of the wearer's emotional state. The characteristic values ​​of each modality of physiological signal acquired under normal conditions, including mean heart rate, skin conductance baseline, body surface temperature, and resting acceleration characteristics, along with the corresponding anxiety index A, are written in real-time to the local storage unit of the intelligent control module. This embodiment uses 8MB SPI Flash as the basic data for subsequent dynamic updates of the individual baseline, model calibration, and evaluation of the adjustment effect. External data synchronization: If the wireless communication module has established a connection with an external smart device, the system will synchronize the emotion record and parameter update log under normal conditions to the smartphone APP according to the preset synchronization cycle (every 30 minutes) to generate a long-term emotion trend report.

[0052] Step S4: Adaptive Multi-Dimensional Haptic Feedback Adjustment When the system determines that the wearer is in a state of anxiety, the intelligent control module adaptively matches and dynamically adjusts the tactile feedback strategy based on the current anxiety level, preset scene modes, and historical adjustment effect data. Specifically, this includes: 1. Initial Solution Matching: Based on the current anxiety level and scenario pattern, an initial adjustment solution is matched from the haptic feedback strategy library: 1) Exam mode: Only the breathing guidance mode with abdominal inflatable cuff is enabled, and the tactile intensity is set to the minimum value; 2) Classroom Mode: Turn off all tactile feedback and enable only cognitive prompts; 3) Home Mode: Activate the corresponding haptic feedback combination based on anxiety level: Mild anxiety: Breathing guidance mode + warming mode; Moderate anxiety: Soothing mode + Breathing guidance mode + Warm mode; Severe anxiety: Soothing mode + Breathing guidance mode + Wrap mode + Warm mode.

[0053] S41-1 Multi-mode Cooperative Control Strategy When the system matches multiple haptic feedback modes, it follows the principle of primary mode dominating, secondary mode accompanying, secondary mode superimposing, and reinforcement mode as a backup, clarifying priorities and fusion methods, and avoiding conflicts between perception and hardware.

[0054] (1) Mode priority (from high to low): Breathing guidance mode (primary mode) > Warming mode (secondary mode) > Soothing mode (secondary mode) > Wrapping mode (enhanced mode); Main mode: The core of all combinations, fully activated, responsible for regulating the autonomic nervous system; Auxiliary mode: Activates throughout the main mode, enhancing basic adjustment effects; Sub-mode: Moderate to severe anxiety is triggered by superposition, providing psychological comfort; Enhanced Mode: Temporarily activated only for severe anxiety; automatically deactivated when anxiety level drops below 0.6.

[0055] (2) Hardware conflict avoidance 1) The back vibration motor and the neck heating unit are physically separated and electrically isolated. The vibration frequency of 100-200Hz and the heating temperature of 36-38℃ are independently controlled, with no perceptible interference; 2) The abdominal airbag, wrist shape memory alloy wire, and other feedback units have no spatial overlap and are powered independently; 3) When multiple modes are activated simultaneously, priority is given to ensuring power supply for the main and auxiliary modes, and power is dynamically allocated to the secondary and enhanced modes.

[0056] (3) Perception fusion control 1) Time synchronization: Using the breathing guidance cycle as the global clock, the stroking path in soothing mode and the wrist contraction in wrapping mode are synchronized with the breathing phase; 2) Intensity ratio: 50% for main mode, 20% for auxiliary mode, 20% for secondary mode, and 10% for enhanced mode, with the total intensity not exceeding the personal calibrated comfort limit; 3) Smooth transition: Mode start / stop and intensity adjustment use a 1-2 second linear gradual change; 4) Dynamically adjust priority: main mode cycle / intensity → auxiliary mode temperature → secondary mode stroking rate / intensity → enhanced mode pressure; if the user does not want to reduce the intensity of the enhanced mode, prioritize reducing the intensity of the enhanced mode. 5) Typical combination synergy Mild anxiety (breathing guidance + heating): The breathing guidance cycle is adaptively adjusted, and the heating temperature is fixed at 37℃; Moderate anxiety (breathing guidance + warmth + soothing): Gently stroke the back in rhythm with breathing, and adjust the intensity dynamically according to the anxiety level; Severe anxiety (full mode): First activate breathing guidance + warmth. If there is no improvement after 30 seconds, add soothing. If there is no improvement after another 30 seconds, temporarily activate the package.

[0057] 2. Real-time evaluation of adjustment effect: During the adjustment process, the system recalculates the anxiety index every 10 seconds and calculates the adjustment effect evaluation indicators: , In the formula, To adjust the initial anxiety level, To adjust to t Anxiety index at all times.

[0058] 3. Dynamic parameter adjustment: When the adjustment effect evaluation index E When the haptic feedback intensity, frequency, and duration are less than the preset effect threshold (0.2 in this embodiment), the system gradually adjusts the intensity, frequency, and duration of the haptic feedback according to a preset step size. The adjustment formula is as follows: , In the formula, for t The intensity of tactile stimulation at any given moment. To preset the effect threshold, k This is for adjusting the coefficient.

[0059] 4. Automatic Stop: When the adjustment effect evaluation index... E When the value is greater than or equal to 0.5 and the duration exceeds 1 minute, the system automatically stops haptic feedback.

[0060] (I) Implementation of Soothing Mode: Soothing mode simulates the feeling of being gently stroked through a back vibration motor array. The specific control steps are as follows: 1) Construct the spatial coordinate matrix of the 16 vibration motors on the back: , The origin of the coordinate system is the upper left corner of the back, the x-axis is to the right, and the y-axis is downward.

[0061] 2) Generate an asymmetric caress path from top left to bottom right. The parametric equation of the path is: , In the formula, ( , (This refers to the coordinates of the starting point.) and They are respectively x shaft and y The stroking rate in the axial direction, and The total stroking speed ranges from 3 to 10 cm / s.

[0062] 3) Calculate the activation intensity of each motor at time t based on the distance between the stroking path and each vibration motor: , In the formula, For the first i Each haptic feedback unit in t Activation intensity at any given moment For maximum vibration intensity, s This is the Gaussian kernel width parameter, with a value range of 1-3cm.

[0063] (II) Specific Implementation of Breathing Guidance Mode: The breathing guidance mode guides the wearer to adjust their breathing through the rhythmic expansion and contraction of the abdominal inflatable bladder. The specific control steps are as follows: 1) The system collects 5 minutes of resting respiratory signals when the wearer uses it for the first time, and calculates the individual's resting respiratory rate. and resting breathing cycle .

[0064] 2) Adaptively adjust the breathing guidance cycle based on the current anxiety level: , in, c The leading coefficient is 0 < c <0.5.

[0065] 3) Based on a 4:2:6 ratio of inhalation: breath-holding: exhalation time, the induction cycle is divided into three phases: a. Inhalation phase: The airbag inflation pressure rises linearly to the preset value; b. Breath-holding phase: The airbag pressure remains constant; c. Exhalation phase: The airbag deflation pressure drops linearly to 0.

[0066] 4) During the guidance process, the system detects the wearer's actual breathing rate in real time through heart rate variability signals. When the deviation between the actual breathing rate and the guidance rate exceeds ±10%, the guidance rate is automatically fine-tuned to achieve synchronization.

[0067] Step S5: Automatic Triggering of Nervous Movements Based on Deep Learning The system simultaneously runs a stress-related micro-movement recognition algorithm based on a lightweight convolutional neural network to automatically trigger emotion regulation. The specific steps are as follows: 1. The triaxial accelerometer collects the wearer's limb movement signals in real time at a sampling frequency of 100Hz; 2. Perform frame segmentation on the motion signal, with a frame length of 1 second and a frame shift of 0.5 seconds. Extract the time domain features (mean, standard deviation, root mean square, zero crossing rate), frequency domain features (centroid of the spectrum, bandwidth of the spectrum), and energy features of each frame signal to construct a motion feature vector with a dimension of 12. 3. Input the action feature vectors into a pre-trained lightweight convolutional neural network model. This model contains two convolutional layers, two pooling layers, and one fully connected layer. The output layer uses the Softmax activation function to identify four types of tension-related small actions: leg shaking, hand picking, restlessness, and nail biting, and outputs the confidence score for each action. , In the formula, For the first c The network output value of the action type, C This represents the total number of action types.

[0068] 4. When the confidence level of a certain tense small movement is greater than 0.7 and the duration exceeds 30 seconds, the system automatically activates the matching tactile feedback mode.

[0069] Step S6: Cognitive-behavioral guidance While adjusting to tactile feedback, the system simultaneously initiates a cognitive behavioral guidance process, which includes: 1. Based on the determined anxiety type and anxiety level, retrieve the cognitive distortion type with the highest matching degree and the corresponding rational rebuttal statement from the cognitive restructuring information database; 2. Control the e-ink cognitive prompt area to first display a cognitive distortion type indicator (such as "catastrophizing"). When the wearer flips the cognitive prompt area, switch to displaying the corresponding rational rebuttal statement (such as "Things won't be as bad as you think, and most of your worries won't happen"). 3. Wearers can write emotional keywords or worrying events in the rewritable and erasable recording area. The pressure sensor array detects the writing and erasing actions in real time and calculates the degree of emotional release. R : , In the formula, S The percentage of the erased area relative to the total recorded area. v The average speed of the erasing motion, t For the duration of the erasure action, These are the weighting coefficients.

[0070] 4. The system will perform correlation analysis between the emotional release level index R and changes in physiological signals, update the personal cognitive regulation effect model, and use it for subsequent intelligent recommendation of cognitive information.

[0071] Step S7: External Interaction and Long-Term Management The system synchronizes all physiological data, emotional state records, and regulation effect data to a smartphone app via a wireless communication module. The app provides the following functions: 1) Real-time display of the wearer's physiological data such as heart rate, skin conductance, and body temperature, as well as their current emotional state; 2) Generate daily, weekly, and monthly mood trend charts, with special annotations for anxiety curves during important events such as exams and speeches; 3) Provide extended content for cognitive behavioral training, including daily challenges, emotion diaries, guided mindfulness meditation, etc.; 4) Allows users to customize parameters of the haptic feedback mode, such as stroking rate, breathing guidance ratio, heating temperature, etc. 5) Supports parental binding, allowing parents to view their child's overall emotional state, but not specific emotional diaries or cognitive content, thus protecting student privacy.

[0072] Example 3: Workflow examples in different scenario modes 1. Examination Mode: Before entering the examination room, students switch the system to examination mode via the app. During the examination, the system only activates the heart rate sensor and abdominal inflatable bladder, disabling all other sensors and haptic feedback units. When the system detects a student's heart rate increase exceeding 20% ​​of their baseline value for more than one minute, it automatically activates the breathing guidance mode. The bladder slowly inflates and deflates in a 4:2:6 ratio to guide the student's breathing. Throughout the process, the bladder's inflation pressure does not exceed 1 kPa, making it completely imperceptible externally and not affecting other examinees.

[0073] 2. Classroom Mode: When students are in class, the system switches to classroom mode. In this mode, all tactile feedback units are turned off, retaining only physiological signal monitoring and cognitive prompting functions. When the system detects that a student is in an anxious state, it displays the corresponding cognitive distortion type and rational rebuttal statement through the e-ink cognitive prompt area. Students can view the content by flipping the cognitive prompt area or write emotional keywords in the erasable recording area and release emotions by erasing.

[0074] 3. Home Mode: When students are studying or resting at home, the system switches to Home Mode. In this mode, the system activates its full functionality, including multimodal physiological signal monitoring, all tactile feedback modes, and cognitive guidance. When the system detects that a student is in an anxious state, it automatically activates the corresponding tactile feedback combination based on the anxiety level and simultaneously provides cognitive guidance to help the student quickly alleviate their emotions.

[0075] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A clothing system for mood regulation targeting students, characterized in that: Including the garment itself; The multimodal physiological signal acquisition module includes a heart rate sensor, a skin conductance sensor, a temperature sensor, and a triaxial accelerometer, which are respectively positioned on the garment body at the corresponding locations on the neck, armpit, chest, and wrist. The multi-dimensional tactile feedback module includes a micro vibration motor array, an inflatable airbag array, shape memory alloy wires, and a heating unit, which are respectively set on the garment body at the corresponding positions of the human back, abdomen, neck, and arms; The intelligent control module is electrically connected to the multimodal physiological signal acquisition module and the multidimensional tactile feedback module, and has a built-in personal calibration hierarchical multimodal fusion algorithm, adaptive tactile feedback strategy and lightweight convolutional neural network model. The intelligent cognitive prompt module includes an electronic ink cognitive prompt area and a pressure-detectable rewritable recording area, which is set on the chest or cuff of the garment. The wireless communication module is electrically connected to the intelligent control module and is used to interact with external intelligent devices to record emotion data, analyze trends, and update cognitive training content.

2. A control method for an emotion regulation clothing system for students, characterized in that, Includes the following steps: S1: Through the multimodal physiological signal acquisition module on the clothing worn by the student, multiple physiological signals and limb movement signals of the wearer are collected simultaneously. S2: Adaptive preprocessing and multi-scale feature extraction are performed on the collected multi-source signals to construct an emotion feature vector containing time domain, frequency domain, and time-frequency domain features; S3: Based on the emotional feature vector, the wearer's dynamic anxiety index is calculated through a personally calibrated hierarchical multimodal fusion algorithm, and the emotional state is graded and judged in combination with the preset multi-level anxiety threshold. S4: When it is determined that the wearer is in an anxious state, the tactile feedback strategy is adaptively matched and dynamically adjusted according to the anxiety level, current scene mode and historical adjustment effect data to provide the wearer with multi-dimensional coordinated tactile stimulation. S5: Synchronously initiate the cognitive behavior guidance process, intelligently push corresponding cognitive reconstruction information according to the type of anxiety, and complete the closed-loop management of emotions through physical interaction.

3. The control method for an emotion regulation clothing system for students according to claim 2, characterized in that, The specific steps of adaptive preprocessing and multi-scale feature extraction of multi-source signals in step S2 include: S21: Perform wavelet threshold denoising and baseline drift correction on the original physiological signal to remove motion artifacts and power frequency interference. The calculation formula for wavelet denoising is as follows: ， In the formula, The signal value after denoising. For the first j Scale No. k Wavelet coefficients of position, For the first j Adaptive threshold for scale For wavelet basis functions, For indicator functions, satisfying =1, =0, J The maximum decomposition scale; S22: The denoised signal is segmented using a sliding time window, with both the window length and the sliding step size being preset range values; S23: Extract multi-scale features from each signal segment, including mean, standard deviation, root mean square and peak factor in the time domain, power spectral density and power spectral entropy of the preset frequency band in the frequency domain, and wavelet packet energy entropy in the time and frequency domain. S24: Standardize the extracted features to construct a D-dimensional emotion feature vector. .

4. The control method for an emotion regulation clothing system for students according to claim 2, characterized in that, The specific steps in step S3 for calculating the dynamic anxiety index using a personally calibrated hierarchical multimodal fusion algorithm include: S31: Perform preliminary emotion probability prediction for each single-modal feature subset to obtain the anxiety probability of the m-th modality. The calculation formula is: ， In the formula, It is the Sigmoid activation function. For the first m A feature subset of each modality For the first m The weight vector of each modality For the first m Bias terms for each modality; S32: Introduction of Personal Calibration Factors α The single-mode probability is corrected, and the corrected probability is: ， in This represents the individual's baseline anxiety probability for this modality. For the first m Calibration coefficients for each mode; S33: The modified single-modal probabilities are fused at the decision level using DS evidence theory to obtain the final dynamic anxiety index A, calculated as follows: ， In the formula, For the first m Global credibility weights for each modality M The total number of modes; S34: Anxiety Index A Compared with the preset three-level anxiety threshold, it is divided into four levels: normal state, mild anxiety, moderate anxiety and severe anxiety.

5. The control method for an emotion regulation clothing system for students according to claim 4, characterized in that, The specific steps in step S4 for adaptive matching and dynamic adjustment of the haptic feedback strategy include: S41: Based on the current anxiety level and the preset scenario mode, match the initial adjustment plan from the tactile feedback strategy library. The scenario modes include exam mode, classroom mode, home mode and exercise mode. S42: During the adjustment process, physiological feedback signals from the wearer are collected in real time, and adjustment effect evaluation indicators are calculated. E The calculation formula is: ， In the formula, To adjust the initial anxiety level, To adjust to t Anxiety level at any moment; S43: When evaluating the adjustment effect index E When the haptic feedback intensity, frequency, and duration are less than the preset effect threshold, the adjustment formula is as follows: ， In the formula, for t The intensity of tactile stimulation at any given moment. To preset the effect threshold, k For adjustment coefficients; S44: When evaluating the adjustment effect index E The haptic feedback will automatically stop after the effect threshold is greater than or equal to the preset effect threshold and the preset duration is reached.

6. The control method for an emotion regulation clothing system for student groups according to claim 2, characterized in that, The multi-dimensional collaborative tactile feedback in step S4 includes any combination of soothing mode, breathing guidance mode, wrapping mode, and warmth mode. The specific control steps for the soothing mode include: S441: Constructing the spatial coordinate matrix of the back tactile feedback unit: ， in N This represents the total number of haptic feedback units. S442: Generates an asymmetric stroking path, with the path direction from the top left to the bottom right. The parametric equation of the stroking path is: ， In the formula, ( , (This refers to the coordinates of the starting point for gently touching the ground.) and They are respectively x shaft and y The stroking rate in the axial direction, and ; S443: Calculate the activation intensity of each unit based on the distance between the stroking path and each tactile feedback unit. The activation intensity formula is: ， In the formula, For the first i Each haptic feedback unit in t Activation intensity at any given moment σ This is the Gaussian kernel width parameter.

7. The control method for an emotion regulation clothing system for student groups according to claim 2, characterized in that, The specific control steps for the breathing guidance mode in step S4 include: S451: Collect the wearer's respiratory signal at rest and calculate the individual's resting respiratory rate. ; S452: Adaptively adjusts the breathing guidance cycle based on the individual's resting respiratory rate. The calculation formula is: ， In the formula, For an individual's resting breathing cycle, γ For guiding coefficients, A Current anxiety index; S453: According to the preset ratio of inhalation: breath-holding: exhalation time, the guidance cycle is divided into three stages, and the intensity of stimulation of the abdominal tactile feedback unit is controlled respectively: the intensity increases linearly during the inhalation stage, remains unchanged during the breath-holding stage, and decreases linearly during the exhalation stage. S454: During the guidance process, the wearer's actual breathing rate is detected in real time. When the deviation between the actual breathing rate and the guidance rate exceeds a preset threshold, the guidance rate is automatically fine-tuned to achieve synchronization.

8. The control method for an emotion regulation clothing system for students according to claim 2, characterized in that, It also includes deep learning-based recognition and automatic triggering of subtle nervous movements, specifically including: S0: Real-time acquisition of the wearer's limb movement signals via a triaxial accelerometer, with a sampling frequency within a preset range; S01: Perform frame segmentation processing on the motion signal, extract the time domain features, frequency domain features, and energy features of each frame signal, and construct the motion feature vector; S02: Input the action feature vector into a pre-trained lightweight convolutional neural network model to identify tension-related small actions, including but not limited to leg shaking, hand picking, restlessness, and nail biting, and output the confidence score of the corresponding action. S03: When the confidence level of a certain type of tense small movement exceeds a preset confidence threshold and the duration exceeds a preset duration threshold, a matching tactile feedback mode is automatically activated based on the anxiety characteristics corresponding to that movement type; wherein, the output layer of the lightweight convolutional neural network uses the Softmax activation function, the first... c The formula for calculating the confidence level of a class of actions is: ， In the formula, For the first c The network output value of the action type, C This represents the total number of action types.

9. The control method for an emotion regulation clothing system for student groups according to claim 2, characterized in that, The specific steps in the cognitive behavior guidance process in step S5 include: S51: Based on the anxiety type and anxiety level determined in step S3, retrieve the cognitive distortion type with the highest matching degree and the corresponding rational rebuttal statement from the cognitive reconstruction information database; S52: Control the electronic ink cognitive prompt area on the garment body, first display the cognitive distortion type indicator, and in response to the wearer's flipping or touch operation, switch to display the corresponding rational rebuttal statement; S53: Detects the wearer's writing and erasing actions in a rewritable recording area using a pressure sensor, and calculates an indicator of emotional release level. R The calculation formula is: ， In the formula, S The percentage of the area to be erased relative to the total area of ​​the recorded area. v The average speed of the erasing motion, t The duration of the erasure action; S54: Indicator of Emotional Release Level R Correlation analysis with changes in physiological signals is performed to update the personal cognitive regulation effect model for subsequent intelligent recommendation of cognitive information.

Images