Emotion regulation system based on multi-dimensional perception and evaluation method thereof

Abstract

The invention discloses an emotion regulation system based on multi-dimensional perception and an evaluation method thereof. In the invention, through the emotion classification and intensity evaluation module using the support vector machine algorithm, the system not only can accurately divide the emotion of the user into different categories, but also can evaluate the intensity of the emotion, so that a more personalized emotion regulation scheme is provided for the user. The accurate emotion recognition enables the system to provide corresponding intervention measures according to the real-time emotion state of the user, such as music relaxation, breathing guidance or emotion catharsis games, so that the user is effectively helped to relieve negative emotion, and the emotion self-management ability is improved. The emotion trend analysis module predicts the emotion change trend by using an autoregressive moving average (ARIMA) model, so that the system can have perspectiveness, identify possible emotion fluctuation in advance and take preventive measures. The method can quickly respond when the emotion of the user is abnormal, and timely assistance is provided, so that the influence of emotion problems on the daily life and work of the user is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of emotion regulation, and specifically relates to an emotion regulation system based on multi-dimensional perception and its evaluation method. Background Art

[0002] In today's fast-paced social life, people are facing increasing pressure and challenges, which often lead to mood swings and psychological problems. Traditional mental health services mainly rely on artificial counseling and self-reporting, and these methods have certain limitations, such as poor real-time performance, strong subjectivity, and uneven resource allocation. Therefore, how to use modern technology to achieve real-time, objective, and personalized emotion monitoring and regulation has become an urgent problem to be solved. Most existing emotion monitoring systems rely on a single perception means, such as only identifying emotions through physiological signals or facial expressions.

[0003] However, such traditional emotion monitoring systems often cannot comprehensively capture the user's emotional state. In addition, these systems usually lack the ability to predict the trend of emotional changes and detect abnormal emotions, so there are lags and inaccuracies in providing emotion regulation suggestions. Summary of the Invention

[0004] The purpose of the present invention is to provide an emotion regulation system based on multi-dimensional perception and its evaluation method to solve the above-mentioned problems.

[0005] The technical solution adopted by the present invention is as follows: An emotion regulation system based on multi-dimensional perception, the system includes: a biosensor module, a visual recognition module, a speech recognition module, an environmental monitoring module, a data preprocessing module, an emotion recognition and analysis module, a regulation strategy generation module, a user management and feedback module, and an execution module;

[0006] A data access module, an emotion classification and intensity evaluation module, an emotion trend analysis module, and an abnormal emotion detection module are arranged inside the emotion recognition and analysis module;

[0007] The biosensor module, the visual recognition module, the speech recognition module, and the environmental monitoring module respectively collect the user's physiological data, visual expressions, speech features, and environmental information,

[0008] The data preprocessing module performs cleaning, synchronization, and formatting on these data in real-time transmission to ensure the quality and consistency of the data;

[0009] The emotion recognition and analysis module receives the data through the data access sub-module, and then uses the emotion classification and intensity evaluation, feature extraction and trend analysis, and abnormal emotion detection sub-modules to accurately identify and deeply analyze the user's emotional state; the analysis result is transmitted to the regulation strategy generation module;

[0010] The adjustment strategy generation module formulates corresponding physiological regulation, psychological intervention, and environmental optimization strategies according to the user's personalized needs and emotional state; these strategies are then transmitted to the execution module, and emotion regulation is implemented through means such as acoustic and optical output, physical regulation, and interactive feedback.

[0011] The user management and feedback module is responsible for collecting the user's personal information, adjustment records, and feedback opinions.

[0012] In a preferred embodiment, the biosensor module includes: an electrocardiogram (ECG) sensor for monitoring heartbeats and heart rate variability, a skin electrical activity (EDA) sensor for detecting the level of emotional arousal, and a blood oxygen saturation sensor and a body temperature sensor; the ECG sensor is attached to the user's chest with conductive gel to capture the waveform of each heartbeat, and the EDA sensor is integrated in a watch or bracelet to infer the user's emotional state by measuring changes in skin resistance.

[0013] The visual recognition module uses a camera to capture the user's facial expressions and body movements and analyzes them through computer vision technology; it includes facial recognition algorithms, expression recognition algorithms, and motion recognition algorithms; the facial recognition algorithm identifies the user's identity, the expression recognition algorithm can distinguish the user's smile and frown expressions to infer the user's emotional state, and motion recognition analyzes the user's behavior patterns, such as whether they are in an active state or showing anxious body movements.

[0014] The speech recognition module collects the user's speech data through a microphone and analyzes the pitch, rhythm, intensity, and lexical content of the speech using speech recognition technology; the speech recognition module consists of a speech-to-text system, an emotion analysis algorithm, and a speech feature extraction unit; the STT system converts the user's speech into text, and the emotion analysis algorithm analyzes the user's emotion from the pitch and rhythm of the speech.

[0015] The environmental monitoring module is responsible for collecting environmental information around the user, such as temperature, humidity, light intensity, and noise level; it includes various environmental sensors, such as temperature and humidity sensors, photosensitive sensors, and sound sensors; the temperature and humidity sensors monitor the comfort of the indoor environment, the photosensitive sensors detect whether the environmental light is suitable for the user's activities, and the sound sensors monitor whether the environmental noise has a negative impact on the user's emotion.

[0016] In a preferred embodiment, the data preprocessing module cleans, synchronizes, and formats the data collected from each sensing module to ensure the quality and consistency of the data; this module may include a data cleaning unit, a time synchronization unit, and a data standardization unit; the data cleaning unit removes outliers and noise, the time synchronization unit ensures that the data collected by different sensors is aligned in time, and the data standardization unit converts the data into a unified format and scale for subsequent analysis.

[0017] In a preferred embodiment, the data access module is responsible for receiving the original data streams from multiple sensing modules such as biosensors, visual recognition, speech recognition, and environmental monitoring, and these data streams contain rich user physiological and psychological state information;

[0018] The emotion classification and intensity assessment module distinguishes different emotion types through the support vector machine (SVM) algorithm, and the calculations include:

[0019] ①. Feature vector extraction: Extract feature vectors from the preprocessed data, and these features may include heart rate variability, skin electrical activity, facial expression features, and speech spectrum features;

[0020] ②. Training the model: Use a training data set with emotion labels to train the SVM model;

[0021] ③. Emotion classification: Use the trained model to classify new feature vectors to determine the emotion type;

[0022] ④. Intensity assessment: Evaluate the intensity of the emotion based on the distance from the support vector to the separating hyperplane;

[0023] The basic formula for SVM to find the optimal hyperplane is:

[0024] w·x + b = 0, where w is the normal vector, which determines the direction of the hyperplane; x is the feature vector; b is the bias term;

[0025] For the classification problem, the goal of SVM is to minimize the following objective function:

[0026] 1 / 2||w|| 2 ,

[0027] Subject to the following constraints:

[0028] yi(w·xi + b) ≥ 1;

[0029] For each sample i, where y_i is the class label (taking values ±1) and x_i is the feature vector of the sample;

[0030] The emotion classification and intensity evaluation module evaluates the emotion intensity by the distance from the support vector to the separation hyperplane. For a given test sample \(x_i\), its emotion intensity is expressed as:

[0031] The larger this value is, the higher the emotion intensity.

[0032] In a preferred embodiment, the emotion recognition and analysis module uses the autoregressive integrated moving average (ARIMA) model algorithm to predict the changing trend of emotions over time. The calculation content includes:

[0033] ①. Time series data preparation: Arrange the emotion classification results in chronological order to form an emotion time series;

[0034] ②. Model identification: Determine the parameters of the ARIMA model, including the number of autoregressive terms (\(p\)), the order of differencing (\(d\)), and the number of moving average terms (\(q\));

[0035] ③. Model estimation: Use historical emotion data to estimate the parameters of the ARIMA model;

[0036] ④. Trend prediction: Use the estimated model to predict the future emotion trend;

[0037] The basic formula of the ARIMA model is:

[0038] (1 - \(\varphi_1L-\varphi_2L^2-\cdots-\varphi_pL^p\))(1 - L)^dX_t = c+\(\theta_1Z_t+\theta_2Z_{t - 1}+\cdots+\theta_qZ_{t - q}\); 1 L - \(\varphi_2L^2\) 2 L^2 2 -…-\(\varphi_pL^p\) p L^p p )(1 - L)^d d X_t t =c+\(\theta_1Z_t\) 1 L^1Z_t t +\(\theta_2Z_{t - 1}\) 2 L^1 2 Z_{t - 1} t +…+\(\theta_qZ_{t - q}\) q L^q q Z_{t - q} t ;

[0039] Where:

[0040] \(X_t\) is the value of the time series data at time point \(t\);

[0041] \(L\) is the lag operator, \(L^kX_t = X_{t - k}\);

[0042] \(\varphi_1,\varphi_2,\cdots,\varphi_p\) are the coefficients of the autoregressive terms;

[0043] \(d\) is the order of differencing;

[0044] \(\theta_1,\theta_2,\cdots,\theta_q\) are the coefficients of the moving average terms;

[0045] $Z_t$ is a white noise sequence;

[0046] $c$ is a constant term.

[0047] In a preferred embodiment, the abnormal emotion detection module uses the Isolation Forest algorithm to isolate observations by randomly selecting features and split values, to identify abnormal emotions that are significantly different from other emotional states.

[0048] The calculations include:

[0049] ①. Isolation tree construction: Randomly select features and split points, and recursively construct isolation trees until each leaf node contains only one data point or reaches the specified maximum depth;

[0050] ②. Path length calculation: For each data point, calculate its average path length in all isolation trees;

[0051] ③. Anomaly score calculation: Calculate the anomaly score based on the path length. The shorter the path, the greater the likelihood that the data point is an anomaly;

[0052] The calculation formulas of the Isolation Forest algorithm include the calculation of path length and anomaly score:

[0053] where the path length $h(x)$ is the path length of the data point $x$ in the isolation tree, which is the number of edges on the path from the root node to the leaf node;

[0054] The calculation formula for the anomaly score $s$ is:

[0055] $s(x,n)=2^{-\frac{h(x)}{c(n)}}$ (-h(x) / c(n)) ;

[0056] where:

[0057] $h(x)$ is the average path length of the data point $x$;

[0058] $n$ is the total number of data points in the data set;

[0059] $c(n)$ is the harmonic mean of the given data set size $n$, and the calculation formula is:

[0060] $c(n)=2\cdot(H(n - 1)+\gamma)-\frac{2\cdot(n - 1)}{n}$;

[0061] where $H(n - 1)$ is the harmonic number of the factorial of $n - 1$, and $\gamma$ is the Euler-Mascheroni constant.

[0062] In a preferred embodiment, the adjustment strategy generation module generates corresponding emotion adjustment strategies according to the user's emotional state and personalized needs; the adjustment strategy generation module includes a user preference analysis unit, an emotional state evaluation unit, and an adjustment strategy recommendation unit; the user preference analysis unit considers the user's historical data and preference settings, the emotional state evaluation unit synthesizes the current emotion analysis results, and the adjustment strategy recommendation unit proposes a series of possible intervention measures, such as changing the environmental light, playing relaxing music, or providing psychological counseling suggestions.

[0063] In a preferred embodiment, the user management and feedback module includes a user information management unit, an emotion regulation record unit, and a user feedback processing unit; the user information management unit stores basic information such as the user's age, gender, and health status, the emotion regulation record unit records the effect of each emotion regulation, and the user feedback processing unit analyzes the user's satisfaction with the adjustment strategy to optimize the system performance.

[0064] In a preferred embodiment, the execution module includes an acoustic-optic output unit, a physical adjustment unit, and an interactive feedback unit; the acoustic-optic output unit controls the indoor lighting and plays music, the physical adjustment unit adjusts the room temperature or the comfort of the seat, and the interactive feedback unit interacts with the user through a display screen or a voice assistant to provide instant emotion regulation feedback and guidance.

[0065] In a preferred embodiment, an evaluation method for emotion regulation based on multi-dimensional perception includes:

[0066] S1: System initialization and user information entry:

[0067] Start the emotion regulation system, perform hardware self-check and software initialization;

[0068] Guide the user to enter personal information, including age, gender, health status, and psychological history, so that the system can better understand the user background and customize adjustment strategies;

[0069] S2: Real-time monitoring and data collection:

[0070] The biosensor module, visual recognition module, speech recognition module, and environmental monitoring module work simultaneously to collect the user's physiological data, facial expressions, speech features, and environmental information in real time;

[0071] The data preprocessing module cleans, synchronizes, and formats the collected data to ensure data quality;

[0072] S3: Emotion state recognition and analysis:

[0073] The emotion recognition and analysis module uses the processed data to identify the user's current emotional state and its intensity through emotion classification and intensity assessment algorithms;

[0074] The emotion trend analysis module predicts the future change trend of the user's emotion, providing a reference for long-term emotion management;

[0075] S4: Abnormal emotion detection and warning:

[0076] The abnormal emotion detection module analyzes the emotion data to identify whether there is an abnormal emotional state;

[0077] If an abnormal emotion is detected, the system immediately issues a warning and activates an emergency adjustment strategy;

[0078] S5: Generation and execution of personalized adjustment strategies:

[0079] The adjustment strategy generation module generates personalized emotion adjustment strategies based on the user's emotional state, personalized needs, and trend analysis results;

[0080] The execution module implements emotion adjustment measures through sound and light output, physical adjustment, and interactive feedback according to the generated strategy;

[0081] S6: Effect evaluation and system optimization:

[0082] The user management and feedback module collects the user's feedback on the emotion adjustment effect and the process data of the system's execution of the adjustment strategy;

[0083] Analyze the feedback and data, evaluate the effect of emotion adjustment, and optimize the performance of the emotion recognition and analysis module, the adjustment strategy generation module, and the execution module according to the evaluation results;

[0084] Update the optimized parameters and strategies into the system to achieve more accurate and personalized emotion adjustment. In summary, due to the adoption of the above technical solutions, the beneficial effects of the present invention are:

[0085] 1. In the present invention, through the emotion classification and intensity assessment module using the support vector machine (SVM) algorithm, the system can not only accurately classify the user's emotions into different categories, but also evaluate the intensity of emotions, thereby providing a more personalized emotion adjustment plan for the user. This precise emotion recognition enables the system to provide corresponding intervention measures according to the user's real-time emotional state, such as relaxation music, breathing guidance, or emotion catharsis games, effectively helping the user relieve negative emotions and improve the ability of emotion self-management.

[0086] 2. In the present invention, the emotional trend analysis module uses the autoregressive integrated moving average model (ARIMA) to predict the emotional change trend, enabling the system to be forward-looking, identify possible emotional fluctuations in advance, and take preventive measures. This predictive ability is crucial for preventing emotional crises and maintaining the long-term mental health of users. At the same time, the abnormal emotion detection module effectively identifies abnormal emotion patterns through the isolation forest algorithm, which is of great significance for timely discovering and handling potential mental health problems. Through the combined action of these modules, the system can quickly respond when the user's emotion is abnormal, provide timely assistance, and thus reduce the impact of emotional problems on the user's daily life and work. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 is the overall system block diagram of the present invention;

[0088] Figure 2 is the system block diagram of the emotion recognition and analysis module in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0089] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0090] Referring to Figure 1-2 ,

[0091] An emotion regulation system based on multi-dimensional perception, the system comprising: a biosensor module, a visual recognition module, a speech recognition module, an environmental monitoring module, a data preprocessing module, an emotion recognition and analysis module, a regulation strategy generation module, a user management and feedback module, and an execution module;

[0092] The emotion recognition and analysis module internally is provided with a data access module, an emotion classification and intensity evaluation module, an emotional trend analysis module, and an abnormal emotion detection module;

[0093] The biosensor module, the visual recognition module, the speech recognition module, and the environmental monitoring module respectively collect the physiological data, visual expressions, speech features, and environmental information of the user,

[0094] The data preprocessing module performs cleaning, synchronization, and formatting on these data in real-time transmission to ensure the quality and consistency of the data.

[0095] The emotion recognition and analysis module receives data through the data access sub-module, and then uses the emotion classification and intensity evaluation, feature extraction and trend analysis, and abnormal emotion detection sub-modules to accurately identify and deeply analyze the user's emotional state. The analysis results are transmitted to the adjustment strategy generation module, which formulates corresponding physiological adjustment, psychological intervention, and environmental optimization strategies according to the user's personalized needs and emotional state. These strategies are then transmitted to the execution module to implement emotion regulation through means such as sound and light output, physical adjustment, and interactive feedback. At the same time, the user management and feedback module is responsible for collecting the user's personal information, adjustment records, and feedback opinions, and these information are used to continuously optimize the algorithms and strategies of the emotion recognition and analysis module and the adjustment strategy generation module, forming a closed-loop feedback optimization system, so as to realize the intelligent management of the user's emotions and the maintenance of mental health

[0096] The biosensor module is the core sensing unit of the system, responsible for real-time monitoring of the user's physiological signals. This module includes multiple sub-units, such as an electrocardiogram (ECG) sensor for monitoring heartbeats and heart rate variability, a skin electroactivity (EDA) sensor for detecting the level of emotional arousal, as well as a blood oxygen saturation sensor and a body temperature sensor, etc. The ECG sensor is attached to the user's chest with conductive gel and can accurately capture the waveform of each heartbeat, while the EDA sensor is usually integrated in a watch or bracelet and infers the user's emotional state by measuring the change in skin resistance;

[0097] The visual recognition module uses a camera to capture the user's facial expressions and body movements and analyzes them through computer vision technology. This module may include face recognition algorithms, expression recognition algorithms, and action recognition algorithms. The face recognition algorithm can identify the user's identity, the expression recognition algorithm can distinguish the user's expressions such as smiling and frowning, thereby inferring the user's emotional state, and action recognition can analyze the user's behavior patterns, such as whether they are in an active state or showing anxious body movements.

[0098] The speech recognition module collects the user's speech data through a microphone and analyzes the pitch, rhythm, intensity, and lexical content of the speech using speech recognition technology. This module may include a speech-to-text (STT) system, an emotion analysis algorithm, and a speech feature extraction unit. The STT system converts the user's speech into text, and the emotion analysis algorithm analyzes the user's emotion from the pitch and rhythm of the speech, such as whether it contains emotions of anger, sadness, or excitement

[0099] The environmental monitoring module is responsible for collecting environmental information around the user, such as temperature, humidity, light intensity, and noise level, etc. This module may include various environmental sensors, such as temperature and humidity sensors, photosensitive sensors, and sound sensors. The temperature and humidity sensors can monitor the comfort level of the indoor environment, the photosensitive sensors can detect whether the environmental light is suitable for the user's activities, and the sound sensors can monitor whether the environmental noise has a negative impact on the user's mood.

[0100] The data preprocessing module cleans, synchronizes, and formats the data collected from each perception module to ensure the quality and consistency of the data. This module may include a data cleaning unit, a time synchronization unit, and a data standardization unit. The data cleaning unit removes outliers and noise, the time synchronization unit ensures that the data collected by different sensors are aligned in time, and the data standardization unit converts the data into a unified format and scale for subsequent analysis.

[0101] The data access module is responsible for receiving the original data streams from multiple perception modules such as biosensors, visual recognition, speech recognition, and environmental monitoring. These data streams contain rich information about the user's physiological and psychological states. Secondly, this module has a data synchronization function to ensure that the data from different sources are aligned on the time axis for accurate cross-modal emotion analysis.

[0102] The emotion classification and intensity evaluation module uses the Support Vector Machine (SVM) algorithm to distinguish different emotion types. The calculations include:

[0103] ①. Feature vector extraction: Extract feature vectors from the preprocessed data. These features may include heart rate variability, skin electrical activity, facial expression features, speech spectrum features, etc.

[0104] ②. Model training: Use a training data set with emotion labels to train the SVM model.

[0105] ③. Emotion classification: Use the trained model to classify new feature vectors to determine the emotion type.

[0106] ④. Intensity evaluation: Evaluate the intensity of the emotion according to the distance from the support vector to the separating hyperplane.

[0107] The basic formula for SVM to find the optimal hyperplane is:

[0108] w·x + b = 0, where w is the normal vector that determines the direction of the hyperplane, x is the feature vector, and b is the bias term.

[0109] For classification problems, the goal of SVM is to minimize the following objective function:

[0110] 1 / 2||w|| 2,

[0111] Subject to the following constraints:

[0112] y_i(w·x_i + b) ≥ 1;

[0113] For each sample i, where y_i is the class label (taking values ±1) and x_i is the feature vector of the sample;

[0114] The emotion classification and intensity evaluation module evaluates the emotion intensity by the distance from the support vector to the separation hyperplane. For a given test sample x_i, its emotion intensity can be expressed as:

[0115] The larger this value, the higher the emotion intensity.

[0116] The emotion recognition and analysis module uses the autoregressive integrated moving average (ARIMA) model algorithm to predict the changing trend of emotions over time. The calculation content includes:

[0117] ①. Time series data preparation: Arrange the emotion classification results in chronological order to form an emotion time series.

[0118] ②. Model identification: Determine the parameters of the ARIMA model, including the number of autoregressive terms (p), the order of differencing (d), and the number of moving average terms (q).

[0119] ③. Model estimation: Use historical emotion data to estimate the parameters of the ARIMA model.

[0120] ④. Trend prediction: Use the estimated model to predict the future emotion trend;

[0121] The basic formula of the ARIMA model is:

[0122] (1 - φ 1 L - φ 2 L 2 -… - φ p L p )(1 - L) d X t = c + θ 1 LZ t + θ 2 L 2 Z t +… + θ q L q Z t ;

[0123] Where:

[0124] X_t is the value of the time series data at time point t.

[0125] L is the lag operator, and \(L^k X_t = X_{t - k}\).

[0126] \(\varphi_1,\varphi_2,\cdots,\varphi_p\) are the coefficients of the autoregressive terms.

[0127] d is the order of differencing.

[0128] \(\theta_1,\theta_2,\cdots,\theta_q\) are the coefficients of the moving average terms.

[0129] \(Z_t\) is a white noise sequence.

[0130] c is the constant term.

[0131] The abnormal emotion detection module uses the Isolation Forest algorithm to isolate observations by randomly selecting features and split values, and to identify abnormal emotions that are significantly different from other emotional states.

[0132] The calculations include:

[0133] ①. Isolation tree construction: Randomly select features and split points, and recursively construct isolation trees until each leaf node contains only one data point or reaches the specified maximum depth.

[0134] ②. Path length calculation: For each data point, calculate its average path length in all isolation trees.

[0135] ③. Abnormal score calculation: Calculate the abnormal score based on the path length. The shorter the path, the greater the likelihood that the data point is abnormal.

[0136] The calculation formula of the Isolation Forest algorithm includes the calculation of path length and abnormal score:

[0137] Among them, the path length \(h(x)\) is the path length of the data point \(x\) in the isolation tree, which is the number of edges on the path from the root node to the leaf node;

[0138] The calculation formula for the abnormal score \(s\) is:

[0139] \(s(x,n)=2\) (-h(x) / c(n)) ;

[0140] Where:

[0141] \(h(x)\) is the average path length of the data point \(x\).

[0142] n is the total number of data points in the data set.

[0143] \(c(n)\) is the harmonic mean of the given data set size \(n\), and the calculation formula is:

[0144] \(c(n)=2\cdot(H(n - 1)+\gamma)-2\cdot(n - 1) / n\);

[0145] Where H(n - 1) is the harmonic number of the factorial of n - 1, and γ is the Euler-Mascheroni constant.

[0146] The adjustment strategy generation module generates corresponding emotion regulation strategies according to the user's emotional state and personalized needs. This module may include a user preference analysis unit, an emotional state assessment unit, and an adjustment strategy recommendation unit. The user preference analysis unit considers the user's historical data and preference settings, the emotional state assessment unit synthesizes the current emotional analysis results, and the adjustment strategy recommendation unit proposes a series of possible intervention measures, such as changing the ambient light, playing relaxing music, or providing psychological counseling suggestions.

[0147] The user management and feedback module is responsible for managing the user's personal information, adjustment records, and collecting user feedback. This module may include a user information management unit, an emotion regulation record unit, and a user feedback processing unit. The user information management unit stores basic information such as the user's age, gender, health status, etc., the emotion regulation record unit records the effect of each emotion regulation, and the user feedback processing unit analyzes the user's satisfaction with the adjustment strategy for optimizing system performance.

[0148] The execution module is responsible for implementing the adjustment measures recommended by the adjustment strategy generation module. This module may include an acoustic-optic output unit, a physical adjustment unit, and an interactive feedback unit. The acoustic-optic output unit can control the indoor lighting and play music, the physical adjustment unit can adjust the room temperature or the comfort of the seat, and the interactive feedback unit interacts with the user through a display screen or a voice assistant to provide instant emotion regulation feedback and guidance.

[0149] An evaluation method for emotion regulation based on multi-dimensional perception, which uses the emotion regulation system of the above embodiment for evaluation. The specific evaluation steps include:

[0150] S1: System initialization and user information entry:

[0151] Start the emotion regulation system for hardware self-check and software initialization.

[0152] Guide the user to enter personal information, including age, gender, health status, psychological history, etc., so that the system can better understand the user background and customize adjustment strategies.

[0153] S2: Real-time monitoring and data collection:

[0154] The biosensor module, visual recognition module, speech recognition module, and environmental monitoring module work simultaneously to collect the user's physiological data, facial expressions, speech features, and environmental information in real time.

[0155] The data preprocessing module cleans, synchronizes, and formats the collected data to ensure data quality.

[0156] S3: Emotional state recognition and analysis:

[0157] The emotion recognition and analysis module uses the processed data to identify the user's current emotional state and its intensity through emotion classification and intensity assessment algorithms.

[0158] The emotion trend analysis module predicts the future change trend of the user's emotion, providing a reference for long-term emotion management.

[0159] S4: Abnormal emotion detection and warning:

[0160] The abnormal emotion detection module analyzes the emotion data to identify whether there is an abnormal emotional state.

[0161] If an abnormal emotion is detected, the system immediately issues a warning and activates an emergency adjustment strategy.

[0162] S5: Generation and execution of personalized adjustment strategies:

[0163] The adjustment strategy generation module generates personalized emotion adjustment strategies based on the user's emotional state, personalized needs, and trend analysis results.

[0164] The execution module implements emotion adjustment measures through sound and light output, physical adjustment, and interactive feedback according to the generated strategy.

[0165] S6: Effect evaluation and system optimization:

[0166] The user management and feedback module collects the user's feedback on the emotion adjustment effect and the process data of the system's execution of the adjustment strategy.

[0167] Analyze the feedback and data, evaluate the effect of emotion adjustment, and optimize the performance of the emotion recognition and analysis module, the adjustment strategy generation module, and the execution module according to the evaluation results.

[0168] Update the optimized parameters and strategies into the system to achieve more accurate and personalized emotion adjustment.

[0169] In the present invention, through the emotion classification and intensity assessment module using the support vector machine (SVM) algorithm, the system can not only accurately classify the user's emotions into different categories, but also evaluate the intensity of emotions, thereby providing a more personalized emotion adjustment plan for the user. This precise emotion recognition enables the system to provide corresponding intervention measures according to the user's real-time emotional state, such as relaxation music, breathing guidance, or emotion catharsis games, effectively helping the user relieve negative emotions and improve the ability of emotion self-management.

[0170] In the present invention, the emotional trend analysis module uses the autoregressive integrated moving average model (ARIMA) to predict the emotional change trend, enabling the system to be forward-looking, identify possible emotional fluctuations in advance, and take preventive measures. This predictive ability is crucial for preventing emotional crises and maintaining the long-term mental health of users. At the same time, the abnormal emotion detection module effectively identifies abnormal emotion patterns through the isolation forest algorithm, which is of great significance for timely detecting and handling potential mental health problems. Through the combined action of these modules, the system can quickly respond when the user's emotion is abnormal, provide timely assistance, thereby reducing the impact of emotional problems on the user's daily life and work.

[0171] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the existence of additional identical elements in the process, method, article, or device comprising the element.

[0172] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An emotion regulation system based on multi-dimensional perception, characterized in that: The system includes: a biosensor module, a visual recognition module, a speech recognition module, an environmental monitoring module, a data preprocessing module, an emotion recognition and analysis module, a regulation strategy generation module, a user management and feedback module, and an execution module; The emotion recognition and analysis module is internally provided with a data access module, an emotion classification and intensity assessment module, an emotion trend analysis module and an abnormal emotion detection module; The biosensor module, visual recognition module, voice recognition module and environmental monitoring module collect the user's physiological data, visual expressions, voice features and environmental information respectively. The data preprocessing module cleans, synchronizes and formats the data in real time to ensure the quality and consistency of the data; The emotion recognition and analysis module receives data through the data access submodule, and then uses the emotion classification and intensity assessment, feature extraction and trend analysis, and abnormal emotion detection submodules to accurately identify and deeply analyze the user's emotional state; the analysis results are transmitted to the adjustment strategy generation module; The regulation strategy generation module formulates corresponding physiological regulation, psychological intervention and environmental optimization strategies according to the user's personalized needs and emotional state; these strategies are then transmitted to the execution module to implement emotional regulation through sound and light output, physical regulation and interactive feedback; The user management and feedback module is responsible for collecting users' personal information, adjustment records and feedback.

2. The emotion regulation system based on multi-dimensional perception as claimed in claim 1, characterized in that: The biosensor module includes: an electrocardiogram (ECG) sensor for monitoring heartbeat and heart rate variability, an electrodermal activity (EDA) sensor for detecting emotional arousal levels, a blood oxygen saturation sensor, and a body temperature sensor; the ECG sensor is attached to the user's chest through conductive adhesive to capture the waveform of each heartbeat, and the EDA sensor is integrated in a watch or bracelet to infer the user's emotional state by measuring changes in skin resistance; The visual recognition module uses a camera to capture the user's facial expressions and body movements, and analyzes them through computer vision technology; it includes a facial recognition algorithm, an expression recognition algorithm, and a motion recognition algorithm; the facial recognition algorithm identifies the user's identity, the expression recognition algorithm can distinguish between the user's smiling and frowning expressions, thereby inferring the user's emotional state, and the motion recognition analyzes the user's behavior patterns, such as whether he is in an active state or shows anxious body movements; The speech recognition module collects the user's speech data through a microphone and uses speech recognition technology to analyze the tone, rhythm, intensity and vocabulary content of the speech; the speech recognition module uses a speech-to-text system, a sentiment analysis algorithm and a speech feature extraction unit; the STT system converts the user's speech into text, and the sentiment analysis algorithm analyzes the user's emotions from the tone and rhythm of the speech; The environmental monitoring module is responsible for collecting environmental information around the user, such as temperature, humidity, light intensity and noise level; it includes various environmental sensors, such as temperature and humidity sensors, light sensors and sound sensors; the temperature and humidity sensors monitor the comfort of the indoor environment, the light sensors detect whether the ambient light is suitable for the user's activities, and the sound sensors monitor whether the ambient noise has a negative impact on the user's emotions.

3. The emotion regulation system based on multi-dimensional perception as claimed in claim 1, characterized in that: The data preprocessing module cleans, synchronizes and formats the data collected from each perception module to ensure the quality and consistency of the data; the module may include a data cleaning unit, a time synchronization unit and a data standardization unit; the data cleaning unit removes outliers and noise, the time synchronization unit ensures that the data collected by different sensors are aligned in time, and the data standardization unit converts the data into a unified format and scale for subsequent analysis.

4. The emotion regulation system based on multi-dimensional perception as claimed in claim 1, characterized in that: The data access module is responsible for receiving raw data streams from multiple perception modules such as biosensors, visual recognition, voice recognition and environmental monitoring. These data streams contain rich information about the user's physiological and psychological states. The emotion classification and intensity assessment module uses the support vector machine (SVM) algorithm to distinguish different emotion types. The calculation content includes: ①. Feature vector extraction: Extract feature vectors from preprocessed data. These features may include heart rate variability, skin electrode activity, facial expression features, and speech spectrum features; ②. Training model: Use the training dataset with emotion labels to train the SVM model; ③. Emotion classification: Use the trained model to classify the new feature vector and determine the emotion type; ④. Intensity evaluation: Evaluate the intensity of the emotion based on the distance from the support vector to the separating hyperplane; The basic formula for SVM to find the optimal hyperplane is: w·x+b=0, where w is the normal vector, which determines the direction of the hyperplane; x is the eigenvector; b is the bias term; For classification problems, the goal of SVM is to minimize the following objective function: 1 / 2||w|| 2 ， Subject to the following constraints: yi(w·xi+b)≥1; For each sample i, y_i is the category label (value ±1), and x_i is the feature vector of the sample; The emotion classification and intensity assessment module assesses the emotion intensity by the distance from the support vector to the separating hyperplane. For a given test sample x_i, its emotion intensity is expressed as: The larger the value, the higher the intensity of the emotion.

5. The emotion regulation system based on multi-dimensional perception as claimed in claim 1, characterized in that: The emotion recognition analysis module uses the autoregressive moving average model algorithm to predict the changing trend of emotions over time. The calculation content includes: ①. Time series data preparation: Arrange the emotion classification results in chronological order to form an emotion time series; ②. Model identification: Determine the parameters of the ARIMA model, including the number of autoregressive terms (p), the order of difference (d), and the number of moving average terms (q); ③. Model estimation: Use historical sentiment data to estimate the parameters of the ARIMA model; ④. Trend prediction: Use the estimated model to predict future sentiment trends; The basic formula of the ARIMA model is: (1-φ1L-φ2L 2 -…-f p L p (1-L) d X t =c+θ1LZ t +θ2L 2 Z t +…+θ q L q Z t ; in: X_t is the value of the time series data at time point t; L is the lag operator, L^k X_t=X_{tk}; φ1, φ2, ..., φp are the coefficients of the autoregressive term; d is the difference order; θ1,θ2,...,θq are the coefficients of the moving average term; Z_t is a white noise sequence; c is a constant term.

6. The emotion regulation system based on multi-dimensional perception as claimed in claim 1, characterized in that: The abnormal emotion detection module uses the isolation forest algorithm to isolate observations by randomly selecting features and cutoff values ​​to identify abnormal emotions that are significantly different from other emotional states. The calculation includes: ①. Isolation tree construction: randomly select features and split points, and recursively build isolation trees until each leaf node contains only one data point or reaches the specified maximum depth; ②. Path length calculation: For each data point, calculate its average path length in all isolation trees; ③. Anomaly score calculation: The anomaly score is calculated based on the path length. The shorter the path, the greater the possibility that the data point is an anomaly. The calculation formula of the isolation forest algorithm includes the calculation of path length and anomaly score: The path length h(x) is the path length of the data point x in the isolation tree, which is the number of edges on the path from the root node to the leaf node; The calculation formula of the anomaly score s is: s(x,n)=2 (-h(x) / c(n)) ; in: h(x) is the average path length of data point x; n is the total number of data points in the dataset; c(n) is the harmonic mean of a given data set of size n and is calculated as: c(n)=2·(H(n-1)+γ)-2·(n-1) / n; where H(n-1) is the harmonic number of the factorial of n-1, and γ is the Euler-Mascheroni constant.

7. The emotion regulation system based on multi-dimensional perception as claimed in claim 1, characterized in that: The regulation strategy generation module generates a corresponding emotion regulation strategy according to the user's emotional state and personalized needs; the regulation strategy generation module includes a user preference analysis unit, an emotional state evaluation unit and a regulation strategy recommendation unit; The user preference analysis unit takes into account the user's historical data and preference settings, the emotional state assessment unit integrates the current emotional analysis results, and the regulation strategy recommendation unit proposes a series of possible intervention measures, such as changing the ambient lighting, playing relaxing music, or providing psychological counseling suggestions.

8. The emotion regulation system based on multi-dimensional perception as claimed in claim 1, characterized in that: The user management and feedback module includes a user information management unit, an emotion regulation recording unit and a user feedback processing unit; the user information management unit will store the user's basic information such as age, gender and health status, the emotion regulation recording unit will record the effect of each emotion regulation, and the user feedback processing unit will analyze the user's satisfaction with the regulation strategy to optimize system performance.

9. The emotion regulation system based on multi-dimensional perception as claimed in claim 1, characterized in that: The execution module includes an audio-visual output unit, a physical adjustment unit and an interactive feedback unit; the audio-visual output unit controls the lighting and music playing in the room, the physical adjustment unit adjusts the room temperature or the comfort of the seat, and the interactive feedback unit interacts with the user through a display screen or a voice assistant to provide instant emotion regulation feedback and guidance.

10. An assessment method for emotion regulation based on multidimensional perception, characterized in that: The evaluation method uses the emotion regulation system according to any one of claims 1 to 9 for evaluation, and the specific evaluation steps include: S1: System initialization and user information entry: Start the emotion regulation system and perform hardware self-check and software initialization; Instruct users to enter personal information, including age, gender, health status, and psychological history, so that the system can better understand the user's background and customize the adjustment strategy; S2: Real-time monitoring and data collection: The biosensor module, visual recognition module, voice recognition module and environmental monitoring module work simultaneously to collect the user's physiological data, facial expressions, voice characteristics and environmental information in real time; The data preprocessing module cleans, synchronizes and formats the collected data to ensure data quality; S3: Emotional state recognition and analysis: The emotion recognition analysis module uses the processed data to identify the user's current emotional state and its intensity through emotion classification and intensity assessment algorithms; The emotion trend analysis module predicts the future trend of user emotions and provides a reference for long-term emotion management; S4: Abnormal emotion detection and early warning: The abnormal emotion detection module analyzes the emotion data to identify whether there is an abnormal emotional state; If abnormal emotions are detected, the system immediately issues an alert and initiates an emergency adjustment strategy; S5: Personalized adjustment strategy generation and execution: The regulation strategy generation module generates personalized emotion regulation strategies based on the user's emotional state, personalized needs and trend analysis results; The execution module implements emotion regulation measures according to the generated strategies through sound and light output, physical adjustment and interactive feedback; S6: Effect evaluation and system optimization: The user management and feedback module collects user feedback on the effect of emotion regulation and the process data of the system executing the regulation strategy; Analyze feedback and data, evaluate the effectiveness of emotion regulation, and optimize the performance of the emotion recognition and analysis module, regulation strategy generation module, and execution module based on the evaluation results; Update the optimized parameters and strategies into the system to achieve more accurate and personalized emotion regulation.

Images